Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning

PubMed Central

Hoo-Chang, Shin; Roth, Holger R.; Gao, Mingchen; Lu, Le; Xu, Ziyue; Nogues, Isabella; Yao, Jianhua; Mollura, Daniel

2016-01-01

Remarkable progress has been made in image recognition, primarily due to the availability of large-scale annotated datasets (i.e. ImageNet) and the revival of deep convolutional neural networks (CNN). CNNs enable learning data-driven, highly representative, layered hierarchical image features from sufficient training data. However, obtaining datasets as comprehensively annotated as ImageNet in the medical imaging domain remains a challenge. There are currently three major techniques that successfully employ CNNs to medical image classification: training the CNN from scratch, using off-the-shelf pre-trained CNN features, and conducting unsupervised CNN pre-training with supervised fine-tuning. Another effective method is transfer learning, i.e., fine-tuning CNN models (supervised) pre-trained from natural image dataset to medical image tasks (although domain transfer between two medical image datasets is also possible). In this paper, we exploit three important, but previously understudied factors of employing deep convolutional neural networks to computer-aided detection problems. We first explore and evaluate different CNN architectures. The studied models contain 5 thousand to 160 million parameters, and vary in numbers of layers. We then evaluate the influence of dataset scale and spatial image context on performance. Finally, we examine when and why transfer learning from pre-trained ImageNet (via fine-tuning) can be useful. We study two specific computeraided detection (CADe) problems, namely thoraco-abdominal lymph node (LN) detection and interstitial lung disease (ILD) classification. We achieve the state-of-the-art performance on the mediastinal LN detection, with 85% sensitivity at 3 false positive per patient, and report the first five-fold cross-validation classification results on predicting axial CT slices with ILD categories. Our extensive empirical evaluation, CNN model analysis and valuable insights can be extended to the design of high performance CAD systems for other medical imaging tasks. PMID:26886976

Deep Convolutional Neural Networks for Computer-Aided Detection: CNN Architectures, Dataset Characteristics and Transfer Learning.

PubMed

Shin, Hoo-Chang; Roth, Holger R; Gao, Mingchen; Lu, Le; Xu, Ziyue; Nogues, Isabella; Yao, Jianhua; Mollura, Daniel; Summers, Ronald M

2016-05-01

Remarkable progress has been made in image recognition, primarily due to the availability of large-scale annotated datasets and deep convolutional neural networks (CNNs). CNNs enable learning data-driven, highly representative, hierarchical image features from sufficient training data. However, obtaining datasets as comprehensively annotated as ImageNet in the medical imaging domain remains a challenge. There are currently three major techniques that successfully employ CNNs to medical image classification: training the CNN from scratch, using off-the-shelf pre-trained CNN features, and conducting unsupervised CNN pre-training with supervised fine-tuning. Another effective method is transfer learning, i.e., fine-tuning CNN models pre-trained from natural image dataset to medical image tasks. In this paper, we exploit three important, but previously understudied factors of employing deep convolutional neural networks to computer-aided detection problems. We first explore and evaluate different CNN architectures. The studied models contain 5 thousand to 160 million parameters, and vary in numbers of layers. We then evaluate the influence of dataset scale and spatial image context on performance. Finally, we examine when and why transfer learning from pre-trained ImageNet (via fine-tuning) can be useful. We study two specific computer-aided detection (CADe) problems, namely thoraco-abdominal lymph node (LN) detection and interstitial lung disease (ILD) classification. We achieve the state-of-the-art performance on the mediastinal LN detection, and report the first five-fold cross-validation classification results on predicting axial CT slices with ILD categories. Our extensive empirical evaluation, CNN model analysis and valuable insights can be extended to the design of high performance CAD systems for other medical imaging tasks.

Deep Learning for Lowtextured Image Matching

NASA Astrophysics Data System (ADS)

Kniaz, V. V.; Fedorenko, V. V.; Fomin, N. A.

2018-05-01

Low-textured objects pose challenges for an automatic 3D model reconstruction. Such objects are common in archeological applications of photogrammetry. Most of the common feature point descriptors fail to match local patches in featureless regions of an object. Hence, automatic documentation of the archeological process using Structure from Motion (SfM) methods is challenging. Nevertheless, such documentation is possible with the aid of a human operator. Deep learning-based descriptors have outperformed most of common feature point descriptors recently. This paper is focused on the development of a new Wide Image Zone Adaptive Robust feature Descriptor (WIZARD) based on the deep learning. We use a convolutional auto-encoder to compress discriminative features of a local path into a descriptor code. We build a codebook to perform point matching on multiple images. The matching is performed using the nearest neighbor search and a modified voting algorithm. We present a new "Multi-view Amphora" (Amphora) dataset for evaluation of point matching algorithms. The dataset includes images of an Ancient Greek vase found at Taman Peninsula in Southern Russia. The dataset provides color images, a ground truth 3D model, and a ground truth optical flow. We evaluated the WIZARD descriptor on the "Amphora" dataset to show that it outperforms the SIFT and SURF descriptors on the complex patch pairs.

NutriNet: A Deep Learning Food and Drink Image Recognition System for Dietary Assessment

PubMed Central

Koroušić Seljak, Barbara

2017-01-01

Automatic food image recognition systems are alleviating the process of food-intake estimation and dietary assessment. However, due to the nature of food images, their recognition is a particularly challenging task, which is why traditional approaches in the field have achieved a low classification accuracy. Deep neural networks have outperformed such solutions, and we present a novel approach to the problem of food and drink image detection and recognition that uses a newly-defined deep convolutional neural network architecture, called NutriNet. This architecture was tuned on a recognition dataset containing 225,953 512 × 512 pixel images of 520 different food and drink items from a broad spectrum of food groups, on which we achieved a classification accuracy of 86.72%, along with an accuracy of 94.47% on a detection dataset containing 130,517 images. We also performed a real-world test on a dataset of self-acquired images, combined with images from Parkinson’s disease patients, all taken using a smartphone camera, achieving a top-five accuracy of 55%, which is an encouraging result for real-world images. Additionally, we tested NutriNet on the University of Milano-Bicocca 2016 (UNIMIB2016) food image dataset, on which we improved upon the provided baseline recognition result. An online training component was implemented to continually fine-tune the food and drink recognition model on new images. The model is being used in practice as part of a mobile app for the dietary assessment of Parkinson’s disease patients. PMID:28653995

Intervertebral disc detection in X-ray images using faster R-CNN.

PubMed

Ruhan Sa; Owens, William; Wiegand, Raymond; Studin, Mark; Capoferri, Donald; Barooha, Kenneth; Greaux, Alexander; Rattray, Robert; Hutton, Adam; Cintineo, John; Chaudhary, Vipin

2017-07-01

Automatic identification of specific osseous landmarks on the spinal radiograph can be used to automate calculations for correcting ligament instability and injury, which affect 75% of patients injured in motor vehicle accidents. In this work, we propose to use deep learning based object detection method as the first step towards identifying landmark points in lateral lumbar X-ray images. The significant breakthrough of deep learning technology has made it a prevailing choice for perception based applications, however, the lack of large annotated training dataset has brought challenges to utilizing the technology in medical image processing field. In this work, we propose to fine tune a deep network, Faster-RCNN, a state-of-the-art deep detection network in natural image domain, using small annotated clinical datasets. In the experiment we show that, by using only 81 lateral lumbar X-Ray training images, one can achieve much better performance compared to traditional sliding window detection method on hand crafted features. Furthermore, we fine-tuned the network using 974 training images and tested on 108 images, which achieved average precision of 0.905 with average computation time of 3 second per image, which greatly outperformed traditional methods in terms of accuracy and efficiency.

DeepPap: Deep Convolutional Networks for Cervical Cell Classification.

PubMed

Zhang, Ling; Le Lu; Nogues, Isabella; Summers, Ronald M; Liu, Shaoxiong; Yao, Jianhua

2017-11-01

Automation-assisted cervical screening via Pap smear or liquid-based cytology (LBC) is a highly effective cell imaging based cancer detection tool, where cells are partitioned into "abnormal" and "normal" categories. However, the success of most traditional classification methods relies on the presence of accurate cell segmentations. Despite sixty years of research in this field, accurate segmentation remains a challenge in the presence of cell clusters and pathologies. Moreover, previous classification methods are only built upon the extraction of hand-crafted features, such as morphology and texture. This paper addresses these limitations by proposing a method to directly classify cervical cells-without prior segmentation-based on deep features, using convolutional neural networks (ConvNets). First, the ConvNet is pretrained on a natural image dataset. It is subsequently fine-tuned on a cervical cell dataset consisting of adaptively resampled image patches coarsely centered on the nuclei. In the testing phase, aggregation is used to average the prediction scores of a similar set of image patches. The proposed method is evaluated on both Pap smear and LBC datasets. Results show that our method outperforms previous algorithms in classification accuracy (98.3%), area under the curve (0.99) values, and especially specificity (98.3%), when applied to the Herlev benchmark Pap smear dataset and evaluated using five-fold cross validation. Similar superior performances are also achieved on the HEMLBC (H&E stained manual LBC) dataset. Our method is promising for the development of automation-assisted reading systems in primary cervical screening.

Deep learning-based fine-grained car make/model classification for visual surveillance

NASA Astrophysics Data System (ADS)

Gundogdu, Erhan; Parıldı, Enes Sinan; Solmaz, Berkan; Yücesoy, Veysel; Koç, Aykut

2017-10-01

Fine-grained object recognition is a potential computer vision problem that has been recently addressed by utilizing deep Convolutional Neural Networks (CNNs). Nevertheless, the main disadvantage of classification methods relying on deep CNN models is the need for considerably large amount of data. In addition, there exists relatively less amount of annotated data for a real world application, such as the recognition of car models in a traffic surveillance system. To this end, we mainly concentrate on the classification of fine-grained car make and/or models for visual scenarios by the help of two different domains. First, a large-scale dataset including approximately 900K images is constructed from a website which includes fine-grained car models. According to their labels, a state-of-the-art CNN model is trained on the constructed dataset. The second domain that is dealt with is the set of images collected from a camera integrated to a traffic surveillance system. These images, which are over 260K, are gathered by a special license plate detection method on top of a motion detection algorithm. An appropriately selected size of the image is cropped from the region of interest provided by the detected license plate location. These sets of images and their provided labels for more than 30 classes are employed to fine-tune the CNN model which is already trained on the large scale dataset described above. To fine-tune the network, the last two fully-connected layers are randomly initialized and the remaining layers are fine-tuned in the second dataset. In this work, the transfer of a learned model on a large dataset to a smaller one has been successfully performed by utilizing both the limited annotated data of the traffic field and a large scale dataset with available annotations. Our experimental results both in the validation dataset and the real field show that the proposed methodology performs favorably against the training of the CNN model from scratch.

Identification of autism spectrum disorder using deep learning and the ABIDE dataset.

PubMed

Heinsfeld, Anibal Sólon; Franco, Alexandre Rosa; Craddock, R Cameron; Buchweitz, Augusto; Meneguzzi, Felipe

2018-01-01

The goal of the present study was to apply deep learning algorithms to identify autism spectrum disorder (ASD) patients from large brain imaging dataset, based solely on the patients brain activation patterns. We investigated ASD patients brain imaging data from a world-wide multi-site database known as ABIDE (Autism Brain Imaging Data Exchange). ASD is a brain-based disorder characterized by social deficits and repetitive behaviors. According to recent Centers for Disease Control data, ASD affects one in 68 children in the United States. We investigated patterns of functional connectivity that objectively identify ASD participants from functional brain imaging data, and attempted to unveil the neural patterns that emerged from the classification. The results improved the state-of-the-art by achieving 70% accuracy in identification of ASD versus control patients in the dataset. The patterns that emerged from the classification show an anticorrelation of brain function between anterior and posterior areas of the brain; the anticorrelation corroborates current empirical evidence of anterior-posterior disruption in brain connectivity in ASD. We present the results and identify the areas of the brain that contributed most to differentiating ASD from typically developing controls as per our deep learning model.

Rapid and accurate intraoperative pathological diagnosis by artificial intelligence with deep learning technology.

PubMed

Zhang, Jing; Song, Yanlin; Xia, Fan; Zhu, Chenjing; Zhang, Yingying; Song, Wenpeng; Xu, Jianguo; Ma, Xuelei

2017-09-01

Frozen section is widely used for intraoperative pathological diagnosis (IOPD), which is essential for intraoperative decision making. However, frozen section suffers from some drawbacks, such as time consuming and high misdiagnosis rate. Recently, artificial intelligence (AI) with deep learning technology has shown bright future in medicine. We hypothesize that AI with deep learning technology could help IOPD, with a computer trained by a dataset of intraoperative lesion images. Evidences supporting our hypothesis included the successful use of AI with deep learning technology in diagnosing skin cancer, and the developed method of deep-learning algorithm. Large size of the training dataset is critical to increase the diagnostic accuracy. The performance of the trained machine could be tested by new images before clinical use. Real-time diagnosis, easy to use and potential high accuracy were the advantages of AI for IOPD. In sum, AI with deep learning technology is a promising method to help rapid and accurate IOPD. Copyright © 2017 Elsevier Ltd. All rights reserved.

Histopathological Breast Cancer Image Classification by Deep Neural Network Techniques Guided by Local Clustering.

PubMed

Nahid, Abdullah-Al; Mehrabi, Mohamad Ali; Kong, Yinan

2018-01-01

Breast Cancer is a serious threat and one of the largest causes of death of women throughout the world. The identification of cancer largely depends on digital biomedical photography analysis such as histopathological images by doctors and physicians. Analyzing histopathological images is a nontrivial task, and decisions from investigation of these kinds of images always require specialised knowledge. However, Computer Aided Diagnosis (CAD) techniques can help the doctor make more reliable decisions. The state-of-the-art Deep Neural Network (DNN) has been recently introduced for biomedical image analysis. Normally each image contains structural and statistical information. This paper classifies a set of biomedical breast cancer images (BreakHis dataset) using novel DNN techniques guided by structural and statistical information derived from the images. Specifically a Convolutional Neural Network (CNN), a Long-Short-Term-Memory (LSTM), and a combination of CNN and LSTM are proposed for breast cancer image classification. Softmax and Support Vector Machine (SVM) layers have been used for the decision-making stage after extracting features utilising the proposed novel DNN models. In this experiment the best Accuracy value of 91.00% is achieved on the 200x dataset, the best Precision value 96.00% is achieved on the 40x dataset, and the best F -Measure value is achieved on both the 40x and 100x datasets.

Data augmentation-assisted deep learning of hand-drawn partially colored sketches for visual search

PubMed Central

Muhammad, Khan; Baik, Sung Wook

2017-01-01

In recent years, image databases are growing at exponential rates, making their management, indexing, and retrieval, very challenging. Typical image retrieval systems rely on sample images as queries. However, in the absence of sample query images, hand-drawn sketches are also used. The recent adoption of touch screen input devices makes it very convenient to quickly draw shaded sketches of objects to be used for querying image databases. This paper presents a mechanism to provide access to visual information based on users’ hand-drawn partially colored sketches using touch screen devices. A key challenge for sketch-based image retrieval systems is to cope with the inherent ambiguity in sketches due to the lack of colors, textures, shading, and drawing imperfections. To cope with these issues, we propose to fine-tune a deep convolutional neural network (CNN) using augmented dataset to extract features from partially colored hand-drawn sketches for query specification in a sketch-based image retrieval framework. The large augmented dataset contains natural images, edge maps, hand-drawn sketches, de-colorized, and de-texturized images which allow CNN to effectively model visual contents presented to it in a variety of forms. The deep features extracted from CNN allow retrieval of images using both sketches and full color images as queries. We also evaluated the role of partial coloring or shading in sketches to improve the retrieval performance. The proposed method is tested on two large datasets for sketch recognition and sketch-based image retrieval and achieved better classification and retrieval performance than many existing methods. PMID:28859140

Deep learning in breast cancer risk assessment: evaluation of convolutional neural networks on a clinical dataset of full-field digital mammograms.

PubMed

Li, Hui; Giger, Maryellen L; Huynh, Benjamin Q; Antropova, Natalia O

2017-10-01

To evaluate deep learning in the assessment of breast cancer risk in which convolutional neural networks (CNNs) with transfer learning are used to extract parenchymal characteristics directly from full-field digital mammographic (FFDM) images instead of using computerized radiographic texture analysis (RTA), 456 clinical FFDM cases were included: a "high-risk" BRCA1/2 gene-mutation carriers dataset (53 cases), a "high-risk" unilateral cancer patients dataset (75 cases), and a "low-risk dataset" (328 cases). Deep learning was compared to the use of features from RTA, as well as to a combination of both in the task of distinguishing between high- and low-risk subjects. Similar classification performances were obtained using CNN [area under the curve [Formula: see text]; standard error [Formula: see text

Representation learning: a unified deep learning framework for automatic prostate MR segmentation.

PubMed

Liao, Shu; Gao, Yaozong; Oto, Aytekin; Shen, Dinggang

2013-01-01

Image representation plays an important role in medical image analysis. The key to the success of different medical image analysis algorithms is heavily dependent on how we represent the input data, namely features used to characterize the input image. In the literature, feature engineering remains as an active research topic, and many novel hand-crafted features are designed such as Haar wavelet, histogram of oriented gradient, and local binary patterns. However, such features are not designed with the guidance of the underlying dataset at hand. To this end, we argue that the most effective features should be designed in a learning based manner, namely representation learning, which can be adapted to different patient datasets at hand. In this paper, we introduce a deep learning framework to achieve this goal. Specifically, a stacked independent subspace analysis (ISA) network is adopted to learn the most effective features in a hierarchical and unsupervised manner. The learnt features are adapted to the dataset at hand and encode high level semantic anatomical information. The proposed method is evaluated on the application of automatic prostate MR segmentation. Experimental results show that significant segmentation accuracy improvement can be achieved by the proposed deep learning method compared to other state-of-the-art segmentation approaches.

Evaluation of a Traffic Sign Detector by Synthetic Image Data for Advanced Driver Assistance Systems

NASA Astrophysics Data System (ADS)

Hanel, A.; Kreuzpaintner, D.; Stilla, U.

2018-05-01

Recently, several synthetic image datasets of street scenes have been published. These datasets contain various traffic signs and can therefore be used to train and test machine learning-based traffic sign detectors. In this contribution, selected datasets are compared regarding ther applicability for traffic sign detection. The comparison covers the process to produce the synthetic images and addresses the virtual worlds, needed to produce the synthetic images, and their environmental conditions. The comparison covers variations in the appearance of traffic signs and the labeling strategies used for the datasets, as well. A deep learning traffic sign detector is trained with multiple training datasets with different ratios between synthetic and real training samples to evaluate the synthetic SYNTHIA dataset. A test of the detector on real samples only has shown that an overall accuracy and ROC AUC of more than 95 % can be achieved for both a small rate of synthetic samples and a large rate of synthetic samples in the training dataset.

Deep machine learning provides state-of-the-art performance in image-based plant phenotyping.

PubMed

Pound, Michael P; Atkinson, Jonathan A; Townsend, Alexandra J; Wilson, Michael H; Griffiths, Marcus; Jackson, Aaron S; Bulat, Adrian; Tzimiropoulos, Georgios; Wells, Darren M; Murchie, Erik H; Pridmore, Tony P; French, Andrew P

2017-10-01

In plant phenotyping, it has become important to be able to measure many features on large image sets in order to aid genetic discovery. The size of the datasets, now often captured robotically, often precludes manual inspection, hence the motivation for finding a fully automated approach. Deep learning is an emerging field that promises unparalleled results on many data analysis problems. Building on artificial neural networks, deep approaches have many more hidden layers in the network, and hence have greater discriminative and predictive power. We demonstrate the use of such approaches as part of a plant phenotyping pipeline. We show the success offered by such techniques when applied to the challenging problem of image-based plant phenotyping and demonstrate state-of-the-art results (>97% accuracy) for root and shoot feature identification and localization. We use fully automated trait identification using deep learning to identify quantitative trait loci in root architecture datasets. The majority (12 out of 14) of manually identified quantitative trait loci were also discovered using our automated approach based on deep learning detection to locate plant features. We have shown deep learning-based phenotyping to have very good detection and localization accuracy in validation and testing image sets. We have shown that such features can be used to derive meaningful biological traits, which in turn can be used in quantitative trait loci discovery pipelines. This process can be completely automated. We predict a paradigm shift in image-based phenotyping bought about by such deep learning approaches, given sufficient training sets. © The Authors 2017. Published by Oxford University Press.

Development of a Deep Learning Algorithm for Automatic Diagnosis of Diabetic Retinopathy.

PubMed

Raju, Manoj; Pagidimarri, Venkatesh; Barreto, Ryan; Kadam, Amrit; Kasivajjala, Vamsichandra; Aswath, Arun

2017-01-01

This paper mainly focuses on the deep learning application in classifying the stage of diabetic retinopathy and detecting the laterality of the eye using funduscopic images. Diabetic retinopathy is a chronic, progressive, sight-threatening disease of the retinal blood vessels. Ophthalmologists diagnose diabetic retinopathy through early funduscopic screening. Normally, there is a time delay in reporting and intervention, apart from the financial cost and risk of blindness associated with it. Using a convolutional neural network based approach for automatic diagnosis of diabetic retinopathy, we trained the prediction network on the publicly available Kaggle dataset. Approximately 35,000 images were used to train the network, which observed a sensitivity of 80.28% and a specificity of 92.29% on the validation dataset of ~53,000 images. Using 8,810 images, the network was trained for detecting the laterality of the eye and observed an accuracy of 93.28% on the validation set of 8,816 images.

On the Multi-Modal Object Tracking and Image Fusion Using Unsupervised Deep Learning Methodologies

NASA Astrophysics Data System (ADS)

LaHaye, N.; Ott, J.; Garay, M. J.; El-Askary, H. M.; Linstead, E.

2017-12-01

The number of different modalities of remote-sensors has been on the rise, resulting in large datasets with different complexity levels. Such complex datasets can provide valuable information separately, yet there is a bigger value in having a comprehensive view of them combined. As such, hidden information can be deduced through applying data mining techniques on the fused data. The curse of dimensionality of such fused data, due to the potentially vast dimension space, hinders our ability to have deep understanding of them. This is because each dataset requires a user to have instrument-specific and dataset-specific knowledge for optimum and meaningful usage. Once a user decides to use multiple datasets together, deeper understanding of translating and combining these datasets in a correct and effective manner is needed. Although there exists data centric techniques, generic automated methodologies that can potentially solve this problem completely don't exist. Here we are developing a system that aims to gain a detailed understanding of different data modalities. Such system will provide an analysis environment that gives the user useful feedback and can aid in research tasks. In our current work, we show the initial outputs our system implementation that leverages unsupervised deep learning techniques so not to burden the user with the task of labeling input data, while still allowing for a detailed machine understanding of the data. Our goal is to be able to track objects, like cloud systems or aerosols, across different image-like data-modalities. The proposed system is flexible, scalable and robust to understand complex likenesses within multi-modal data in a similar spatio-temporal range, and also to be able to co-register and fuse these images when needed.

Active appearance model and deep learning for more accurate prostate segmentation on MRI

NASA Astrophysics Data System (ADS)

Cheng, Ruida; Roth, Holger R.; Lu, Le; Wang, Shijun; Turkbey, Baris; Gandler, William; McCreedy, Evan S.; Agarwal, Harsh K.; Choyke, Peter; Summers, Ronald M.; McAuliffe, Matthew J.

2016-03-01

Prostate segmentation on 3D MR images is a challenging task due to image artifacts, large inter-patient prostate shape and texture variability, and lack of a clear prostate boundary specifically at apex and base levels. We propose a supervised machine learning model that combines atlas based Active Appearance Model (AAM) with a Deep Learning model to segment the prostate on MR images. The performance of the segmentation method is evaluated on 20 unseen MR image datasets. The proposed method combining AAM and Deep Learning achieves a mean Dice Similarity Coefficient (DSC) of 0.925 for whole 3D MR images of the prostate using axial cross-sections. The proposed model utilizes the adaptive atlas-based AAM model and Deep Learning to achieve significant segmentation accuracy.

Deep learning based classification for head and neck cancer detection with hyperspectral imaging in an animal model

NASA Astrophysics Data System (ADS)

Ma, Ling; Lu, Guolan; Wang, Dongsheng; Wang, Xu; Chen, Zhuo Georgia; Muller, Susan; Chen, Amy; Fei, Baowei

2017-03-01

Hyperspectral imaging (HSI) is an emerging imaging modality that can provide a noninvasive tool for cancer detection and image-guided surgery. HSI acquires high-resolution images at hundreds of spectral bands, providing big data to differentiating different types of tissue. We proposed a deep learning based method for the detection of head and neck cancer with hyperspectral images. Since the deep learning algorithm can learn the feature hierarchically, the learned features are more discriminative and concise than the handcrafted features. In this study, we adopt convolutional neural networks (CNN) to learn the deep feature of pixels for classifying each pixel into tumor or normal tissue. We evaluated our proposed classification method on the dataset containing hyperspectral images from 12 tumor-bearing mice. Experimental results show that our method achieved an average accuracy of 91.36%. The preliminary study demonstrated that our deep learning method can be applied to hyperspectral images for detecting head and neck tumors in animal models.

Deep learning in breast cancer risk assessment: evaluation of fine-tuned convolutional neural networks on a clinical dataset of FFDMs

NASA Astrophysics Data System (ADS)

Li, Hui; Mendel, Kayla R.; Lee, John H.; Lan, Li; Giger, Maryellen L.

2018-02-01

We evaluated the potential of deep learning in the assessment of breast cancer risk using convolutional neural networks (CNNs) fine-tuned on full-field digital mammographic (FFDM) images. This study included 456 clinical FFDM cases from two high-risk datasets: BRCA1/2 gene-mutation carriers (53 cases) and unilateral cancer patients (75 cases), and a low-risk dataset as the control group (328 cases). All FFDM images (12-bit quantization and 100 micron pixel) were acquired with a GE Senographe 2000D system and were retrospectively collected under an IRB-approved, HIPAA-compliant protocol. Regions of interest of 256x256 pixels were selected from the central breast region behind the nipple in the craniocaudal projection. VGG19 pre-trained on the ImageNet dataset was used to classify the images either as high-risk or as low-risk subjects. The last fully-connected layer of pre-trained VGG19 was fine-tuned on FFDM images for breast cancer risk assessment. Performance was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC) in the task of distinguishing between high-risk and low-risk subjects. AUC values of 0.84 (SE=0.05) and 0.72 (SE=0.06) were obtained in the task of distinguishing between the BRCA1/2 gene-mutation carriers and low-risk women and between unilateral cancer patients and low-risk women, respectively. Deep learning with CNNs appears to be able to extract parenchymal characteristics directly from FFDMs which are relevant to the task of distinguishing between cancer risk populations, and therefore has potential to aid clinicians in assessing mammographic parenchymal patterns for cancer risk assessment.

Adaptive template generation for amyloid PET using a deep learning approach.

PubMed

Kang, Seung Kwan; Seo, Seongho; Shin, Seong A; Byun, Min Soo; Lee, Dong Young; Kim, Yu Kyeong; Lee, Dong Soo; Lee, Jae Sung

2018-05-11

Accurate spatial normalization (SN) of amyloid positron emission tomography (PET) images for Alzheimer's disease assessment without coregistered anatomical magnetic resonance imaging (MRI) of the same individual is technically challenging. In this study, we applied deep neural networks to generate individually adaptive PET templates for robust and accurate SN of amyloid PET without using matched 3D MR images. Using 681 pairs of simultaneously acquired 11 C-PIB PET and T1-weighted 3D MRI scans of AD, MCI, and cognitively normal subjects, we trained and tested two deep neural networks [convolutional auto-encoder (CAE) and generative adversarial network (GAN)] that produce adaptive best PET templates. More specifically, the networks were trained using 685,100 pieces of augmented data generated by rotating 527 randomly selected datasets and validated using 154 datasets. The input to the supervised neural networks was the 3D PET volume in native space and the label was the spatially normalized 3D PET image using the transformation parameters obtained from MRI-based SN. The proposed deep learning approach significantly enhanced the quantitative accuracy of MRI-less amyloid PET assessment by reducing the SN error observed when an average amyloid PET template is used. Given an input image, the trained deep neural networks rapidly provide individually adaptive 3D PET templates without any discontinuity between the slices (in 0.02 s). As the proposed method does not require 3D MRI for the SN of PET images, it has great potential for use in routine analysis of amyloid PET images in clinical practice and research. © 2018 Wiley Periodicals, Inc.

Deep multi-scale convolutional neural network for hyperspectral image classification

NASA Astrophysics Data System (ADS)

Zhang, Feng-zhe; Yang, Xia

2018-04-01

In this paper, we proposed a multi-scale convolutional neural network for hyperspectral image classification task. Firstly, compared with conventional convolution, we utilize multi-scale convolutions, which possess larger respective fields, to extract spectral features of hyperspectral image. We design a deep neural network with a multi-scale convolution layer which contains 3 different convolution kernel sizes. Secondly, to avoid overfitting of deep neural network, dropout is utilized, which randomly sleeps neurons, contributing to improve the classification accuracy a bit. In addition, new skills like ReLU in deep learning is utilized in this paper. We conduct experiments on University of Pavia and Salinas datasets, and obtained better classification accuracy compared with other methods.

Deep ensemble learning of virtual endoluminal views for polyp detection in CT colonography

NASA Astrophysics Data System (ADS)

Umehara, Kensuke; Näppi, Janne J.; Hironaka, Toru; Regge, Daniele; Ishida, Takayuki; Yoshida, Hiroyuki

2017-03-01

Robust training of a deep convolutional neural network (DCNN) requires a very large number of annotated datasets that are currently not available in CT colonography (CTC). We previously demonstrated that deep transfer learning provides an effective approach for robust application of a DCNN in CTC. However, at high detection accuracy, the differentiation of small polyps from non-polyps was still challenging. In this study, we developed and evaluated a deep ensemble learning (DEL) scheme for reviewing of virtual endoluminal images to improve the performance of computer-aided detection (CADe) of polyps in CTC. Nine different types of image renderings were generated from virtual endoluminal images of polyp candidates detected by a conventional CADe system. Eleven DCNNs that represented three types of publically available pre-trained DCNN models were re-trained by transfer learning to identify polyps from the virtual endoluminal images. A DEL scheme that determines the final detected polyps by a review of the nine types of VE images was developed by combining the DCNNs using a random forest classifier as a meta-classifier. For evaluation, we sampled 154 CTC cases from a large CTC screening trial and divided the cases randomly into a training dataset and a test dataset. At 3.9 falsepositive (FP) detections per patient on average, the detection sensitivities of the conventional CADe system, the highestperforming single DCNN, and the DEL scheme were 81.3%, 90.7%, and 93.5%, respectively, for polyps ≥6 mm in size. For small polyps, the DEL scheme reduced the number of false positives by up to 83% over that of using a single DCNN alone. These preliminary results indicate that the DEL scheme provides an effective approach for improving the polyp detection performance of CADe in CTC, especially for small polyps.

ClimateNet: A Machine Learning dataset for Climate Science Research

NASA Astrophysics Data System (ADS)

Prabhat, M.; Biard, J.; Ganguly, S.; Ames, S.; Kashinath, K.; Kim, S. K.; Kahou, S.; Maharaj, T.; Beckham, C.; O'Brien, T. A.; Wehner, M. F.; Williams, D. N.; Kunkel, K.; Collins, W. D.

2017-12-01

Deep Learning techniques have revolutionized commercial applications in Computer vision, speech recognition and control systems. The key for all of these developments was the creation of a curated, labeled dataset ImageNet, for enabling multiple research groups around the world to develop methods, benchmark performance and compete with each other. The success of Deep Learning can be largely attributed to the broad availability of this dataset. Our empirical investigations have revealed that Deep Learning is similarly poised to benefit the task of pattern detection in climate science. Unfortunately, labeled datasets, a key pre-requisite for training, are hard to find. Individual research groups are typically interested in specialized weather patterns, making it hard to unify, and share datasets across groups and institutions. In this work, we are proposing ClimateNet: a labeled dataset that provides labeled instances of extreme weather patterns, as well as associated raw fields in model and observational output. We develop a schema in NetCDF to enumerate weather pattern classes/types, store bounding boxes, and pixel-masks. We are also working on a TensorFlow implementation to natively import such NetCDF datasets, and are providing a reference convolutional architecture for binary classification tasks. Our hope is that researchers in Climate Science, as well as ML/DL, will be able to use (and extend) ClimateNet to make rapid progress in the application of Deep Learning for Climate Science research.

Computer-aided classification of mammographic masses using the deep learning technology: a preliminary study

NASA Astrophysics Data System (ADS)

Qiu, Yuchen; Yan, Shiju; Tan, Maxine; Cheng, Samuel; Liu, Hong; Zheng, Bin

2016-03-01

Although mammography is the only clinically acceptable imaging modality used in the population-based breast cancer screening, its efficacy is quite controversy. One of the major challenges is how to help radiologists more accurately classify between benign and malignant lesions. The purpose of this study is to investigate a new mammographic mass classification scheme based on a deep learning method. In this study, we used an image dataset involving 560 regions of interest (ROIs) extracted from digital mammograms, which includes 280 malignant and 280 benign mass ROIs, respectively. An eight layer deep learning network was applied, which employs three pairs of convolution-max-pooling layers for automatic feature extraction and a multiple layer perception (MLP) classifier for feature categorization. In order to improve robustness of selected features, each convolution layer is connected with a max-pooling layer. A number of 20, 10, and 5 feature maps were utilized for the 1st, 2nd and 3rd convolution layer, respectively. The convolution networks are followed by a MLP classifier, which generates a classification score to predict likelihood of a ROI depicting a malignant mass. Among 560 ROIs, 420 ROIs were used as a training dataset and the remaining 140 ROIs were used as a validation dataset. The result shows that the new deep learning based classifier yielded an area under the receiver operation characteristic curve (AUC) of 0.810+/-0.036. This study demonstrated the potential superiority of using a deep learning based classifier to distinguish malignant and benign breast masses without segmenting the lesions and extracting the pre-defined image features.

White blood cells identification system based on convolutional deep neural learning networks.

PubMed

Shahin, A I; Guo, Yanhui; Amin, K M; Sharawi, Amr A

2017-11-16

White blood cells (WBCs) differential counting yields valued information about human health and disease. The current developed automated cell morphology equipments perform differential count which is based on blood smear image analysis. Previous identification systems for WBCs consist of successive dependent stages; pre-processing, segmentation, feature extraction, feature selection, and classification. There is a real need to employ deep learning methodologies so that the performance of previous WBCs identification systems can be increased. Classifying small limited datasets through deep learning systems is a major challenge and should be investigated. In this paper, we propose a novel identification system for WBCs based on deep convolutional neural networks. Two methodologies based on transfer learning are followed: transfer learning based on deep activation features and fine-tuning of existed deep networks. Deep acrivation featues are extracted from several pre-trained networks and employed in a traditional identification system. Moreover, a novel end-to-end convolutional deep architecture called "WBCsNet" is proposed and built from scratch. Finally, a limited balanced WBCs dataset classification is performed through the WBCsNet as a pre-trained network. During our experiments, three different public WBCs datasets (2551 images) have been used which contain 5 healthy WBCs types. The overall system accuracy achieved by the proposed WBCsNet is (96.1%) which is more than different transfer learning approaches or even the previous traditional identification system. We also present features visualization for the WBCsNet activation which reflects higher response than the pre-trained activated one. a novel WBCs identification system based on deep learning theory is proposed and a high performance WBCsNet can be employed as a pre-trained network. Copyright © 2017. Published by Elsevier B.V.

SAR image classification based on CNN in real and simulation datasets

NASA Astrophysics Data System (ADS)

Peng, Lijiang; Liu, Ming; Liu, Xiaohua; Dong, Liquan; Hui, Mei; Zhao, Yuejin

2018-04-01

Convolution neural network (CNN) has made great success in image classification tasks. Even in the field of synthetic aperture radar automatic target recognition (SAR-ATR), state-of-art results has been obtained by learning deep representation of features on the MSTAR benchmark. However, the raw data of MSTAR have shortcomings in training a SAR-ATR model because of high similarity in background among the SAR images of each kind. This indicates that the CNN would learn the hierarchies of features of backgrounds as well as the targets. To validate the influence of the background, some other SAR images datasets have been made which contains the simulation SAR images of 10 manufactured targets such as tank and fighter aircraft, and the backgrounds of simulation SAR images are sampled from the whole original MSTAR data. The simulation datasets contain the dataset that the backgrounds of each kind images correspond to the one kind of backgrounds of MSTAR targets or clutters and the dataset that each image shares the random background of whole MSTAR targets or clutters. In addition, mixed datasets of MSTAR and simulation datasets had been made to use in the experiments. The CNN architecture proposed in this paper are trained on all datasets mentioned above. The experimental results shows that the architecture can get high performances on all datasets even the backgrounds of the images are miscellaneous, which indicates the architecture can learn a good representation of the targets even though the drastic changes on background.

Deep learning approach to bacterial colony classification.

PubMed

Zieliński, Bartosz; Plichta, Anna; Misztal, Krzysztof; Spurek, Przemysław; Brzychczy-Włoch, Monika; Ochońska, Dorota

2017-01-01

In microbiology it is diagnostically useful to recognize various genera and species of bacteria. It can be achieved using computer-aided methods, which make the recognition processes more automatic and thus significantly reduce the time necessary for the classification. Moreover, in case of diagnostic uncertainty (the misleading similarity in shape or structure of bacterial cells), such methods can minimize the risk of incorrect recognition. In this article, we apply the state of the art method for texture analysis to classify genera and species of bacteria. This method uses deep Convolutional Neural Networks to obtain image descriptors, which are then encoded and classified with Support Vector Machine or Random Forest. To evaluate this approach and to make it comparable with other approaches, we provide a new dataset of images. DIBaS dataset (Digital Image of Bacterial Species) contains 660 images with 33 different genera and species of bacteria.

Deep image mining for diabetic retinopathy screening.

PubMed

Quellec, Gwenolé; Charrière, Katia; Boudi, Yassine; Cochener, Béatrice; Lamard, Mathieu

2017-07-01

Deep learning is quickly becoming the leading methodology for medical image analysis. Given a large medical archive, where each image is associated with a diagnosis, efficient pathology detectors or classifiers can be trained with virtually no expert knowledge about the target pathologies. However, deep learning algorithms, including the popular ConvNets, are black boxes: little is known about the local patterns analyzed by ConvNets to make a decision at the image level. A solution is proposed in this paper to create heatmaps showing which pixels in images play a role in the image-level predictions. In other words, a ConvNet trained for image-level classification can be used to detect lesions as well. A generalization of the backpropagation method is proposed in order to train ConvNets that produce high-quality heatmaps. The proposed solution is applied to diabetic retinopathy (DR) screening in a dataset of almost 90,000 fundus photographs from the 2015 Kaggle Diabetic Retinopathy competition and a private dataset of almost 110,000 photographs (e-ophtha). For the task of detecting referable DR, very good detection performance was achieved: A z =0.954 in Kaggle's dataset and A z =0.949 in e-ophtha. Performance was also evaluated at the image level and at the lesion level in the DiaretDB1 dataset, where four types of lesions are manually segmented: microaneurysms, hemorrhages, exudates and cotton-wool spots. For the task of detecting images containing these four lesion types, the proposed detector, which was trained to detect referable DR, outperforms recent algorithms trained to detect those lesions specifically, with pixel-level supervision. At the lesion level, the proposed detector outperforms heatmap generation algorithms for ConvNets. This detector is part of the Messidor® system for mobile eye pathology screening. Because it does not rely on expert knowledge or manual segmentation for detecting relevant patterns, the proposed solution is a promising image mining tool, which has the potential to discover new biomarkers in images. Copyright © 2017 Elsevier B.V. All rights reserved.

Study on the Classification of GAOFEN-3 Polarimetric SAR Images Using Deep Neural Network

NASA Astrophysics Data System (ADS)

Zhang, J.; Zhang, J.; Zhao, Z.

2018-04-01

Polarimetric Synthetic Aperture Radar (POLSAR) imaging principle determines that the image quality will be affected by speckle noise. So the recognition accuracy of traditional image classification methods will be reduced by the effect of this interference. Since the date of submission, Deep Convolutional Neural Network impacts on the traditional image processing methods and brings the field of computer vision to a new stage with the advantages of a strong ability to learn deep features and excellent ability to fit large datasets. Based on the basic characteristics of polarimetric SAR images, the paper studied the types of the surface cover by using the method of Deep Learning. We used the fully polarimetric SAR features of different scales to fuse RGB images to the GoogLeNet model based on convolution neural network Iterative training, and then use the trained model to test the classification of data validation.First of all, referring to the optical image, we mark the surface coverage type of GF-3 POLSAR image with 8m resolution, and then collect the samples according to different categories. To meet the GoogLeNet model requirements of 256 × 256 pixel image input and taking into account the lack of full-resolution SAR resolution, the original image should be pre-processed in the process of resampling. In this paper, POLSAR image slice samples of different scales with sampling intervals of 2 m and 1 m to be trained separately and validated by the verification dataset. Among them, the training accuracy of GoogLeNet model trained with resampled 2-m polarimetric SAR image is 94.89 %, and that of the trained SAR image with resampled 1 m is 92.65 %.

Application of a deep-learning method to the forecast of daily solar flare occurrence using Convolution Neural Network

NASA Astrophysics Data System (ADS)

Shin, Seulki; Moon, Yong-Jae; Chu, Hyoungseok

2017-08-01

As the application of deep-learning methods has been succeeded in various fields, they have a high potential to be applied to space weather forecasting. Convolutional neural network, one of deep learning methods, is specialized in image recognition. In this study, we apply the AlexNet architecture, which is a winner of Imagenet Large Scale Virtual Recognition Challenge (ILSVRC) 2012, to the forecast of daily solar flare occurrence using the MatConvNet software of MATLAB. Our input images are SOHO/MDI, EIT 195Å, and 304Å from January 1996 to December 2010, and output ones are yes or no of flare occurrence. We select training dataset from Jan 1996 to Dec 2000 and from Jan 2003 to Dec 2008. Testing dataset is chosen from Jan 2001 to Dec 2002 and from Jan 2009 to Dec 2010 in order to consider the solar cycle effect. In training dataset, we randomly select one fifth of training data for validation dataset to avoid the overfitting problem. Our model successfully forecasts the flare occurrence with about 0.90 probability of detection (POD) for common flares (C-, M-, and X-class). While POD of major flares (M- and X-class) forecasting is 0.96, false alarm rate (FAR) also scores relatively high(0.60). We also present several statistical parameters such as critical success index (CSI) and true skill statistics (TSS). Our model can immediately be applied to automatic forecasting service when image data are available.

Large-Scale Image Analytics Using Deep Learning

NASA Astrophysics Data System (ADS)

Ganguly, S.; Nemani, R. R.; Basu, S.; Mukhopadhyay, S.; Michaelis, A.; Votava, P.

2014-12-01

High resolution land cover classification maps are needed to increase the accuracy of current Land ecosystem and climate model outputs. Limited studies are in place that demonstrates the state-of-the-art in deriving very high resolution (VHR) land cover products. In addition, most methods heavily rely on commercial softwares that are difficult to scale given the region of study (e.g. continents to globe). Complexities in present approaches relate to (a) scalability of the algorithm, (b) large image data processing (compute and memory intensive), (c) computational cost, (d) massively parallel architecture, and (e) machine learning automation. In addition, VHR satellite datasets are of the order of terabytes and features extracted from these datasets are of the order of petabytes. In our present study, we have acquired the National Agricultural Imaging Program (NAIP) dataset for the Continental United States at a spatial resolution of 1-m. This data comes as image tiles (a total of quarter million image scenes with ~60 million pixels) and has a total size of ~100 terabytes for a single acquisition. Features extracted from the entire dataset would amount to ~8-10 petabytes. In our proposed approach, we have implemented a novel semi-automated machine learning algorithm rooted on the principles of "deep learning" to delineate the percentage of tree cover. In order to perform image analytics in such a granular system, it is mandatory to devise an intelligent archiving and query system for image retrieval, file structuring, metadata processing and filtering of all available image scenes. Using the Open NASA Earth Exchange (NEX) initiative, which is a partnership with Amazon Web Services (AWS), we have developed an end-to-end architecture for designing the database and the deep belief network (following the distbelief computing model) to solve a grand challenge of scaling this process across quarter million NAIP tiles that cover the entire Continental United States. The AWS core components that we use to solve this problem are DynamoDB along with S3 for database query and storage, ElastiCache shared memory architecture for image segmentation, Elastic Map Reduce (EMR) for image feature extraction, and the memory optimized Elastic Cloud Compute (EC2) for the learning algorithm.

Part-based deep representation for product tagging and search

NASA Astrophysics Data System (ADS)

Chen, Keqing

2017-06-01

Despite previous studies, tagging and indexing the product images remain challenging due to the large inner-class variation of the products. In the traditional methods, the quantized hand-crafted features such as SIFTs are extracted as the representation of the product images, which are not discriminative enough to handle the inner-class variation. For discriminative image representation, this paper firstly presents a novel deep convolutional neural networks (DCNNs) architect true pre-trained on a large-scale general image dataset. Compared to the traditional features, our DCNNs representation is of more discriminative power with fewer dimensions. Moreover, we incorporate the part-based model into the framework to overcome the negative effect of bad alignment and cluttered background and hence the descriptive ability of the deep representation is further enhanced. Finally, we collect and contribute a well-labeled shoe image database, i.e., the TBShoes, on which we apply the part-based deep representation for product image tagging and search, respectively. The experimental results highlight the advantages of the proposed part-based deep representation.

A Wavelet Polarization Decomposition Net Model for Polarimetric SAR Image Classification

NASA Astrophysics Data System (ADS)

He, Chu; Ou, Dan; Yang, Teng; Wu, Kun; Liao, Mingsheng; Chen, Erxue

2014-11-01

In this paper, a deep model based on wavelet texture has been proposed for Polarimetric Synthetic Aperture Radar (PolSAR) image classification inspired by recent successful deep learning method. Our model is supposed to learn powerful and informative representations to improve the generalization ability for the complex scene classification tasks. Given the influence of speckle noise in Polarimetric SAR image, wavelet polarization decomposition is applied first to obtain basic and discriminative texture features which are then embedded into a Deep Neural Network (DNN) in order to compose multi-layer higher representations. We demonstrate that the model can produce a powerful representation which can capture some untraceable information from Polarimetric SAR images and show a promising achievement in comparison with other traditional SAR image classification methods for the SAR image dataset.

Comparison Between NPP-VIIRS Aerosol Data Products and the MODIS AQUA Deep Blue Collection 6 Dataset Over Land

NASA Technical Reports Server (NTRS)

Sayer, Andrew M.; Hsu, N. C.; Bettenhausen, C.; Lee, J.; Kondragunta, S.

2013-01-01

Aerosols are small particles suspended in the atmosphere and have a variety of natural and man-made sources. Knowledge of aerosol optical depth (AOD), which is a measure of the amount of aerosol in the atmosphere, and its change over time, is important for multiple reasons. These include climate change, air quality (pollution) monitoring, monitoring hazards such as dust storms and volcanic ash, monitoring smoke from biomass burning, determining potential energy yields from solar plants, determining visibility at sea, estimating fertilization of oceans and rainforests by transported mineral dust, understanding changes in weather brought upon by the interaction of aerosols and clouds, and more. The Suomi-NPP satellite was launched late in 2011. The Visible Infrared Imaging Radiometer Suite (VIIRS) aboard Suomi-NPP is being used, among other things, to determine AOD. This study compares the VIIRS dataset to ground-based measurements of AOD, along with a state-of-the-art satellite AOD dataset (the new version of the Moderate Resolution Imaging Spectrometer Deep Blue algorithm) to assess its reliability. The Suomi-NPP satellite was launched late in 2011, carrying several instruments designed to continue the biogeophysical data records of current and previous satellite sensors. The Visible Infrared Imaging Radiometer Suite (VIIRS) aboard Suomi-NPP is being used, among other things, to determine aerosol optical depth (AOD), and related activities since launch have been focused towards validating and understanding this new dataset through comparisons with other satellite and ground-based products. The operational VIIRS AOD product is compared over land with AOD derived from Moderate Resolution Imaging Spectrometer (MODIS) observations using the Deep Blue (DB) algorithm from the forthcoming Collection 6 of MODIS data

Deep Convolutional Neural Networks for Endotracheal Tube Position and X-ray Image Classification: Challenges and Opportunities.

PubMed

Lakhani, Paras

2017-08-01

The goal of this study is to evaluate the efficacy of deep convolutional neural networks (DCNNs) in differentiating subtle, intermediate, and more obvious image differences in radiography. Three different datasets were created, which included presence/absence of the endotracheal (ET) tube (n = 300), low/normal position of the ET tube (n = 300), and chest/abdominal radiographs (n = 120). The datasets were split into training, validation, and test. Both untrained and pre-trained deep neural networks were employed, including AlexNet and GoogLeNet classifiers, using the Caffe framework. Data augmentation was performed for the presence/absence and low/normal ET tube datasets. Receiver operating characteristic (ROC), area under the curves (AUC), and 95% confidence intervals were calculated. Statistical differences of the AUCs were determined using a non-parametric approach. The pre-trained AlexNet and GoogLeNet classifiers had perfect accuracy (AUC 1.00) in differentiating chest vs. abdominal radiographs, using only 45 training cases. For more difficult datasets, including the presence/absence and low/normal position endotracheal tubes, more training cases, pre-trained networks, and data-augmentation approaches were helpful to increase accuracy. The best-performing network for classifying presence vs. absence of an ET tube was still very accurate with an AUC of 0.99. However, for the most difficult dataset, such as low vs. normal position of the endotracheal tube, DCNNs did not perform as well, but achieved a reasonable AUC of 0.81.

Multi-categorical deep learning neural network to classify retinal images: A pilot study employing small database.

PubMed

Choi, Joon Yul; Yoo, Tae Keun; Seo, Jeong Gi; Kwak, Jiyong; Um, Terry Taewoong; Rim, Tyler Hyungtaek

2017-01-01

Deep learning emerges as a powerful tool for analyzing medical images. Retinal disease detection by using computer-aided diagnosis from fundus image has emerged as a new method. We applied deep learning convolutional neural network by using MatConvNet for an automated detection of multiple retinal diseases with fundus photographs involved in STructured Analysis of the REtina (STARE) database. Dataset was built by expanding data on 10 categories, including normal retina and nine retinal diseases. The optimal outcomes were acquired by using a random forest transfer learning based on VGG-19 architecture. The classification results depended greatly on the number of categories. As the number of categories increased, the performance of deep learning models was diminished. When all 10 categories were included, we obtained results with an accuracy of 30.5%, relative classifier information (RCI) of 0.052, and Cohen's kappa of 0.224. Considering three integrated normal, background diabetic retinopathy, and dry age-related macular degeneration, the multi-categorical classifier showed accuracy of 72.8%, 0.283 RCI, and 0.577 kappa. In addition, several ensemble classifiers enhanced the multi-categorical classification performance. The transfer learning incorporated with ensemble classifier of clustering and voting approach presented the best performance with accuracy of 36.7%, 0.053 RCI, and 0.225 kappa in the 10 retinal diseases classification problem. First, due to the small size of datasets, the deep learning techniques in this study were ineffective to be applied in clinics where numerous patients suffering from various types of retinal disorders visit for diagnosis and treatment. Second, we found that the transfer learning incorporated with ensemble classifiers can improve the classification performance in order to detect multi-categorical retinal diseases. Further studies should confirm the effectiveness of algorithms with large datasets obtained from hospitals.

The Livermore Brain: Massive Deep Learning Networks Enabled by High Performance Computing

DOE Office of Scientific and Technical Information (OSTI.GOV)

Chen, Barry Y.

The proliferation of inexpensive sensor technologies like the ubiquitous digital image sensors has resulted in the collection and sharing of vast amounts of unsorted and unexploited raw data. Companies and governments who are able to collect and make sense of large datasets to help them make better decisions more rapidly will have a competitive advantage in the information era. Machine Learning technologies play a critical role for automating the data understanding process; however, to be maximally effective, useful intermediate representations of the data are required. These representations or “features” are transformations of the raw data into a form where patternsmore » are more easily recognized. Recent breakthroughs in Deep Learning have made it possible to learn these features from large amounts of labeled data. The focus of this project is to develop and extend Deep Learning algorithms for learning features from vast amounts of unlabeled data and to develop the HPC neural network training platform to support the training of massive network models. This LDRD project succeeded in developing new unsupervised feature learning algorithms for images and video and created a scalable neural network training toolkit for HPC. Additionally, this LDRD helped create the world’s largest freely-available image and video dataset supporting open multimedia research and used this dataset for training our deep neural networks. This research helped LLNL capture several work-for-others (WFO) projects, attract new talent, and establish collaborations with leading academic and commercial partners. Finally, this project demonstrated the successful training of the largest unsupervised image neural network using HPC resources and helped establish LLNL leadership at the intersection of Machine Learning and HPC research.« less

Fine-grained leukocyte classification with deep residual learning for microscopic images.

PubMed

Qin, Feiwei; Gao, Nannan; Peng, Yong; Wu, Zizhao; Shen, Shuying; Grudtsin, Artur

2018-08-01

Leukocyte classification and cytometry have wide applications in medical domain, previous researches usually exploit machine learning techniques to classify leukocytes automatically. However, constrained by the past development of machine learning techniques, for example, extracting distinctive features from raw microscopic images are difficult, the widely used SVM classifier only has relative few parameters to tune, these methods cannot efficiently handle fine-grained classification cases when the white blood cells have up to 40 categories. Based on deep learning theory, a systematic study is conducted on finer leukocyte classification in this paper. A deep residual neural network based leukocyte classifier is constructed at first, which can imitate the domain expert's cell recognition process, and extract salient features robustly and automatically. Then the deep neural network classifier's topology is adjusted according to the prior knowledge of white blood cell test. After that the microscopic image dataset with almost one hundred thousand labeled leukocytes belonging to 40 categories is built, and combined training strategies are adopted to make the designed classifier has good generalization ability. The proposed deep residual neural network based classifier was tested on microscopic image dataset with 40 leukocyte categories. It achieves top-1 accuracy of 77.80%, top-5 accuracy of 98.75% during the training procedure. The average accuracy on the test set is nearly 76.84%. This paper presents a fine-grained leukocyte classification method for microscopic images, based on deep residual learning theory and medical domain knowledge. Experimental results validate the feasibility and effectiveness of our approach. Extended experiments support that the fine-grained leukocyte classifier could be used in real medical applications, assist doctors in diagnosing diseases, reduce human power significantly. Copyright © 2018 Elsevier B.V. All rights reserved.

Automatically identifying, counting, and describing wild animals in camera-trap images with deep learning.

PubMed

Norouzzadeh, Mohammad Sadegh; Nguyen, Anh; Kosmala, Margaret; Swanson, Alexandra; Palmer, Meredith S; Packer, Craig; Clune, Jeff

2018-06-19

Having accurate, detailed, and up-to-date information about the location and behavior of animals in the wild would improve our ability to study and conserve ecosystems. We investigate the ability to automatically, accurately, and inexpensively collect such data, which could help catalyze the transformation of many fields of ecology, wildlife biology, zoology, conservation biology, and animal behavior into "big data" sciences. Motion-sensor "camera traps" enable collecting wildlife pictures inexpensively, unobtrusively, and frequently. However, extracting information from these pictures remains an expensive, time-consuming, manual task. We demonstrate that such information can be automatically extracted by deep learning, a cutting-edge type of artificial intelligence. We train deep convolutional neural networks to identify, count, and describe the behaviors of 48 species in the 3.2 million-image Snapshot Serengeti dataset. Our deep neural networks automatically identify animals with >93.8% accuracy, and we expect that number to improve rapidly in years to come. More importantly, if our system classifies only images it is confident about, our system can automate animal identification for 99.3% of the data while still performing at the same 96.6% accuracy as that of crowdsourced teams of human volunteers, saving >8.4 y (i.e., >17,000 h at 40 h/wk) of human labeling effort on this 3.2 million-image dataset. Those efficiency gains highlight the importance of using deep neural networks to automate data extraction from camera-trap images, reducing a roadblock for this widely used technology. Our results suggest that deep learning could enable the inexpensive, unobtrusive, high-volume, and even real-time collection of a wealth of information about vast numbers of animals in the wild. Copyright © 2018 the Author(s). Published by PNAS.

HCP: A Flexible CNN Framework for Multi-label Image Classification.

PubMed

Wei, Yunchao; Xia, Wei; Lin, Min; Huang, Junshi; Ni, Bingbing; Dong, Jian; Zhao, Yao; Yan, Shuicheng

2015-10-26

Convolutional Neural Network (CNN) has demonstrated promising performance in single-label image classification tasks. However, how CNN best copes with multi-label images still remains an open problem, mainly due to the complex underlying object layouts and insufficient multi-label training images. In this work, we propose a flexible deep CNN infrastructure, called Hypotheses-CNN-Pooling (HCP), where an arbitrary number of object segment hypotheses are taken as the inputs, then a shared CNN is connected with each hypothesis, and finally the CNN output results from different hypotheses are aggregated with max pooling to produce the ultimate multi-label predictions. Some unique characteristics of this flexible deep CNN infrastructure include: 1) no ground-truth bounding box information is required for training; 2) the whole HCP infrastructure is robust to possibly noisy and/or redundant hypotheses; 3) the shared CNN is flexible and can be well pre-trained with a large-scale single-label image dataset, e.g., ImageNet; and 4) it may naturally output multi-label prediction results. Experimental results on Pascal VOC 2007 and VOC 2012 multi-label image datasets well demonstrate the superiority of the proposed HCP infrastructure over other state-of-the-arts. In particular, the mAP reaches 90.5% by HCP only and 93.2% after the fusion with our complementary result in [44] based on hand-crafted features on the VOC 2012 dataset.

Assessing microscope image focus quality with deep learning.

PubMed

Yang, Samuel J; Berndl, Marc; Michael Ando, D; Barch, Mariya; Narayanaswamy, Arunachalam; Christiansen, Eric; Hoyer, Stephan; Roat, Chris; Hung, Jane; Rueden, Curtis T; Shankar, Asim; Finkbeiner, Steven; Nelson, Philip

2018-03-15

Large image datasets acquired on automated microscopes typically have some fraction of low quality, out-of-focus images, despite the use of hardware autofocus systems. Identification of these images using automated image analysis with high accuracy is important for obtaining a clean, unbiased image dataset. Complicating this task is the fact that image focus quality is only well-defined in foreground regions of images, and as a result, most previous approaches only enable a computation of the relative difference in quality between two or more images, rather than an absolute measure of quality. We present a deep neural network model capable of predicting an absolute measure of image focus on a single image in isolation, without any user-specified parameters. The model operates at the image-patch level, and also outputs a measure of prediction certainty, enabling interpretable predictions. The model was trained on only 384 in-focus Hoechst (nuclei) stain images of U2OS cells, which were synthetically defocused to one of 11 absolute defocus levels during training. The trained model can generalize on previously unseen real Hoechst stain images, identifying the absolute image focus to within one defocus level (approximately 3 pixel blur diameter difference) with 95% accuracy. On a simpler binary in/out-of-focus classification task, the trained model outperforms previous approaches on both Hoechst and Phalloidin (actin) stain images (F-scores of 0.89 and 0.86, respectively over 0.84 and 0.83), despite only having been presented Hoechst stain images during training. Lastly, we observe qualitatively that the model generalizes to two additional stains, Hoechst and Tubulin, of an unseen cell type (Human MCF-7) acquired on a different instrument. Our deep neural network enables classification of out-of-focus microscope images with both higher accuracy and greater precision than previous approaches via interpretable patch-level focus and certainty predictions. The use of synthetically defocused images precludes the need for a manually annotated training dataset. The model also generalizes to different image and cell types. The framework for model training and image prediction is available as a free software library and the pre-trained model is available for immediate use in Fiji (ImageJ) and CellProfiler.

The MIND PALACE: A Multi-Spectral Imaging and Spectroscopy Database for Planetary Science

NASA Astrophysics Data System (ADS)

Eshelman, E.; Doloboff, I.; Hara, E. K.; Uckert, K.; Sapers, H. M.; Abbey, W.; Beegle, L. W.; Bhartia, R.

2017-12-01

The Multi-Instrument Database (MIND) is the web-based home to a well-characterized set of analytical data collected by a suite of deep-UV fluorescence/Raman instruments built at the Jet Propulsion Laboratory (JPL). Samples derive from a growing body of planetary surface analogs, mineral and microbial standards, meteorites, spacecraft materials, and other astrobiologically relevant materials. In addition to deep-UV spectroscopy, datasets stored in MIND are obtained from a variety of analytical techniques obtained over multiple spatial and spectral scales including electron microscopy, optical microscopy, infrared spectroscopy, X-ray fluorescence, and direct fluorescence imaging. Multivariate statistical analysis techniques, primarily Principal Component Analysis (PCA), are used to guide interpretation of these large multi-analytical spectral datasets. Spatial co-referencing of integrated spectral/visual maps is performed using QGIS (geographic information system software). Georeferencing techniques transform individual instrument data maps into a layered co-registered data cube for analysis across spectral and spatial scales. The body of data in MIND is intended to serve as a permanent, reliable, and expanding database of deep-UV spectroscopy datasets generated by this unique suite of JPL-based instruments on samples of broad planetary science interest.

Lesion Detection in CT Images Using Deep Learning Semantic Segmentation Technique

NASA Astrophysics Data System (ADS)

Kalinovsky, A.; Liauchuk, V.; Tarasau, A.

2017-05-01

In this paper, the problem of automatic detection of tuberculosis lesion on 3D lung CT images is considered as a benchmark for testing out algorithms based on a modern concept of Deep Learning. For training and testing of the algorithms a domestic dataset of 338 3D CT scans of tuberculosis patients with manually labelled lesions was used. The algorithms which are based on using Deep Convolutional Networks were implemented and applied in three different ways including slice-wise lesion detection in 2D images using semantic segmentation, slice-wise lesion detection in 2D images using sliding window technique as well as straightforward detection of lesions via semantic segmentation in whole 3D CT scans. The algorithms demonstrate superior performance compared to algorithms based on conventional image analysis methods.

High resolution depth reconstruction from monocular images and sparse point clouds using deep convolutional neural network

NASA Astrophysics Data System (ADS)

Dimitrievski, Martin; Goossens, Bart; Veelaert, Peter; Philips, Wilfried

2017-09-01

Understanding the 3D structure of the environment is advantageous for many tasks in the field of robotics and autonomous vehicles. From the robot's point of view, 3D perception is often formulated as a depth image reconstruction problem. In the literature, dense depth images are often recovered deterministically from stereo image disparities. Other systems use an expensive LiDAR sensor to produce accurate, but semi-sparse depth images. With the advent of deep learning there have also been attempts to estimate depth by only using monocular images. In this paper we combine the best of the two worlds, focusing on a combination of monocular images and low cost LiDAR point clouds. We explore the idea that very sparse depth information accurately captures the global scene structure while variations in image patches can be used to reconstruct local depth to a high resolution. The main contribution of this paper is a supervised learning depth reconstruction system based on a deep convolutional neural network. The network is trained on RGB image patches reinforced with sparse depth information and the output is a depth estimate for each pixel. Using image and point cloud data from the KITTI vision dataset we are able to learn a correspondence between local RGB information and local depth, while at the same time preserving the global scene structure. Our results are evaluated on sequences from the KITTI dataset and our own recordings using a low cost camera and LiDAR setup.

Application of deep learning to the classification of images from colposcopy.

PubMed

Sato, Masakazu; Horie, Koji; Hara, Aki; Miyamoto, Yuichiro; Kurihara, Kazuko; Tomio, Kensuke; Yokota, Harushige

2018-03-01

The objective of the present study was to investigate whether deep learning could be applied successfully to the classification of images from colposcopy. For this purpose, a total of 158 patients who underwent conization were enrolled, and medical records and data from the gynecological oncology database were retrospectively reviewed. Deep learning was performed with the Keras neural network and TensorFlow libraries. Using preoperative images from colposcopy as the input data and deep learning technology, the patients were classified into three groups [severe dysplasia, carcinoma in situ (CIS) and invasive cancer (IC)]. A total of 485 images were obtained for the analysis, of which 142 images were of severe dysplasia (2.9 images/patient), 257 were of CIS (3.3 images/patient), and 86 were of IC (4.1 images/patient). Of these, 233 images were captured with a green filter, and the remaining 252 were captured without a green filter. Following the application of L2 regularization, L1 regularization, dropout and data augmentation, the accuracy of the validation dataset was ~50%. Although the present study is preliminary, the results indicated that deep learning may be applied to classify colposcopy images.

Application of deep learning to the classification of images from colposcopy

PubMed Central

Sato, Masakazu; Horie, Koji; Hara, Aki; Miyamoto, Yuichiro; Kurihara, Kazuko; Tomio, Kensuke; Yokota, Harushige

2018-01-01

The objective of the present study was to investigate whether deep learning could be applied successfully to the classification of images from colposcopy. For this purpose, a total of 158 patients who underwent conization were enrolled, and medical records and data from the gynecological oncology database were retrospectively reviewed. Deep learning was performed with the Keras neural network and TensorFlow libraries. Using preoperative images from colposcopy as the input data and deep learning technology, the patients were classified into three groups [severe dysplasia, carcinoma in situ (CIS) and invasive cancer (IC)]. A total of 485 images were obtained for the analysis, of which 142 images were of severe dysplasia (2.9 images/patient), 257 were of CIS (3.3 images/patient), and 86 were of IC (4.1 images/patient). Of these, 233 images were captured with a green filter, and the remaining 252 were captured without a green filter. Following the application of L2 regularization, L1 regularization, dropout and data augmentation, the accuracy of the validation dataset was ~50%. Although the present study is preliminary, the results indicated that deep learning may be applied to classify colposcopy images. PMID:29456725

Accurate segmentation of lung fields on chest radiographs using deep convolutional networks

NASA Astrophysics Data System (ADS)

Arbabshirani, Mohammad R.; Dallal, Ahmed H.; Agarwal, Chirag; Patel, Aalpan; Moore, Gregory

2017-02-01

Accurate segmentation of lung fields on chest radiographs is the primary step for computer-aided detection of various conditions such as lung cancer and tuberculosis. The size, shape and texture of lung fields are key parameters for chest X-ray (CXR) based lung disease diagnosis in which the lung field segmentation is a significant primary step. Although many methods have been proposed for this problem, lung field segmentation remains as a challenge. In recent years, deep learning has shown state of the art performance in many visual tasks such as object detection, image classification and semantic image segmentation. In this study, we propose a deep convolutional neural network (CNN) framework for segmentation of lung fields. The algorithm was developed and tested on 167 clinical posterior-anterior (PA) CXR images collected retrospectively from picture archiving and communication system (PACS) of Geisinger Health System. The proposed multi-scale network is composed of five convolutional and two fully connected layers. The framework achieved IOU (intersection over union) of 0.96 on the testing dataset as compared to manual segmentation. The suggested framework outperforms state of the art registration-based segmentation by a significant margin. To our knowledge, this is the first deep learning based study of lung field segmentation on CXR images developed on a heterogeneous clinical dataset. The results suggest that convolutional neural networks could be employed reliably for lung field segmentation.

Beyond Retinal Layers: A Deep Voting Model for Automated Geographic Atrophy Segmentation in SD-OCT Images

PubMed Central

Ji, Zexuan; Chen, Qiang; Niu, Sijie; Leng, Theodore; Rubin, Daniel L.

2018-01-01

Purpose To automatically and accurately segment geographic atrophy (GA) in spectral-domain optical coherence tomography (SD-OCT) images by constructing a voting system with deep neural networks without the use of retinal layer segmentation. Methods An automatic GA segmentation method for SD-OCT images based on the deep network was constructed. The structure of the deep network was composed of five layers, including one input layer, three hidden layers, and one output layer. During the training phase, the labeled A-scans with 1024 features were directly fed into the network as the input layer to obtain the deep representations. Then a soft-max classifier was trained to determine the label of each individual pixel. Finally, a voting decision strategy was used to refine the segmentation results among 10 trained models. Results Two image data sets with GA were used to evaluate the model. For the first dataset, our algorithm obtained a mean overlap ratio (OR) 86.94% ± 8.75%, absolute area difference (AAD) 11.49% ± 11.50%, and correlation coefficients (CC) 0.9857; for the second dataset, the mean OR, AAD, and CC of the proposed method were 81.66% ± 10.93%, 8.30% ± 9.09%, and 0.9952, respectively. The proposed algorithm was capable of improving over 5% and 10% segmentation accuracy, respectively, when compared with several state-of-the-art algorithms on two data sets. Conclusions Without retinal layer segmentation, the proposed algorithm could produce higher segmentation accuracy and was more stable when compared with state-of-the-art methods that relied on retinal layer segmentation results. Our model may provide reliable GA segmentations from SD-OCT images and be useful in the clinical diagnosis of advanced nonexudative AMD. Translational Relevance Based on the deep neural networks, this study presents an accurate GA segmentation method for SD-OCT images without using any retinal layer segmentation results, and may contribute to improved understanding of advanced nonexudative AMD. PMID:29302382

Beyond Retinal Layers: A Deep Voting Model for Automated Geographic Atrophy Segmentation in SD-OCT Images.

PubMed

Ji, Zexuan; Chen, Qiang; Niu, Sijie; Leng, Theodore; Rubin, Daniel L

2018-01-01

To automatically and accurately segment geographic atrophy (GA) in spectral-domain optical coherence tomography (SD-OCT) images by constructing a voting system with deep neural networks without the use of retinal layer segmentation. An automatic GA segmentation method for SD-OCT images based on the deep network was constructed. The structure of the deep network was composed of five layers, including one input layer, three hidden layers, and one output layer. During the training phase, the labeled A-scans with 1024 features were directly fed into the network as the input layer to obtain the deep representations. Then a soft-max classifier was trained to determine the label of each individual pixel. Finally, a voting decision strategy was used to refine the segmentation results among 10 trained models. Two image data sets with GA were used to evaluate the model. For the first dataset, our algorithm obtained a mean overlap ratio (OR) 86.94% ± 8.75%, absolute area difference (AAD) 11.49% ± 11.50%, and correlation coefficients (CC) 0.9857; for the second dataset, the mean OR, AAD, and CC of the proposed method were 81.66% ± 10.93%, 8.30% ± 9.09%, and 0.9952, respectively. The proposed algorithm was capable of improving over 5% and 10% segmentation accuracy, respectively, when compared with several state-of-the-art algorithms on two data sets. Without retinal layer segmentation, the proposed algorithm could produce higher segmentation accuracy and was more stable when compared with state-of-the-art methods that relied on retinal layer segmentation results. Our model may provide reliable GA segmentations from SD-OCT images and be useful in the clinical diagnosis of advanced nonexudative AMD. Based on the deep neural networks, this study presents an accurate GA segmentation method for SD-OCT images without using any retinal layer segmentation results, and may contribute to improved understanding of advanced nonexudative AMD.

Deep Convolutional Neural Network-Based Early Automated Detection of Diabetic Retinopathy Using Fundus Image.

PubMed

Xu, Kele; Feng, Dawei; Mi, Haibo

2017-11-23

The automatic detection of diabetic retinopathy is of vital importance, as it is the main cause of irreversible vision loss in the working-age population in the developed world. The early detection of diabetic retinopathy occurrence can be very helpful for clinical treatment; although several different feature extraction approaches have been proposed, the classification task for retinal images is still tedious even for those trained clinicians. Recently, deep convolutional neural networks have manifested superior performance in image classification compared to previous handcrafted feature-based image classification methods. Thus, in this paper, we explored the use of deep convolutional neural network methodology for the automatic classification of diabetic retinopathy using color fundus image, and obtained an accuracy of 94.5% on our dataset, outperforming the results obtained by using classical approaches.

Automatic segmentation of the prostate on CT images using deep learning and multi-atlas fusion

NASA Astrophysics Data System (ADS)

Ma, Ling; Guo, Rongrong; Zhang, Guoyi; Tade, Funmilayo; Schuster, David M.; Nieh, Peter; Master, Viraj; Fei, Baowei

2017-02-01

Automatic segmentation of the prostate on CT images has many applications in prostate cancer diagnosis and therapy. However, prostate CT image segmentation is challenging because of the low contrast of soft tissue on CT images. In this paper, we propose an automatic segmentation method by combining a deep learning method and multi-atlas refinement. First, instead of segmenting the whole image, we extract the region of interesting (ROI) to delete irrelevant regions. Then, we use the convolutional neural networks (CNN) to learn the deep features for distinguishing the prostate pixels from the non-prostate pixels in order to obtain the preliminary segmentation results. CNN can automatically learn the deep features adapting to the data, which are different from some handcrafted features. Finally, we select some similar atlases to refine the initial segmentation results. The proposed method has been evaluated on a dataset of 92 prostate CT images. Experimental results show that our method achieved a Dice similarity coefficient of 86.80% as compared to the manual segmentation. The deep learning based method can provide a useful tool for automatic segmentation of the prostate on CT images and thus can have a variety of clinical applications.

DeepMitosis: Mitosis detection via deep detection, verification and segmentation networks.

PubMed

Li, Chao; Wang, Xinggang; Liu, Wenyu; Latecki, Longin Jan

2018-04-01

Mitotic count is a critical predictor of tumor aggressiveness in the breast cancer diagnosis. Nowadays mitosis counting is mainly performed by pathologists manually, which is extremely arduous and time-consuming. In this paper, we propose an accurate method for detecting the mitotic cells from histopathological slides using a novel multi-stage deep learning framework. Our method consists of a deep segmentation network for generating mitosis region when only a weak label is given (i.e., only the centroid pixel of mitosis is annotated), an elaborately designed deep detection network for localizing mitosis by using contextual region information, and a deep verification network for improving detection accuracy by removing false positives. We validate the proposed deep learning method on two widely used Mitosis Detection in Breast Cancer Histological Images (MITOSIS) datasets. Experimental results show that we can achieve the highest F-score on the MITOSIS dataset from ICPR 2012 grand challenge merely using the deep detection network. For the ICPR 2014 MITOSIS dataset that only provides the centroid location of mitosis, we employ the segmentation model to estimate the bounding box annotation for training the deep detection network. We also apply the verification model to eliminate some false positives produced from the detection model. By fusing scores of the detection and verification models, we achieve the state-of-the-art results. Moreover, our method is very fast with GPU computing, which makes it feasible for clinical practice. Copyright © 2018 Elsevier B.V. All rights reserved.

Geographical topic learning for social images with a deep neural network

NASA Astrophysics Data System (ADS)

Feng, Jiangfan; Xu, Xin

2017-03-01

The use of geographical tagging in social-media images is becoming a part of image metadata and a great interest for geographical information science. It is well recognized that geographical topic learning is crucial for geographical annotation. Existing methods usually exploit geographical characteristics using image preprocessing, pixel-based classification, and feature recognition. How to effectively exploit the high-level semantic feature and underlying correlation among different types of contents is a crucial task for geographical topic learning. Deep learning (DL) has recently demonstrated robust capabilities for image tagging and has been introduced into geoscience. It extracts high-level features computed from a whole image component, where the cluttered background may dominate spatial features in the deep representation. Therefore, a method of spatial-attentional DL for geographical topic learning is provided and we can regard it as a special case of DL combined with various deep networks and tuning tricks. Results demonstrated that the method is discriminative for different types of geographical topic learning. In addition, it outperforms other sequential processing models in a tagging task for a geographical image dataset.

Epithelium-Stroma Classification via Convolutional Neural Networks and Unsupervised Domain Adaptation in Histopathological Images.

PubMed

Huang, Yue; Zheng, Han; Liu, Chi; Ding, Xinghao; Rohde, Gustavo K

2017-11-01

Epithelium-stroma classification is a necessary preprocessing step in histopathological image analysis. Current deep learning based recognition methods for histology data require collection of large volumes of labeled data in order to train a new neural network when there are changes to the image acquisition procedure. However, it is extremely expensive for pathologists to manually label sufficient volumes of data for each pathology study in a professional manner, which results in limitations in real-world applications. A very simple but effective deep learning method, that introduces the concept of unsupervised domain adaptation to a simple convolutional neural network (CNN), has been proposed in this paper. Inspired by transfer learning, our paper assumes that the training data and testing data follow different distributions, and there is an adaptation operation to more accurately estimate the kernels in CNN in feature extraction, in order to enhance performance by transferring knowledge from labeled data in source domain to unlabeled data in target domain. The model has been evaluated using three independent public epithelium-stroma datasets by cross-dataset validations. The experimental results demonstrate that for epithelium-stroma classification, the proposed framework outperforms the state-of-the-art deep neural network model, and it also achieves better performance than other existing deep domain adaptation methods. The proposed model can be considered to be a better option for real-world applications in histopathological image analysis, since there is no longer a requirement for large-scale labeled data in each specified domain.

Chest x-ray generation and data augmentation for cardiovascular abnormality classification

NASA Astrophysics Data System (ADS)

Madani, Ali; Moradi, Mehdi; Karargyris, Alexandros; Syeda-Mahmood, Tanveer

2018-03-01

Medical imaging datasets are limited in size due to privacy issues and the high cost of obtaining annotations. Augmentation is a widely used practice in deep learning to enrich the data in data-limited scenarios and to avoid overfitting. However, standard augmentation methods that produce new examples of data by varying lighting, field of view, and spatial rigid transformations do not capture the biological variance of medical imaging data and could result in unrealistic images. Generative adversarial networks (GANs) provide an avenue to understand the underlying structure of image data which can then be utilized to generate new realistic samples. In this work, we investigate the use of GANs for producing chest X-ray images to augment a dataset. This dataset is then used to train a convolutional neural network to classify images for cardiovascular abnormalities. We compare our augmentation strategy with traditional data augmentation and show higher accuracy for normal vs abnormal classification in chest X-rays.

Segmenting Brain Tissues from Chinese Visible Human Dataset by Deep-Learned Features with Stacked Autoencoder

PubMed Central

Zhao, Guangjun; Wang, Xuchu; Niu, Yanmin; Tan, Liwen; Zhang, Shao-Xiang

2016-01-01

Cryosection brain images in Chinese Visible Human (CVH) dataset contain rich anatomical structure information of tissues because of its high resolution (e.g., 0.167 mm per pixel). Fast and accurate segmentation of these images into white matter, gray matter, and cerebrospinal fluid plays a critical role in analyzing and measuring the anatomical structures of human brain. However, most existing automated segmentation methods are designed for computed tomography or magnetic resonance imaging data, and they may not be applicable for cryosection images due to the imaging difference. In this paper, we propose a supervised learning-based CVH brain tissues segmentation method that uses stacked autoencoder (SAE) to automatically learn the deep feature representations. Specifically, our model includes two successive parts where two three-layer SAEs take image patches as input to learn the complex anatomical feature representation, and then these features are sent to Softmax classifier for inferring the labels. Experimental results validated the effectiveness of our method and showed that it outperformed four other classical brain tissue detection strategies. Furthermore, we reconstructed three-dimensional surfaces of these tissues, which show their potential in exploring the high-resolution anatomical structures of human brain. PMID:27057543

Segmenting Brain Tissues from Chinese Visible Human Dataset by Deep-Learned Features with Stacked Autoencoder.

PubMed

Zhao, Guangjun; Wang, Xuchu; Niu, Yanmin; Tan, Liwen; Zhang, Shao-Xiang

2016-01-01

Cryosection brain images in Chinese Visible Human (CVH) dataset contain rich anatomical structure information of tissues because of its high resolution (e.g., 0.167 mm per pixel). Fast and accurate segmentation of these images into white matter, gray matter, and cerebrospinal fluid plays a critical role in analyzing and measuring the anatomical structures of human brain. However, most existing automated segmentation methods are designed for computed tomography or magnetic resonance imaging data, and they may not be applicable for cryosection images due to the imaging difference. In this paper, we propose a supervised learning-based CVH brain tissues segmentation method that uses stacked autoencoder (SAE) to automatically learn the deep feature representations. Specifically, our model includes two successive parts where two three-layer SAEs take image patches as input to learn the complex anatomical feature representation, and then these features are sent to Softmax classifier for inferring the labels. Experimental results validated the effectiveness of our method and showed that it outperformed four other classical brain tissue detection strategies. Furthermore, we reconstructed three-dimensional surfaces of these tissues, which show their potential in exploring the high-resolution anatomical structures of human brain.

Multi-categorical deep learning neural network to classify retinal images: A pilot study employing small database

PubMed Central

Seo, Jeong Gi; Kwak, Jiyong; Um, Terry Taewoong; Rim, Tyler Hyungtaek

2017-01-01

Deep learning emerges as a powerful tool for analyzing medical images. Retinal disease detection by using computer-aided diagnosis from fundus image has emerged as a new method. We applied deep learning convolutional neural network by using MatConvNet for an automated detection of multiple retinal diseases with fundus photographs involved in STructured Analysis of the REtina (STARE) database. Dataset was built by expanding data on 10 categories, including normal retina and nine retinal diseases. The optimal outcomes were acquired by using a random forest transfer learning based on VGG-19 architecture. The classification results depended greatly on the number of categories. As the number of categories increased, the performance of deep learning models was diminished. When all 10 categories were included, we obtained results with an accuracy of 30.5%, relative classifier information (RCI) of 0.052, and Cohen’s kappa of 0.224. Considering three integrated normal, background diabetic retinopathy, and dry age-related macular degeneration, the multi-categorical classifier showed accuracy of 72.8%, 0.283 RCI, and 0.577 kappa. In addition, several ensemble classifiers enhanced the multi-categorical classification performance. The transfer learning incorporated with ensemble classifier of clustering and voting approach presented the best performance with accuracy of 36.7%, 0.053 RCI, and 0.225 kappa in the 10 retinal diseases classification problem. First, due to the small size of datasets, the deep learning techniques in this study were ineffective to be applied in clinics where numerous patients suffering from various types of retinal disorders visit for diagnosis and treatment. Second, we found that the transfer learning incorporated with ensemble classifiers can improve the classification performance in order to detect multi-categorical retinal diseases. Further studies should confirm the effectiveness of algorithms with large datasets obtained from hospitals. PMID:29095872

Automatical and accurate segmentation of cerebral tissues in fMRI dataset with combination of image processing and deep learning

NASA Astrophysics Data System (ADS)

Kong, Zhenglun; Luo, Junyi; Xu, Shengpu; Li, Ting

2018-02-01

Image segmentation plays an important role in medical science. One application is multimodality imaging, especially the fusion of structural imaging with functional imaging, which includes CT, MRI and new types of imaging technology such as optical imaging to obtain functional images. The fusion process require precisely extracted structural information, in order to register the image to it. Here we used image enhancement, morphometry methods to extract the accurate contours of different tissues such as skull, cerebrospinal fluid (CSF), grey matter (GM) and white matter (WM) on 5 fMRI head image datasets. Then we utilized convolutional neural network to realize automatic segmentation of images in deep learning way. Such approach greatly reduced the processing time compared to manual and semi-automatic segmentation and is of great importance in improving speed and accuracy as more and more samples being learned. The contours of the borders of different tissues on all images were accurately extracted and 3D visualized. This can be used in low-level light therapy and optical simulation software such as MCVM. We obtained a precise three-dimensional distribution of brain, which offered doctors and researchers quantitative volume data and detailed morphological characterization for personal precise medicine of Cerebral atrophy/expansion. We hope this technique can bring convenience to visualization medical and personalized medicine.

BreakingNews: Article Annotation by Image and Text Processing.

PubMed

Ramisa, Arnau; Yan, Fei; Moreno-Noguer, Francesc; Mikolajczyk, Krystian

2018-05-01

Building upon recent Deep Neural Network architectures, current approaches lying in the intersection of Computer Vision and Natural Language Processing have achieved unprecedented breakthroughs in tasks like automatic captioning or image retrieval. Most of these learning methods, though, rely on large training sets of images associated with human annotations that specifically describe the visual content. In this paper we propose to go a step further and explore the more complex cases where textual descriptions are loosely related to the images. We focus on the particular domain of news articles in which the textual content often expresses connotative and ambiguous relations that are only suggested but not directly inferred from images. We introduce an adaptive CNN architecture that shares most of the structure for multiple tasks including source detection, article illustration and geolocation of articles. Deep Canonical Correlation Analysis is deployed for article illustration, and a new loss function based on Great Circle Distance is proposed for geolocation. Furthermore, we present BreakingNews, a novel dataset with approximately 100K news articles including images, text and captions, and enriched with heterogeneous meta-data (such as GPS coordinates and user comments). We show this dataset to be appropriate to explore all aforementioned problems, for which we provide a baseline performance using various Deep Learning architectures, and different representations of the textual and visual features. We report very promising results and bring to light several limitations of current state-of-the-art in this kind of domain, which we hope will help spur progress in the field.

Saliency Detection and Deep Learning-Based Wildfire Identification in UAV Imagery.

PubMed

Zhao, Yi; Ma, Jiale; Li, Xiaohui; Zhang, Jie

2018-02-27

An unmanned aerial vehicle (UAV) equipped with global positioning systems (GPS) can provide direct georeferenced imagery, mapping an area with high resolution. So far, the major difficulty in wildfire image classification is the lack of unified identification marks, the fire features of color, shape, texture (smoke, flame, or both) and background can vary significantly from one scene to another. Deep learning (e.g., DCNN for Deep Convolutional Neural Network) is very effective in high-level feature learning, however, a substantial amount of training images dataset is obligatory in optimizing its weights value and coefficients. In this work, we proposed a new saliency detection algorithm for fast location and segmentation of core fire area in aerial images. As the proposed method can effectively avoid feature loss caused by direct resizing; it is used in data augmentation and formation of a standard fire image dataset 'UAV_Fire'. A 15-layered self-learning DCNN architecture named 'Fire_Net' is then presented as a self-learning fire feature exactor and classifier. We evaluated different architectures and several key parameters (drop out ratio, batch size, etc.) of the DCNN model regarding its validation accuracy. The proposed architecture outperformed previous methods by achieving an overall accuracy of 98%. Furthermore, 'Fire_Net' guarantied an average processing speed of 41.5 ms per image for real-time wildfire inspection. To demonstrate its practical utility, Fire_Net is tested on 40 sampled images in wildfire news reports and all of them have been accurately identified.

Saliency Detection and Deep Learning-Based Wildfire Identification in UAV Imagery

PubMed Central

Zhao, Yi; Ma, Jiale; Li, Xiaohui

2018-01-01

An unmanned aerial vehicle (UAV) equipped with global positioning systems (GPS) can provide direct georeferenced imagery, mapping an area with high resolution. So far, the major difficulty in wildfire image classification is the lack of unified identification marks, the fire features of color, shape, texture (smoke, flame, or both) and background can vary significantly from one scene to another. Deep learning (e.g., DCNN for Deep Convolutional Neural Network) is very effective in high-level feature learning, however, a substantial amount of training images dataset is obligatory in optimizing its weights value and coefficients. In this work, we proposed a new saliency detection algorithm for fast location and segmentation of core fire area in aerial images. As the proposed method can effectively avoid feature loss caused by direct resizing; it is used in data augmentation and formation of a standard fire image dataset ‘UAV_Fire’. A 15-layered self-learning DCNN architecture named ‘Fire_Net’ is then presented as a self-learning fire feature exactor and classifier. We evaluated different architectures and several key parameters (drop out ratio, batch size, etc.) of the DCNN model regarding its validation accuracy. The proposed architecture outperformed previous methods by achieving an overall accuracy of 98%. Furthermore, ‘Fire_Net’ guarantied an average processing speed of 41.5 ms per image for real-time wildfire inspection. To demonstrate its practical utility, Fire_Net is tested on 40 sampled images in wildfire news reports and all of them have been accurately identified. PMID:29495504

TuMore: generation of synthetic brain tumor MRI data for deep learning based segmentation approaches

NASA Astrophysics Data System (ADS)

Lindner, Lydia; Pfarrkirchner, Birgit; Gsaxner, Christina; Schmalstieg, Dieter; Egger, Jan

2018-03-01

Accurate segmentation and measurement of brain tumors plays an important role in clinical practice and research, as it is critical for treatment planning and monitoring of tumor growth. However, brain tumor segmentation is one of the most challenging tasks in medical image analysis. Since manual segmentations are subjective, time consuming and neither accurate nor reliable, there exists a need for objective, robust and fast automated segmentation methods that provide competitive performance. Therefore, deep learning based approaches are gaining interest in the field of medical image segmentation. When the training data set is large enough, deep learning approaches can be extremely effective, but in domains like medicine, only limited data is available in the majority of cases. Due to this reason, we propose a method that allows to create a large dataset of brain MRI (Magnetic Resonance Imaging) images containing synthetic brain tumors - glioblastomas more specifically - and the corresponding ground truth, that can be subsequently used to train deep neural networks.

Deep and Structured Robust Information Theoretic Learning for Image Analysis.

PubMed

Deng, Yue; Bao, Feng; Deng, Xuesong; Wang, Ruiping; Kong, Youyong; Dai, Qionghai

2016-07-07

This paper presents a robust information theoretic (RIT) model to reduce the uncertainties, i.e. missing and noisy labels, in general discriminative data representation tasks. The fundamental pursuit of our model is to simultaneously learn a transformation function and a discriminative classifier that maximize the mutual information of data and their labels in the latent space. In this general paradigm, we respectively discuss three types of the RIT implementations with linear subspace embedding, deep transformation and structured sparse learning. In practice, the RIT and deep RIT are exploited to solve the image categorization task whose performances will be verified on various benchmark datasets. The structured sparse RIT is further applied to a medical image analysis task for brain MRI segmentation that allows group-level feature selections on the brain tissues.

Nonlinear Deep Kernel Learning for Image Annotation.

PubMed

Jiu, Mingyuan; Sahbi, Hichem

2017-02-08

Multiple kernel learning (MKL) is a widely used technique for kernel design. Its principle consists in learning, for a given support vector classifier, the most suitable convex (or sparse) linear combination of standard elementary kernels. However, these combinations are shallow and often powerless to capture the actual similarity between highly semantic data, especially for challenging classification tasks such as image annotation. In this paper, we redefine multiple kernels using deep multi-layer networks. In this new contribution, a deep multiple kernel is recursively defined as a multi-layered combination of nonlinear activation functions, each one involves a combination of several elementary or intermediate kernels, and results into a positive semi-definite deep kernel. We propose four different frameworks in order to learn the weights of these networks: supervised, unsupervised, kernel-based semisupervised and Laplacian-based semi-supervised. When plugged into support vector machines (SVMs), the resulting deep kernel networks show clear gain, compared to several shallow kernels for the task of image annotation. Extensive experiments and analysis on the challenging ImageCLEF photo annotation benchmark, the COREL5k database and the Banana dataset validate the effectiveness of the proposed method.

A New Approach to Develop Computer-aided Diagnosis Scheme of Breast Mass Classification Using Deep Learning Technology

PubMed Central

Qiu, Yuchen; Yan, Shiju; Gundreddy, Rohith Reddy; Wang, Yunzhi; Cheng, Samuel; Liu, Hong; Zheng, Bin

2017-01-01

PURPOSE To develop and test a deep learning based computer-aided diagnosis (CAD) scheme of mammograms for classifying between malignant and benign masses. METHODS An image dataset involving 560 regions of interest (ROIs) extracted from digital mammograms was used. After down-sampling each ROI from 512×512 to 64×64 pixel size, we applied an 8 layer deep learning network that involves 3 pairs of convolution-max-pooling layers for automatic feature extraction and a multiple layer perceptron (MLP) classifier for feature categorization to process ROIs. The 3 pairs of convolution layers contain 20, 10, and 5 feature maps, respectively. Each convolution layer is connected with a max-pooling layer to improve the feature robustness. The output of the sixth layer is fully connected with a MLP classifier, which is composed of one hidden layer and one logistic regression layer. The network then generates a classification score to predict the likelihood of ROI depicting a malignant mass. A four-fold cross validation method was applied to train and test this deep learning network. RESULTS The results revealed that this CAD scheme yields an area under the receiver operation characteristic curve (AUC) of 0.696±0.044, 0.802±0.037, 0.836±0.036, and 0.822±0.035 for fold 1 to 4 testing datasets, respectively. The overall AUC of the entire dataset is 0.790±0.019. CONCLUSIONS This study demonstrates the feasibility of applying a deep learning based CAD scheme to classify between malignant and benign breast masses without a lesion segmentation, image feature computation and selection process. PMID:28436410

A new approach to develop computer-aided diagnosis scheme of breast mass classification using deep learning technology.

PubMed

Qiu, Yuchen; Yan, Shiju; Gundreddy, Rohith Reddy; Wang, Yunzhi; Cheng, Samuel; Liu, Hong; Zheng, Bin

2017-01-01

To develop and test a deep learning based computer-aided diagnosis (CAD) scheme of mammograms for classifying between malignant and benign masses. An image dataset involving 560 regions of interest (ROIs) extracted from digital mammograms was used. After down-sampling each ROI from 512×512 to 64×64 pixel size, we applied an 8 layer deep learning network that involves 3 pairs of convolution-max-pooling layers for automatic feature extraction and a multiple layer perceptron (MLP) classifier for feature categorization to process ROIs. The 3 pairs of convolution layers contain 20, 10, and 5 feature maps, respectively. Each convolution layer is connected with a max-pooling layer to improve the feature robustness. The output of the sixth layer is fully connected with a MLP classifier, which is composed of one hidden layer and one logistic regression layer. The network then generates a classification score to predict the likelihood of ROI depicting a malignant mass. A four-fold cross validation method was applied to train and test this deep learning network. The results revealed that this CAD scheme yields an area under the receiver operation characteristic curve (AUC) of 0.696±0.044, 0.802±0.037, 0.836±0.036, and 0.822±0.035 for fold 1 to 4 testing datasets, respectively. The overall AUC of the entire dataset is 0.790±0.019. This study demonstrates the feasibility of applying a deep learning based CAD scheme to classify between malignant and benign breast masses without a lesion segmentation, image feature computation and selection process.

Exemplar-Based Image and Video Stylization Using Fully Convolutional Semantic Features.

PubMed

Zhu, Feida; Yan, Zhicheng; Bu, Jiajun; Yu, Yizhou

2017-05-10

Color and tone stylization in images and videos strives to enhance unique themes with artistic color and tone adjustments. It has a broad range of applications from professional image postprocessing to photo sharing over social networks. Mainstream photo enhancement softwares, such as Adobe Lightroom and Instagram, provide users with predefined styles, which are often hand-crafted through a trial-and-error process. Such photo adjustment tools lack a semantic understanding of image contents and the resulting global color transform limits the range of artistic styles it can represent. On the other hand, stylistic enhancement needs to apply distinct adjustments to various semantic regions. Such an ability enables a broader range of visual styles. In this paper, we first propose a novel deep learning architecture for exemplar-based image stylization, which learns local enhancement styles from image pairs. Our deep learning architecture consists of fully convolutional networks (FCNs) for automatic semantics-aware feature extraction and fully connected neural layers for adjustment prediction. Image stylization can be efficiently accomplished with a single forward pass through our deep network. To extend our deep network from image stylization to video stylization, we exploit temporal superpixels (TSPs) to facilitate the transfer of artistic styles from image exemplars to videos. Experiments on a number of datasets for image stylization as well as a diverse set of video clips demonstrate the effectiveness of our deep learning architecture.

Deep learning with non-medical training used for chest pathology identification

NASA Astrophysics Data System (ADS)

Bar, Yaniv; Diamant, Idit; Wolf, Lior; Greenspan, Hayit

2015-03-01

In this work, we examine the strength of deep learning approaches for pathology detection in chest radiograph data. Convolutional neural networks (CNN) deep architecture classification approaches have gained popularity due to their ability to learn mid and high level image representations. We explore the ability of a CNN to identify different types of pathologies in chest x-ray images. Moreover, since very large training sets are generally not available in the medical domain, we explore the feasibility of using a deep learning approach based on non-medical learning. We tested our algorithm on a dataset of 93 images. We use a CNN that was trained with ImageNet, a well-known large scale nonmedical image database. The best performance was achieved using a combination of features extracted from the CNN and a set of low-level features. We obtained an area under curve (AUC) of 0.93 for Right Pleural Effusion detection, 0.89 for Enlarged heart detection and 0.79 for classification between healthy and abnormal chest x-ray, where all pathologies are combined into one large class. This is a first-of-its-kind experiment that shows that deep learning with large scale non-medical image databases may be sufficient for general medical image recognition tasks.

A Benchmark Dataset and Saliency-guided Stacked Autoencoders for Video-based Salient Object Detection.

PubMed

Li, Jia; Xia, Changqun; Chen, Xiaowu

2017-10-12

Image-based salient object detection (SOD) has been extensively studied in past decades. However, video-based SOD is much less explored due to the lack of large-scale video datasets within which salient objects are unambiguously defined and annotated. Toward this end, this paper proposes a video-based SOD dataset that consists of 200 videos. In constructing the dataset, we manually annotate all objects and regions over 7,650 uniformly sampled keyframes and collect the eye-tracking data of 23 subjects who free-view all videos. From the user data, we find that salient objects in a video can be defined as objects that consistently pop-out throughout the video, and objects with such attributes can be unambiguously annotated by combining manually annotated object/region masks with eye-tracking data of multiple subjects. To the best of our knowledge, it is currently the largest dataset for videobased salient object detection. Based on this dataset, this paper proposes an unsupervised baseline approach for video-based SOD by using saliencyguided stacked autoencoders. In the proposed approach, multiple spatiotemporal saliency cues are first extracted at the pixel, superpixel and object levels. With these saliency cues, stacked autoencoders are constructed in an unsupervised manner that automatically infers a saliency score for each pixel by progressively encoding the high-dimensional saliency cues gathered from the pixel and its spatiotemporal neighbors. In experiments, the proposed unsupervised approach is compared with 31 state-of-the-art models on the proposed dataset and outperforms 30 of them, including 19 imagebased classic (unsupervised or non-deep learning) models, six image-based deep learning models, and five video-based unsupervised models. Moreover, benchmarking results show that the proposed dataset is very challenging and has the potential to boost the development of video-based SOD.

Performance evaluation of 2D and 3D deep learning approaches for automatic segmentation of multiple organs on CT images

NASA Astrophysics Data System (ADS)

Zhou, Xiangrong; Yamada, Kazuma; Kojima, Takuya; Takayama, Ryosuke; Wang, Song; Zhou, Xinxin; Hara, Takeshi; Fujita, Hiroshi

2018-02-01

The purpose of this study is to evaluate and compare the performance of modern deep learning techniques for automatically recognizing and segmenting multiple organ regions on 3D CT images. CT image segmentation is one of the important task in medical image analysis and is still very challenging. Deep learning approaches have demonstrated the capability of scene recognition and semantic segmentation on nature images and have been used to address segmentation problems of medical images. Although several works showed promising results of CT image segmentation by using deep learning approaches, there is no comprehensive evaluation of segmentation performance of the deep learning on segmenting multiple organs on different portions of CT scans. In this paper, we evaluated and compared the segmentation performance of two different deep learning approaches that used 2D- and 3D deep convolutional neural networks (CNN) without- and with a pre-processing step. A conventional approach that presents the state-of-the-art performance of CT image segmentation without deep learning was also used for comparison. A dataset that includes 240 CT images scanned on different portions of human bodies was used for performance evaluation. The maximum number of 17 types of organ regions in each CT scan were segmented automatically and compared to the human annotations by using ratio of intersection over union (IU) as the criterion. The experimental results demonstrated the IUs of the segmentation results had a mean value of 79% and 67% by averaging 17 types of organs that segmented by a 3D- and 2D deep CNN, respectively. All the results of the deep learning approaches showed a better accuracy and robustness than the conventional segmentation method that used probabilistic atlas and graph-cut methods. The effectiveness and the usefulness of deep learning approaches were demonstrated for solving multiple organs segmentation problem on 3D CT images.

A light and faster regional convolutional neural network for object detection in optical remote sensing images

NASA Astrophysics Data System (ADS)

Ding, Peng; Zhang, Ye; Deng, Wei-Jian; Jia, Ping; Kuijper, Arjan

2018-07-01

Detection of objects from satellite optical remote sensing images is very important for many commercial and governmental applications. With the development of deep convolutional neural networks (deep CNNs), the field of object detection has seen tremendous advances. Currently, objects in satellite remote sensing images can be detected using deep CNNs. In general, optical remote sensing images contain many dense and small objects, and the use of the original Faster Regional CNN framework does not yield a suitably high precision. Therefore, after careful analysis we adopt dense convoluted networks, a multi-scale representation and various combinations of improvement schemes to enhance the structure of the base VGG16-Net for improving the precision. We propose an approach to reduce the test-time (detection time) and memory requirements. To validate the effectiveness of our approach, we perform experiments using satellite remote sensing image datasets of aircraft and automobiles. The results show that the improved network structure can detect objects in satellite optical remote sensing images more accurately and efficiently.

A model of traffic signs recognition with convolutional neural network

NASA Astrophysics Data System (ADS)

Hu, Haihe; Li, Yujian; Zhang, Ting; Huo, Yi; Kuang, Wenqing

2016-10-01

In real traffic scenes, the quality of captured images are generally low due to some factors such as lighting conditions, and occlusion on. All of these factors are challengeable for automated recognition algorithms of traffic signs. Deep learning has provided a new way to solve this kind of problems recently. The deep network can automatically learn features from a large number of data samples and obtain an excellent recognition performance. We therefore approach this task of recognition of traffic signs as a general vision problem, with few assumptions related to road signs. We propose a model of Convolutional Neural Network (CNN) and apply the model to the task of traffic signs recognition. The proposed model adopts deep CNN as the supervised learning model, directly takes the collected traffic signs image as the input, alternates the convolutional layer and subsampling layer, and automatically extracts the features for the recognition of the traffic signs images. The proposed model includes an input layer, three convolutional layers, three subsampling layers, a fully-connected layer, and an output layer. To validate the proposed model, the experiments are implemented using the public dataset of China competition of fuzzy image processing. Experimental results show that the proposed model produces a recognition accuracy of 99.01 % on the training dataset, and yield a record of 92% on the preliminary contest within the fourth best.

Landcover Classification Using Deep Fully Convolutional Neural Networks

NASA Astrophysics Data System (ADS)

Wang, J.; Li, X.; Zhou, S.; Tang, J.

2017-12-01

Land cover classification has always been an essential application in remote sensing. Certain image features are needed for land cover classification whether it is based on pixel or object-based methods. Different from other machine learning methods, deep learning model not only extracts useful information from multiple bands/attributes, but also learns spatial characteristics. In recent years, deep learning methods have been developed rapidly and widely applied in image recognition, semantic understanding, and other application domains. However, there are limited studies applying deep learning methods in land cover classification. In this research, we used fully convolutional networks (FCN) as the deep learning model to classify land covers. The National Land Cover Database (NLCD) within the state of Kansas was used as training dataset and Landsat images were classified using the trained FCN model. We also applied an image segmentation method to improve the original results from the FCN model. In addition, the pros and cons between deep learning and several machine learning methods were compared and explored. Our research indicates: (1) FCN is an effective classification model with an overall accuracy of 75%; (2) image segmentation improves the classification results with better match of spatial patterns; (3) FCN has an excellent ability of learning which can attains higher accuracy and better spatial patterns compared with several machine learning methods.

Deep 3D convolution neural network for CT brain hemorrhage classification

NASA Astrophysics Data System (ADS)

Jnawali, Kamal; Arbabshirani, Mohammad R.; Rao, Navalgund; Patel, Alpen A.

2018-02-01

Intracranial hemorrhage is a critical conditional with the high mortality rate that is typically diagnosed based on head computer tomography (CT) images. Deep learning algorithms, in particular, convolution neural networks (CNN), are becoming the methodology of choice in medical image analysis for a variety of applications such as computer-aided diagnosis, and segmentation. In this study, we propose a fully automated deep learning framework which learns to detect brain hemorrhage based on cross sectional CT images. The dataset for this work consists of 40,367 3D head CT studies (over 1.5 million 2D images) acquired retrospectively over a decade from multiple radiology facilities at Geisinger Health System. The proposed algorithm first extracts features using 3D CNN and then detects brain hemorrhage using the logistic function as the last layer of the network. Finally, we created an ensemble of three different 3D CNN architectures to improve the classification accuracy. The area under the curve (AUC) of the receiver operator characteristic (ROC) curve of the ensemble of three architectures was 0.87. Their results are very promising considering the fact that the head CT studies were not controlled for slice thickness, scanner type, study protocol or any other settings. Moreover, the proposed algorithm reliably detected various types of hemorrhage within the skull. This work is one of the first applications of 3D CNN trained on a large dataset of cross sectional medical images for detection of a critical radiological condition

Localization and diagnosis framework for pediatric cataracts based on slit-lamp images using deep features of a convolutional neural network

PubMed Central

Zhang, Kai; Long, Erping; Cui, Jiangtao; Zhu, Mingmin; An, Yingying; Zhang, Jia; Liu, Zhenzhen; Lin, Zhuoling; Li, Xiaoyan; Chen, Jingjing; Cao, Qianzhong; Li, Jing; Wu, Xiaohang; Wang, Dongni

2017-01-01

Slit-lamp images play an essential role for diagnosis of pediatric cataracts. We present a computer vision-based framework for the automatic localization and diagnosis of slit-lamp images by identifying the lens region of interest (ROI) and employing a deep learning convolutional neural network (CNN). First, three grading degrees for slit-lamp images are proposed in conjunction with three leading ophthalmologists. The lens ROI is located in an automated manner in the original image using two successive applications of Candy detection and the Hough transform, which are cropped, resized to a fixed size and used to form pediatric cataract datasets. These datasets are fed into the CNN to extract high-level features and implement automatic classification and grading. To demonstrate the performance and effectiveness of the deep features extracted in the CNN, we investigate the features combined with support vector machine (SVM) and softmax classifier and compare these with the traditional representative methods. The qualitative and quantitative experimental results demonstrate that our proposed method offers exceptional mean accuracy, sensitivity and specificity: classification (97.07%, 97.28%, and 96.83%) and a three-degree grading area (89.02%, 86.63%, and 90.75%), density (92.68%, 91.05%, and 93.94%) and location (89.28%, 82.70%, and 93.08%). Finally, we developed and deployed a potential automatic diagnostic software for ophthalmologists and patients in clinical applications to implement the validated model. PMID:28306716

Metal surface corrosion grade estimation from single image

NASA Astrophysics Data System (ADS)

Chen, Yijun; Qi, Lin; Sun, Huyuan; Fan, Hao; Dong, Junyu

2018-04-01

Metal corrosion can cause many problems, how to quickly and effectively assess the grade of metal corrosion and timely remediation is a very important issue. Typically, this is done by trained surveyors at great cost. Assisting them in the inspection process by computer vision and artificial intelligence would decrease the inspection cost. In this paper, we propose a dataset of metal surface correction used for computer vision detection and present a comparison between standard computer vision techniques by using OpenCV and deep learning method for automatic metal surface corrosion grade estimation from single image on this dataset. The test has been performed by classifying images and calculating the accuracy for the two different approaches.

Computer aided lung cancer diagnosis with deep learning algorithms

NASA Astrophysics Data System (ADS)

Sun, Wenqing; Zheng, Bin; Qian, Wei

2016-03-01

Deep learning is considered as a popular and powerful method in pattern recognition and classification. However, there are not many deep structured applications used in medical imaging diagnosis area, because large dataset is not always available for medical images. In this study we tested the feasibility of using deep learning algorithms for lung cancer diagnosis with the cases from Lung Image Database Consortium (LIDC) database. The nodules on each computed tomography (CT) slice were segmented according to marks provided by the radiologists. After down sampling and rotating we acquired 174412 samples with 52 by 52 pixel each and the corresponding truth files. Three deep learning algorithms were designed and implemented, including Convolutional Neural Network (CNN), Deep Belief Networks (DBNs), Stacked Denoising Autoencoder (SDAE). To compare the performance of deep learning algorithms with traditional computer aided diagnosis (CADx) system, we designed a scheme with 28 image features and support vector machine. The accuracies of CNN, DBNs, and SDAE are 0.7976, 0.8119, and 0.7929, respectively; the accuracy of our designed traditional CADx is 0.7940, which is slightly lower than CNN and DBNs. We also noticed that the mislabeled nodules using DBNs are 4% larger than using traditional CADx, this might be resulting from down sampling process lost some size information of the nodules.

Object recognition using deep convolutional neural networks with complete transfer and partial frozen layers

NASA Astrophysics Data System (ADS)

Kruithof, Maarten C.; Bouma, Henri; Fischer, Noëlle M.; Schutte, Klamer

2016-10-01

Object recognition is important to understand the content of video and allow flexible querying in a large number of cameras, especially for security applications. Recent benchmarks show that deep convolutional neural networks are excellent approaches for object recognition. This paper describes an approach of domain transfer, where features learned from a large annotated dataset are transferred to a target domain where less annotated examples are available as is typical for the security and defense domain. Many of these networks trained on natural images appear to learn features similar to Gabor filters and color blobs in the first layer. These first-layer features appear to be generic for many datasets and tasks while the last layer is specific. In this paper, we study the effect of copying all layers and fine-tuning a variable number. We performed an experiment with a Caffe-based network on 1000 ImageNet classes that are randomly divided in two equal subgroups for the transfer from one to the other. We copy all layers and vary the number of layers that is fine-tuned and the size of the target dataset. We performed additional experiments with the Keras platform on CIFAR-10 dataset to validate general applicability. We show with both platforms and both datasets that the accuracy on the target dataset improves when more target data is used. When the target dataset is large, it is beneficial to freeze only a few layers. For a large target dataset, the network without transfer learning performs better than the transfer network, especially if many layers are frozen. When the target dataset is small, it is beneficial to transfer (and freeze) many layers. For a small target dataset, the transfer network boosts generalization and it performs much better than the network without transfer learning. Learning time can be reduced by freezing many layers in a network.

A deep convolutional neural network approach to single-particle recognition in cryo-electron microscopy.

PubMed

Zhu, Yanan; Ouyang, Qi; Mao, Youdong

2017-07-21

Single-particle cryo-electron microscopy (cryo-EM) has become a mainstream tool for the structural determination of biological macromolecular complexes. However, high-resolution cryo-EM reconstruction often requires hundreds of thousands of single-particle images. Particle extraction from experimental micrographs thus can be laborious and presents a major practical bottleneck in cryo-EM structural determination. Existing computational methods for particle picking often use low-resolution templates for particle matching, making them susceptible to reference-dependent bias. It is critical to develop a highly efficient template-free method for the automatic recognition of particle images from cryo-EM micrographs. We developed a deep learning-based algorithmic framework, DeepEM, for single-particle recognition from noisy cryo-EM micrographs, enabling automated particle picking, selection and verification in an integrated fashion. The kernel of DeepEM is built upon a convolutional neural network (CNN) composed of eight layers, which can be recursively trained to be highly "knowledgeable". Our approach exhibits an improved performance and accuracy when tested on the standard KLH dataset. Application of DeepEM to several challenging experimental cryo-EM datasets demonstrated its ability to avoid the selection of un-wanted particles and non-particles even when true particles contain fewer features. The DeepEM methodology, derived from a deep CNN, allows automated particle extraction from raw cryo-EM micrographs in the absence of a template. It demonstrates an improved performance, objectivity and accuracy. Application of this novel method is expected to free the labor involved in single-particle verification, significantly improving the efficiency of cryo-EM data processing.

Deep Learning of Atomically Resolved Scanning Transmission Electron Microscopy Images: Chemical Identification and Tracking Local Transformations

DOE PAGES

Ziatdinov, Maxim; Dyck, Ondrej; Maksov, Artem; ...

2017-12-07

Recent advances in scanning transmission electron and scanning probe microscopies have opened unprecedented opportunities in probing the materials structural parameters and various functional properties in real space with an angstrom-level precision. This progress has been accompanied by exponential increase in the size and quality of datasets produced by microscopic and spectroscopic experimental techniques. These developments necessitate adequate methods for extracting relevant physical and chemical information from the large datasets, for which a priori information on the structures of various atomic configurations and lattice defects is limited or absent. Here we demonstrate an application of deep neural networks to extracting informationmore » from atomically resolved images including location of the atomic species and type of defects. We develop a “weakly-supervised” approach that uses information on the coordinates of all atomic species in the image, extracted via a deep neural network, to identify a rich variety of defects that are not part of an initial training set. We further apply our approach to interpret complex atomic and defect transformation, including switching between different coordination of silicon dopants in graphene as a function of time, formation of peculiar silicon dimer with mixed 3-fold and 4-fold coordination, and the motion of molecular “rotor”. In conclusion, this deep learning based approach resembles logic of a human operator, but can be scaled leading to significant shift in the way of extracting and analyzing information from raw experimental data.« less

Deep Learning of Atomically Resolved Scanning Transmission Electron Microscopy Images: Chemical Identification and Tracking Local Transformations

DOE Office of Scientific and Technical Information (OSTI.GOV)

Ziatdinov, Maxim; Dyck, Ondrej; Maksov, Artem

Recent advances in scanning transmission electron and scanning probe microscopies have opened unprecedented opportunities in probing the materials structural parameters and various functional properties in real space with an angstrom-level precision. This progress has been accompanied by exponential increase in the size and quality of datasets produced by microscopic and spectroscopic experimental techniques. These developments necessitate adequate methods for extracting relevant physical and chemical information from the large datasets, for which a priori information on the structures of various atomic configurations and lattice defects is limited or absent. Here we demonstrate an application of deep neural networks to extracting informationmore » from atomically resolved images including location of the atomic species and type of defects. We develop a “weakly-supervised” approach that uses information on the coordinates of all atomic species in the image, extracted via a deep neural network, to identify a rich variety of defects that are not part of an initial training set. We further apply our approach to interpret complex atomic and defect transformation, including switching between different coordination of silicon dopants in graphene as a function of time, formation of peculiar silicon dimer with mixed 3-fold and 4-fold coordination, and the motion of molecular “rotor”. In conclusion, this deep learning based approach resembles logic of a human operator, but can be scaled leading to significant shift in the way of extracting and analyzing information from raw experimental data.« less

Automatic tissue image segmentation based on image processing and deep learning

NASA Astrophysics Data System (ADS)

Kong, Zhenglun; Luo, Junyi; Xu, Shengpu; Li, Ting

2018-02-01

Image segmentation plays an important role in multimodality imaging, especially in fusion structural images offered by CT, MRI with functional images collected by optical technologies or other novel imaging technologies. Plus, image segmentation also provides detailed structure description for quantitative visualization of treating light distribution in the human body when incorporated with 3D light transport simulation method. Here we used image enhancement, operators, and morphometry methods to extract the accurate contours of different tissues such as skull, cerebrospinal fluid (CSF), grey matter (GM) and white matter (WM) on 5 fMRI head image datasets. Then we utilized convolutional neural network to realize automatic segmentation of images in a deep learning way. We also introduced parallel computing. Such approaches greatly reduced the processing time compared to manual and semi-automatic segmentation and is of great importance in improving speed and accuracy as more and more samples being learned. Our results can be used as a criteria when diagnosing diseases such as cerebral atrophy, which is caused by pathological changes in gray matter or white matter. We demonstrated the great potential of such image processing and deep leaning combined automatic tissue image segmentation in personalized medicine, especially in monitoring, and treatments.

Image inpainting and super-resolution using non-local recursive deep convolutional network with skip connections

NASA Astrophysics Data System (ADS)

Liu, Miaofeng

2017-07-01

In recent years, deep convolutional neural networks come into use in image inpainting and super-resolution in many fields. Distinct to most of the former methods requiring to know beforehand the local information for corrupted pixels, we propose a 20-depth fully convolutional network to learn an end-to-end mapping a dataset of damaged/ground truth subimage pairs realizing non-local blind inpainting and super-resolution. As there often exist image with huge corruptions or inpainting on a low-resolution image that the existing approaches unable to perform well, we also share parameters in local area of layers to achieve spatial recursion and enlarge the receptive field. To avoid the difficulty of training this deep neural network, skip-connections between symmetric convolutional layers are designed. Experimental results shows that the proposed method outperforms state-of-the-art methods for diverse corrupting and low-resolution conditions, it works excellently when realizing super-resolution and image inpainting simultaneously

Urban Area Detection in Very High Resolution Remote Sensing Images Using Deep Convolutional Neural Networks.

PubMed

Tian, Tian; Li, Chang; Xu, Jinkang; Ma, Jiayi

2018-03-18

Detecting urban areas from very high resolution (VHR) remote sensing images plays an important role in the field of Earth observation. The recently-developed deep convolutional neural networks (DCNNs), which can extract rich features from training data automatically, have achieved outstanding performance on many image classification databases. Motivated by this fact, we propose a new urban area detection method based on DCNNs in this paper. The proposed method mainly includes three steps: (i) a visual dictionary is obtained based on the deep features extracted by pre-trained DCNNs; (ii) urban words are learned from labeled images; (iii) the urban regions are detected in a new image based on the nearest dictionary word criterion. The qualitative and quantitative experiments on different datasets demonstrate that the proposed method can obtain a remarkable overall accuracy (OA) and kappa coefficient. Moreover, it can also strike a good balance between the true positive rate (TPR) and false positive rate (FPR).

Deep learning for brain tumor classification

NASA Astrophysics Data System (ADS)

Paul, Justin S.; Plassard, Andrew J.; Landman, Bennett A.; Fabbri, Daniel

2017-03-01

Recent research has shown that deep learning methods have performed well on supervised machine learning, image classification tasks. The purpose of this study is to apply deep learning methods to classify brain images with different tumor types: meningioma, glioma, and pituitary. A dataset was publicly released containing 3,064 T1-weighted contrast enhanced MRI (CE-MRI) brain images from 233 patients with either meningioma, glioma, or pituitary tumors split across axial, coronal, or sagittal planes. This research focuses on the 989 axial images from 191 patients in order to avoid confusing the neural networks with three different planes containing the same diagnosis. Two types of neural networks were used in classification: fully connected and convolutional neural networks. Within these two categories, further tests were computed via the augmentation of the original 512×512 axial images. Training neural networks over the axial data has proven to be accurate in its classifications with an average five-fold cross validation of 91.43% on the best trained neural network. This result demonstrates that a more general method (i.e. deep learning) can outperform specialized methods that require image dilation and ring-forming subregions on tumors.

Sleep spindle detection using deep learning: A validation study based on crowdsourcing.

PubMed

Dakun Tan; Rui Zhao; Jinbo Sun; Wei Qin

2015-08-01

Sleep spindles are significant transient oscillations observed on the electroencephalogram (EEG) in stage 2 of non-rapid eye movement sleep. Deep belief network (DBN) gaining great successes in images and speech is still a novel method to develop sleep spindle detection system. In this paper, crowdsourcing replacing gold standard was applied to generate three different labeled samples and constructed three classes of datasets with a combination of these samples. An F1-score measure was estimated to compare the performance of DBN to other three classifiers on classifying these samples, with the DBN obtaining an result of 92.78%. Then a comparison of two feature extraction methods based on power spectrum density was made on same dataset using DBN. In addition, the DBN trained in dataset was applied to detect sleep spindle from raw EEG recordings and performed a comparable capacity to expert group consensus.

Deep Convolutional Neural Networks for breast cancer screening.

PubMed

Chougrad, Hiba; Zouaki, Hamid; Alheyane, Omar

2018-04-01

Radiologists often have a hard time classifying mammography mass lesions which leads to unnecessary breast biopsies to remove suspicions and this ends up adding exorbitant expenses to an already burdened patient and health care system. In this paper we developed a Computer-aided Diagnosis (CAD) system based on deep Convolutional Neural Networks (CNN) that aims to help the radiologist classify mammography mass lesions. Deep learning usually requires large datasets to train networks of a certain depth from scratch. Transfer learning is an effective method to deal with relatively small datasets as in the case of medical images, although it can be tricky as we can easily start overfitting. In this work, we explore the importance of transfer learning and we experimentally determine the best fine-tuning strategy to adopt when training a CNN model. We were able to successfully fine-tune some of the recent, most powerful CNNs and achieved better results compared to other state-of-the-art methods which classified the same public datasets. For instance we achieved 97.35% accuracy and 0.98 AUC on the DDSM database, 95.50% accuracy and 0.97 AUC on the INbreast database and 96.67% accuracy and 0.96 AUC on the BCDR database. Furthermore, after pre-processing and normalizing all the extracted Regions of Interest (ROIs) from the full mammograms, we merged all the datasets to build one large set of images and used it to fine-tune our CNNs. The CNN model which achieved the best results, a 98.94% accuracy, was used as a baseline to build the Breast Cancer Screening Framework. To evaluate the proposed CAD system and its efficiency to classify new images, we tested it on an independent database (MIAS) and got 98.23% accuracy and 0.99 AUC. The results obtained demonstrate that the proposed framework is performant and can indeed be used to predict if the mass lesions are benign or malignant. Copyright © 2018 Elsevier B.V. All rights reserved.

Discrimination of Breast Cancer with Microcalcifications on Mammography by Deep Learning.

PubMed

Wang, Jinhua; Yang, Xi; Cai, Hongmin; Tan, Wanchang; Jin, Cangzheng; Li, Li

2016-06-07

Microcalcification is an effective indicator of early breast cancer. To improve the diagnostic accuracy of microcalcifications, this study evaluates the performance of deep learning-based models on large datasets for its discrimination. A semi-automated segmentation method was used to characterize all microcalcifications. A discrimination classifier model was constructed to assess the accuracies of microcalcifications and breast masses, either in isolation or combination, for classifying breast lesions. Performances were compared to benchmark models. Our deep learning model achieved a discriminative accuracy of 87.3% if microcalcifications were characterized alone, compared to 85.8% with a support vector machine. The accuracies were 61.3% for both methods with masses alone and improved to 89.7% and 85.8% after the combined analysis with microcalcifications. Image segmentation with our deep learning model yielded 15, 26 and 41 features for the three scenarios, respectively. Overall, deep learning based on large datasets was superior to standard methods for the discrimination of microcalcifications. Accuracy was increased by adopting a combinatorial approach to detect microcalcifications and masses simultaneously. This may have clinical value for early detection and treatment of breast cancer.

Discrimination of Breast Cancer with Microcalcifications on Mammography by Deep Learning

PubMed Central

Wang, Jinhua; Yang, Xi; Cai, Hongmin; Tan, Wanchang; Jin, Cangzheng; Li, Li

2016-01-01

Microcalcification is an effective indicator of early breast cancer. To improve the diagnostic accuracy of microcalcifications, this study evaluates the performance of deep learning-based models on large datasets for its discrimination. A semi-automated segmentation method was used to characterize all microcalcifications. A discrimination classifier model was constructed to assess the accuracies of microcalcifications and breast masses, either in isolation or combination, for classifying breast lesions. Performances were compared to benchmark models. Our deep learning model achieved a discriminative accuracy of 87.3% if microcalcifications were characterized alone, compared to 85.8% with a support vector machine. The accuracies were 61.3% for both methods with masses alone and improved to 89.7% and 85.8% after the combined analysis with microcalcifications. Image segmentation with our deep learning model yielded 15, 26 and 41 features for the three scenarios, respectively. Overall, deep learning based on large datasets was superior to standard methods for the discrimination of microcalcifications. Accuracy was increased by adopting a combinatorial approach to detect microcalcifications and masses simultaneously. This may have clinical value for early detection and treatment of breast cancer. PMID:27273294

Combined in-depth, 3D, en face imaging of the optic disc, optic disc pits and optic disc pit maculopathy using swept-source megahertz OCT at 1050 nm.

PubMed

Maertz, Josef; Kolb, Jan Philip; Klein, Thomas; Mohler, Kathrin J; Eibl, Matthias; Wieser, Wolfgang; Huber, Robert; Priglinger, Siegfried; Wolf, Armin

2018-02-01

To demonstrate papillary imaging of eyes with optic disc pits (ODP) or optic disc pit associated maculopathy (ODP-M) with ultrahigh-speed swept-source optical coherence tomography (SS-OCT) at 1.68 million A-scans/s. To generate 3D-renderings of the papillary area with 3D volume-reconstructions of the ODP and highly resolved en face images from a single densely-sampled megahertz-OCT (MHz-OCT) dataset for investigation of ODP-characteristics. A 1.68 MHz-prototype SS-MHz-OCT system at 1050 nm based on a Fourier-domain mode-locked laser was employed to acquire high-definition, 3D datasets with a dense sampling of 1600 × 1600 A-scans over a 45° field of view. Six eyes with ODPs, and two further eyes with glaucomatous alteration or without ocular pathology are presented. 3D-rendering of the deep papillary structures, virtual 3D-reconstructions of the ODPs and depth resolved isotropic en face images were generated using semiautomatic segmentation. 3D-rendering and en face imaging of the optic disc, ODPs and ODP associated pathologies showed a broad spectrum regarding ODP characteristics. Between individuals the shape of the ODP and the appending pathologies varied considerably. MHz-OCT en face imaging generates distinct top-view images of ODPs and ODP-M. MHz-OCT generates high resolution images of retinal pathologies associated with ODP-M and allows visualizing ODPs with depths of up to 2.7 mm. Different patterns of ODPs can be visualized in patients for the first time using 3D-reconstructions and co-registered high-definition en face images extracted from a single densely sampled 1050 nm megahertz-OCT (MHz-OCT) dataset. As the immediate vicinity to the SAS and the site of intrapapillary proliferation is located at the bottom of the ODP it is crucial to image the complete structure and the whole depth of ODPs. Especially in very deep pits, where non-swept-source OCT fails to reach the bottom, conventional swept-source devices and the MHz-OCT alike are feasible and beneficial methods to examine deep details of optic disc pathologies, while the MHz-OCT bears the advantage of an essentially swifter imaging process.

A Generic Deep-Learning-Based Approach for Automated Surface Inspection.

PubMed

Ren, Ruoxu; Hung, Terence; Tan, Kay Chen

2018-03-01

Automated surface inspection (ASI) is a challenging task in industry, as collecting training dataset is usually costly and related methods are highly dataset-dependent. In this paper, a generic approach that requires small training data for ASI is proposed. First, this approach builds classifier on the features of image patches, where the features are transferred from a pretrained deep learning network. Next, pixel-wise prediction is obtained by convolving the trained classifier over input image. An experiment on three public and one industrial data set is carried out. The experiment involves two tasks: 1) image classification and 2) defect segmentation. The results of proposed algorithm are compared against several best benchmarks in literature. In the classification tasks, the proposed method improves accuracy by 0.66%-25.50%. In the segmentation tasks, the proposed method reduces error escape rates by 6.00%-19.00% in three defect types and improves accuracies by 2.29%-9.86% in all seven defect types. In addition, the proposed method achieves 0.0% error escape rate in the segmentation task of industrial data.

Applications of deep convolutional neural networks to digitized natural history collections.

PubMed

Schuettpelz, Eric; Frandsen, Paul B; Dikow, Rebecca B; Brown, Abel; Orli, Sylvia; Peters, Melinda; Metallo, Adam; Funk, Vicki A; Dorr, Laurence J

2017-01-01

Natural history collections contain data that are critical for many scientific endeavors. Recent efforts in mass digitization are generating large datasets from these collections that can provide unprecedented insight. Here, we present examples of how deep convolutional neural networks can be applied in analyses of imaged herbarium specimens. We first demonstrate that a convolutional neural network can detect mercury-stained specimens across a collection with 90% accuracy. We then show that such a network can correctly distinguish two morphologically similar plant families 96% of the time. Discarding the most challenging specimen images increases accuracy to 94% and 99%, respectively. These results highlight the importance of mass digitization and deep learning approaches and reveal how they can together deliver powerful new investigative tools.

ChemNet: A Transferable and Generalizable Deep Neural Network for Small-Molecule Property Prediction

DOE Office of Scientific and Technical Information (OSTI.GOV)

Goh, Garrett B.; Siegel, Charles M.; Vishnu, Abhinav

With access to large datasets, deep neural networks through representation learning have been able to identify patterns from raw data, achieving human-level accuracy in image and speech recognition tasks. However, in chemistry, availability of large standardized and labelled datasets is scarce, and with a multitude of chemical properties of interest, chemical data is inherently small and fragmented. In this work, we explore transfer learning techniques in conjunction with the existing Chemception CNN model, to create a transferable and generalizable deep neural network for small-molecule property prediction. Our latest model, ChemNet learns in a semi-supervised manner from inexpensive labels computed frommore » the ChEMBL database. When fine-tuned to the Tox21, HIV and FreeSolv dataset, which are 3 separate chemical tasks that ChemNet was not originally trained on, we demonstrate that ChemNet exceeds the performance of existing Chemception models, contemporary MLP models that trains on molecular fingerprints, and it matches the performance of the ConvGraph algorithm, the current state-of-the-art. Furthermore, as ChemNet has been pre-trained on a large diverse chemical database, it can be used as a universal “plug-and-play” deep neural network, which accelerates the deployment of deep neural networks for the prediction of novel small-molecule chemical properties.« less

Diagnosis and prediction of periodontally compromised teeth using a deep learning-based convolutional neural network algorithm.

PubMed

Lee, Jae-Hong; Kim, Do-Hyung; Jeong, Seong-Nyum; Choi, Seong-Ho

2018-04-01

The aim of the current study was to develop a computer-assisted detection system based on a deep convolutional neural network (CNN) algorithm and to evaluate the potential usefulness and accuracy of this system for the diagnosis and prediction of periodontally compromised teeth (PCT). Combining pretrained deep CNN architecture and a self-trained network, periapical radiographic images were used to determine the optimal CNN algorithm and weights. The diagnostic and predictive accuracy, sensitivity, specificity, positive predictive value, negative predictive value, receiver operating characteristic (ROC) curve, area under the ROC curve, confusion matrix, and 95% confidence intervals (CIs) were calculated using our deep CNN algorithm, based on a Keras framework in Python. The periapical radiographic dataset was split into training (n=1,044), validation (n=348), and test (n=348) datasets. With the deep learning algorithm, the diagnostic accuracy for PCT was 81.0% for premolars and 76.7% for molars. Using 64 premolars and 64 molars that were clinically diagnosed as severe PCT, the accuracy of predicting extraction was 82.8% (95% CI, 70.1%-91.2%) for premolars and 73.4% (95% CI, 59.9%-84.0%) for molars. We demonstrated that the deep CNN algorithm was useful for assessing the diagnosis and predictability of PCT. Therefore, with further optimization of the PCT dataset and improvements in the algorithm, a computer-aided detection system can be expected to become an effective and efficient method of diagnosing and predicting PCT.

Automatic diagnosis of imbalanced ophthalmic images using a cost-sensitive deep convolutional neural network.

PubMed

Jiang, Jiewei; Liu, Xiyang; Zhang, Kai; Long, Erping; Wang, Liming; Li, Wangting; Liu, Lin; Wang, Shuai; Zhu, Mingmin; Cui, Jiangtao; Liu, Zhenzhen; Lin, Zhuoling; Li, Xiaoyan; Chen, Jingjing; Cao, Qianzhong; Li, Jing; Wu, Xiaohang; Wang, Dongni; Wang, Jinghui; Lin, Haotian

2017-11-21

Ocular images play an essential role in ophthalmological diagnoses. Having an imbalanced dataset is an inevitable issue in automated ocular diseases diagnosis; the scarcity of positive samples always tends to result in the misdiagnosis of severe patients during the classification task. Exploring an effective computer-aided diagnostic method to deal with imbalanced ophthalmological dataset is crucial. In this paper, we develop an effective cost-sensitive deep residual convolutional neural network (CS-ResCNN) classifier to diagnose ophthalmic diseases using retro-illumination images. First, the regions of interest (crystalline lens) are automatically identified via twice-applied Canny detection and Hough transformation. Then, the localized zones are fed into the CS-ResCNN to extract high-level features for subsequent use in automatic diagnosis. Second, the impacts of cost factors on the CS-ResCNN are further analyzed using a grid-search procedure to verify that our proposed system is robust and efficient. Qualitative analyses and quantitative experimental results demonstrate that our proposed method outperforms other conventional approaches and offers exceptional mean accuracy (92.24%), specificity (93.19%), sensitivity (89.66%) and AUC (97.11%) results. Moreover, the sensitivity of the CS-ResCNN is enhanced by over 13.6% compared to the native CNN method. Our study provides a practical strategy for addressing imbalanced ophthalmological datasets and has the potential to be applied to other medical images. The developed and deployed CS-ResCNN could serve as computer-aided diagnosis software for ophthalmologists in clinical application.

A combined deep-learning and deformable-model approach to fully automatic segmentation of the left ventricle in cardiac MRI.

PubMed

Avendi, M R; Kheradvar, Arash; Jafarkhani, Hamid

2016-05-01

Segmentation of the left ventricle (LV) from cardiac magnetic resonance imaging (MRI) datasets is an essential step for calculation of clinical indices such as ventricular volume and ejection fraction. In this work, we employ deep learning algorithms combined with deformable models to develop and evaluate a fully automatic LV segmentation tool from short-axis cardiac MRI datasets. The method employs deep learning algorithms to learn the segmentation task from the ground true data. Convolutional networks are employed to automatically detect the LV chamber in MRI dataset. Stacked autoencoders are used to infer the LV shape. The inferred shape is incorporated into deformable models to improve the accuracy and robustness of the segmentation. We validated our method using 45 cardiac MR datasets from the MICCAI 2009 LV segmentation challenge and showed that it outperforms the state-of-the art methods. Excellent agreement with the ground truth was achieved. Validation metrics, percentage of good contours, Dice metric, average perpendicular distance and conformity, were computed as 96.69%, 0.94, 1.81 mm and 0.86, versus those of 79.2-95.62%, 0.87-0.9, 1.76-2.97 mm and 0.67-0.78, obtained by other methods, respectively. Copyright © 2016 Elsevier B.V. All rights reserved.

An Embodied Multi-Sensor Fusion Approach to Visual Motion Estimation Using Unsupervised Deep Networks.

PubMed

Shamwell, E Jared; Nothwang, William D; Perlis, Donald

2018-05-04

Aimed at improving size, weight, and power (SWaP)-constrained robotic vision-aided state estimation, we describe our unsupervised, deep convolutional-deconvolutional sensor fusion network, Multi-Hypothesis DeepEfference (MHDE). MHDE learns to intelligently combine noisy heterogeneous sensor data to predict several probable hypotheses for the dense, pixel-level correspondence between a source image and an unseen target image. We show how our multi-hypothesis formulation provides increased robustness against dynamic, heteroscedastic sensor and motion noise by computing hypothesis image mappings and predictions at 76⁻357 Hz depending on the number of hypotheses being generated. MHDE fuses noisy, heterogeneous sensory inputs using two parallel, inter-connected architectural pathways and n (1⁻20 in this work) multi-hypothesis generating sub-pathways to produce n global correspondence estimates between a source and a target image. We evaluated MHDE on the KITTI Odometry dataset and benchmarked it against the vision-only DeepMatching and Deformable Spatial Pyramids algorithms and were able to demonstrate a significant runtime decrease and a performance increase compared to the next-best performing method.

Deep feature classification of angiomyolipoma without visible fat and renal cell carcinoma in abdominal contrast-enhanced CT images with texture image patches and hand-crafted feature concatenation.

PubMed

Lee, Hansang; Hong, Helen; Kim, Junmo; Jung, Dae Chul

2018-04-01

To develop an automatic deep feature classification (DFC) method for distinguishing benign angiomyolipoma without visible fat (AMLwvf) from malignant clear cell renal cell carcinoma (ccRCC) from abdominal contrast-enhanced computer tomography (CE CT) images. A dataset including 80 abdominal CT images of 39 AMLwvf and 41 ccRCC patients was used. We proposed a DFC method for differentiating the small renal masses (SRM) into AMLwvf and ccRCC using the combination of hand-crafted and deep features, and machine learning classifiers. First, 71-dimensional hand-crafted features (HCF) of texture and shape were extracted from the SRM contours. Second, 1000-4000-dimensional deep features (DF) were extracted from the ImageNet pretrained deep learning model with the SRM image patches. In DF extraction, we proposed the texture image patches (TIP) to emphasize the texture information inside the mass in DFs and reduce the mass size variability. Finally, the two features were concatenated and the random forest (RF) classifier was trained on these concatenated features to classify the types of SRMs. The proposed method was tested on our dataset using leave-one-out cross-validation and evaluated using accuracy, sensitivity, specificity, positive predictive values (PPV), negative predictive values (NPV), and area under receiver operating characteristics curve (AUC). In experiments, the combinations of four deep learning models, AlexNet, VGGNet, GoogleNet, and ResNet, and four input image patches, including original, masked, mass-size, and texture image patches, were compared and analyzed. In qualitative evaluation, we observed the change in feature distributions between the proposed and comparative methods using tSNE method. In quantitative evaluation, we evaluated and compared the classification results, and observed that (a) the proposed HCF + DF outperformed HCF-only and DF-only, (b) AlexNet showed generally the best performances among the CNN models, and (c) the proposed TIPs not only achieved the competitive performances among the input patches, but also steady performance regardless of CNN models. As a result, the proposed method achieved the accuracy of 76.6 ± 1.4% for the proposed HCF + DF with AlexNet and TIPs, which improved the accuracy by 6.6%p and 8.3%p compared to HCF-only and DF-only, respectively. The proposed shape features and TIPs improved the HCFs and DFs, respectively, and the feature concatenation further enhanced the quality of features for differentiating AMLwvf from ccRCC in abdominal CE CT images. © 2018 American Association of Physicists in Medicine.

Classify epithelium-stroma in histopathological images based on deep transferable network.

PubMed

Yu, X; Zheng, H; Liu, C; Huang, Y; Ding, X

2018-04-20

Recently, the deep learning methods have received more attention in histopathological image analysis. However, the traditional deep learning methods assume that training data and test data have the same distributions, which causes certain limitations in real-world histopathological applications. However, it is costly to recollect a large amount of labeled histology data to train a new neural network for each specified image acquisition procedure even for similar tasks. In this paper, an unsupervised domain adaptation is introduced into a typical deep convolutional neural network (CNN) model to mitigate the repeating of the labels. The unsupervised domain adaptation is implemented by adding two regularisation terms, namely the feature-based adaptation and entropy minimisation, to the object function of a widely used CNN model called the AlexNet. Three independent public epithelium-stroma datasets were used to verify the proposed method. The experimental results have demonstrated that in the epithelium-stroma classification, the proposed method can achieve better performance than the commonly used deep learning methods and some existing deep domain adaptation methods. Therefore, the proposed method can be considered as a better option for the real-world applications of histopathological image analysis because there is no requirement for recollection of large-scale labeled data for every specified domain. © 2018 The Authors Journal of Microscopy © 2018 Royal Microscopical Society.

An analysis of image storage systems for scalable training of deep neural networks

DOE Office of Scientific and Technical Information (OSTI.GOV)

Lim, Seung-Hwan; Young, Steven R; Patton, Robert M

This study presents a principled empirical evaluation of image storage systems for training deep neural networks. We employ the Caffe deep learning framework to train neural network models for three different data sets, MNIST, CIFAR-10, and ImageNet. While training the models, we evaluate five different options to retrieve training image data: (1) PNG-formatted image files on local file system; (2) pushing pixel arrays from image files into a single HDF5 file on local file system; (3) in-memory arrays to hold the pixel arrays in Python and C++; (4) loading the training data into LevelDB, a log-structured merge tree based key-valuemore » storage; and (5) loading the training data into LMDB, a B+tree based key-value storage. The experimental results quantitatively highlight the disadvantage of using normal image files on local file systems to train deep neural networks and demonstrate reliable performance with key-value storage based storage systems. When training a model on the ImageNet dataset, the image file option was more than 17 times slower than the key-value storage option. Along with measurements on training time, this study provides in-depth analysis on the cause of performance advantages/disadvantages of each back-end to train deep neural networks. We envision the provided measurements and analysis will shed light on the optimal way to architect systems for training neural networks in a scalable manner.« less

Binary patterns encoded convolutional neural networks for texture recognition and remote sensing scene classification

NASA Astrophysics Data System (ADS)

Anwer, Rao Muhammad; Khan, Fahad Shahbaz; van de Weijer, Joost; Molinier, Matthieu; Laaksonen, Jorma

2018-04-01

Designing discriminative powerful texture features robust to realistic imaging conditions is a challenging computer vision problem with many applications, including material recognition and analysis of satellite or aerial imagery. In the past, most texture description approaches were based on dense orderless statistical distribution of local features. However, most recent approaches to texture recognition and remote sensing scene classification are based on Convolutional Neural Networks (CNNs). The de facto practice when learning these CNN models is to use RGB patches as input with training performed on large amounts of labeled data (ImageNet). In this paper, we show that Local Binary Patterns (LBP) encoded CNN models, codenamed TEX-Nets, trained using mapped coded images with explicit LBP based texture information provide complementary information to the standard RGB deep models. Additionally, two deep architectures, namely early and late fusion, are investigated to combine the texture and color information. To the best of our knowledge, we are the first to investigate Binary Patterns encoded CNNs and different deep network fusion architectures for texture recognition and remote sensing scene classification. We perform comprehensive experiments on four texture recognition datasets and four remote sensing scene classification benchmarks: UC-Merced with 21 scene categories, WHU-RS19 with 19 scene classes, RSSCN7 with 7 categories and the recently introduced large scale aerial image dataset (AID) with 30 aerial scene types. We demonstrate that TEX-Nets provide complementary information to standard RGB deep model of the same network architecture. Our late fusion TEX-Net architecture always improves the overall performance compared to the standard RGB network on both recognition problems. Furthermore, our final combination leads to consistent improvement over the state-of-the-art for remote sensing scene classification.

Deep convolutional neural network training enrichment using multi-view object-based analysis of Unmanned Aerial systems imagery for wetlands classification

NASA Astrophysics Data System (ADS)

Liu, Tao; Abd-Elrahman, Amr

2018-05-01

Deep convolutional neural network (DCNN) requires massive training datasets to trigger its image classification power, while collecting training samples for remote sensing application is usually an expensive process. When DCNN is simply implemented with traditional object-based image analysis (OBIA) for classification of Unmanned Aerial systems (UAS) orthoimage, its power may be undermined if the number training samples is relatively small. This research aims to develop a novel OBIA classification approach that can take advantage of DCNN by enriching the training dataset automatically using multi-view data. Specifically, this study introduces a Multi-View Object-based classification using Deep convolutional neural network (MODe) method to process UAS images for land cover classification. MODe conducts the classification on multi-view UAS images instead of directly on the orthoimage, and gets the final results via a voting procedure. 10-fold cross validation results show the mean overall classification accuracy increasing substantially from 65.32%, when DCNN was applied on the orthoimage to 82.08% achieved when MODe was implemented. This study also compared the performances of the support vector machine (SVM) and random forest (RF) classifiers with DCNN under traditional OBIA and the proposed multi-view OBIA frameworks. The results indicate that the advantage of DCNN over traditional classifiers in terms of accuracy is more obvious when these classifiers were applied with the proposed multi-view OBIA framework than when these classifiers were applied within the traditional OBIA framework.

Deep Learning Nuclei Detection in Digitized Histology Images by Superpixels.

PubMed

Sornapudi, Sudhir; Stanley, Ronald Joe; Stoecker, William V; Almubarak, Haidar; Long, Rodney; Antani, Sameer; Thoma, George; Zuna, Rosemary; Frazier, Shelliane R

2018-01-01

Advances in image analysis and computational techniques have facilitated automatic detection of critical features in histopathology images. Detection of nuclei is critical for squamous epithelium cervical intraepithelial neoplasia (CIN) classification into normal, CIN1, CIN2, and CIN3 grades. In this study, a deep learning (DL)-based nuclei segmentation approach is investigated based on gathering localized information through the generation of superpixels using a simple linear iterative clustering algorithm and training with a convolutional neural network. The proposed approach was evaluated on a dataset of 133 digitized histology images and achieved an overall nuclei detection (object-based) accuracy of 95.97%, with demonstrated improvement over imaging-based and clustering-based benchmark techniques. The proposed DL-based nuclei segmentation Method with superpixel analysis has shown improved segmentation results in comparison to state-of-the-art methods.

Cnn Based Retinal Image Upscaling Using Zero Component Analysis

NASA Astrophysics Data System (ADS)

Nasonov, A.; Chesnakov, K.; Krylov, A.

2017-05-01

The aim of the paper is to obtain high quality of image upscaling for noisy images that are typical in medical image processing. A new training scenario for convolutional neural network based image upscaling method is proposed. Its main idea is a novel dataset preparation method for deep learning. The dataset contains pairs of noisy low-resolution images and corresponding noiseless highresolution images. To achieve better results at edges and textured areas, Zero Component Analysis is applied to these images. The upscaling results are compared with other state-of-the-art methods like DCCI, SI-3 and SRCNN on noisy medical ophthalmological images. Objective evaluation of the results confirms high quality of the proposed method. Visual analysis shows that fine details and structures like blood vessels are preserved, noise level is reduced and no artifacts or non-existing details are added. These properties are essential in retinal diagnosis establishment, so the proposed algorithm is recommended to be used in real medical applications.

A Deep Learning Algorithm for Prediction of Age-Related Eye Disease Study Severity Scale for Age-Related Macular Degeneration from Color Fundus Photography.

PubMed

Grassmann, Felix; Mengelkamp, Judith; Brandl, Caroline; Harsch, Sebastian; Zimmermann, Martina E; Linkohr, Birgit; Peters, Annette; Heid, Iris M; Palm, Christoph; Weber, Bernhard H F

2018-04-10

Age-related macular degeneration (AMD) is a common threat to vision. While classification of disease stages is critical to understanding disease risk and progression, several systems based on color fundus photographs are known. Most of these require in-depth and time-consuming analysis of fundus images. Herein, we present an automated computer-based classification algorithm. Algorithm development for AMD classification based on a large collection of color fundus images. Validation is performed on a cross-sectional, population-based study. We included 120 656 manually graded color fundus images from 3654 Age-Related Eye Disease Study (AREDS) participants. AREDS participants were >55 years of age, and non-AMD sight-threatening diseases were excluded at recruitment. In addition, performance of our algorithm was evaluated in 5555 fundus images from the population-based Kooperative Gesundheitsforschung in der Region Augsburg (KORA; Cooperative Health Research in the Region of Augsburg) study. We defined 13 classes (9 AREDS steps, 3 late AMD stages, and 1 for ungradable images) and trained several convolution deep learning architectures. An ensemble of network architectures improved prediction accuracy. An independent dataset was used to evaluate the performance of our algorithm in a population-based study. κ Statistics and accuracy to evaluate the concordance between predicted and expert human grader classification. A network ensemble of 6 different neural net architectures predicted the 13 classes in the AREDS test set with a quadratic weighted κ of 92% (95% confidence interval, 89%-92%) and an overall accuracy of 63.3%. In the independent KORA dataset, images wrongly classified as AMD were mainly the result of a macular reflex observed in young individuals. By restricting the KORA analysis to individuals >55 years of age and prior exclusion of other retinopathies, the weighted and unweighted κ increased to 50% and 63%, respectively. Importantly, the algorithm detected 84.2% of all fundus images with definite signs of early or late AMD. Overall, 94.3% of healthy fundus images were classified correctly. Our deep learning algoritm revealed a weighted κ outperforming human graders in the AREDS study and is suitable to classify AMD fundus images in other datasets using individuals >55 years of age. Copyright © 2018 American Academy of Ophthalmology. Published by Elsevier Inc. All rights reserved.

Deep-learning-based classification of FDG-PET data for Alzheimer's disease categories

NASA Astrophysics Data System (ADS)

Singh, Shibani; Srivastava, Anant; Mi, Liang; Caselli, Richard J.; Chen, Kewei; Goradia, Dhruman; Reiman, Eric M.; Wang, Yalin

2017-11-01

Fluorodeoxyglucose (FDG) positron emission tomography (PET) measures the decline in the regional cerebral metabolic rate for glucose, offering a reliable metabolic biomarker even on presymptomatic Alzheimer's disease (AD) patients. PET scans provide functional information that is unique and unavailable using other types of imaging. However, the computational efficacy of FDG-PET data alone, for the classification of various Alzheimers Diagnostic categories, has not been well studied. This motivates us to correctly discriminate various AD Diagnostic categories using FDG-PET data. Deep learning has improved state-of-the-art classification accuracies in the areas of speech, signal, image, video, text mining and recognition. We propose novel methods that involve probabilistic principal component analysis on max-pooled data and mean-pooled data for dimensionality reduction, and multilayer feed forward neural network which performs binary classification. Our experimental dataset consists of baseline data of subjects including 186 cognitively unimpaired (CU) subjects, 336 mild cognitive impairment (MCI) subjects with 158 Late MCI and 178 Early MCI, and 146 AD patients from Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset. We measured F1-measure, precision, recall, negative and positive predictive values with a 10-fold cross validation scheme. Our results indicate that our designed classifiers achieve competitive results while max pooling achieves better classification performance compared to mean-pooled features. Our deep model based research may advance FDG-PET analysis by demonstrating their potential as an effective imaging biomarker of AD.

Early Juno Era Optical Imaging and Analysis of Jupiter's Atmospheric Structure and Color with the NMSU Acousto-optic Imaging Camera

NASA Astrophysics Data System (ADS)

Dahl, E.; Chanover, N.; Voelz, D.; Kuehn, D.; Strycker, P.

2016-12-01

Jupiter's upper atmosphere is a highly dynamic system in which clouds and storms change color, shape, and size on variable timescales. The exact mechanism by which the deep atmosphere affects these changes in the uppermost cloud deck is still unknown. However, with Juno's arrival in July 2016, it is now possible to take detailed observations of the deep atmosphere with the spacecraft's Microwave Radiometer. By taking detailed optical measurements of Jupiter's uppermost cloud deck in conjunction with these microwave observations, we can provide a context in which to better understand these observations. Ultimately, we can utilize these two complementary datasets in order to thoroughly characterize Jupiter's atmosphere in terms of its vertical cloud structure, color distribution, and dynamical state throughout the Juno era. These optical data will also provide a complement to the near-IR sensitivity of the Jovian InfraRed Auroral Mapper and will expand on the limited spectral coverage of JunoCam. In order to obtain high spectral resolution images of Jupiter's atmosphere in the optical regime we use the New Mexico State University Acousto-optic Imaging Camera (NAIC). NAIC's acousto-optic tunable filter allows us to take hyperspectral image cubes of Jupiter from 450-950 nm at an average spectral resolution (λ/dλ) of 242. We present a preliminary analysis of two datasets obtained with NAIC at the Apache Point Observatory 3.5-m telescope: one pre-Juno dataset from March 2016 and the other from November 2016. From these data we derive low-resolution optical spectra of the Great Red Spot and a representative belt and zone to compare with previous work and laboratory measurements of candidate chromophore materials. Additionally, we compare these two datasets to inspect how the atmosphere has changed since before Juno arrived at Jupiter. NASA supported this work through award number NNX15AP34A.

A comparative study of two prediction models for brain tumor progression

NASA Astrophysics Data System (ADS)

Zhou, Deqi; Tran, Loc; Wang, Jihong; Li, Jiang

2015-03-01

MR diffusion tensor imaging (DTI) technique together with traditional T1 or T2 weighted MRI scans supplies rich information sources for brain cancer diagnoses. These images form large-scale, high-dimensional data sets. Due to the fact that significant correlations exist among these images, we assume low-dimensional geometry data structures (manifolds) are embedded in the high-dimensional space. Those manifolds might be hidden from radiologists because it is challenging for human experts to interpret high-dimensional data. Identification of the manifold is a critical step for successfully analyzing multimodal MR images. We have developed various manifold learning algorithms (Tran et al. 2011; Tran et al. 2013) for medical image analysis. This paper presents a comparative study of an incremental manifold learning scheme (Tran. et al. 2013) versus the deep learning model (Hinton et al. 2006) in the application of brain tumor progression prediction. The incremental manifold learning is a variant of manifold learning algorithm to handle large-scale datasets in which a representative subset of original data is sampled first to construct a manifold skeleton and remaining data points are then inserted into the skeleton by following their local geometry. The incremental manifold learning algorithm aims at mitigating the computational burden associated with traditional manifold learning methods for large-scale datasets. Deep learning is a recently developed multilayer perceptron model that has achieved start-of-the-art performances in many applications. A recent technique named "Dropout" can further boost the deep model by preventing weight coadaptation to avoid over-fitting (Hinton et al. 2012). We applied the two models on multiple MRI scans from four brain tumor patients to predict tumor progression and compared the performances of the two models in terms of average prediction accuracy, sensitivity, specificity and precision. The quantitative performance metrics were calculated as average over the four patients. Experimental results show that both the manifold learning and deep neural network models produced better results compared to using raw data and principle component analysis (PCA), and the deep learning model is a better method than manifold learning on this data set. The averaged sensitivity and specificity by deep learning are comparable with these by the manifold learning approach while its precision is considerably higher. This means that the predicted abnormal points by deep learning are more likely to correspond to the actual progression region.

Scalable High Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning

PubMed Central

Wu, Guorong; Kim, Minjeong; Wang, Qian; Munsell, Brent C.

2015-01-01

Feature selection is a critical step in deformable image registration. In particular, selecting the most discriminative features that accurately and concisely describe complex morphological patterns in image patches improves correspondence detection, which in turn improves image registration accuracy. Furthermore, since more and more imaging modalities are being invented to better identify morphological changes in medical imaging data,, the development of deformable image registration method that scales well to new image modalities or new image applications with little to no human intervention would have a significant impact on the medical image analysis community. To address these concerns, a learning-based image registration framework is proposed that uses deep learning to discover compact and highly discriminative features upon observed imaging data. Specifically, the proposed feature selection method uses a convolutional stacked auto-encoder to identify intrinsic deep feature representations in image patches. Since deep learning is an unsupervised learning method, no ground truth label knowledge is required. This makes the proposed feature selection method more flexible to new imaging modalities since feature representations can be directly learned from the observed imaging data in a very short amount of time. Using the LONI and ADNI imaging datasets, image registration performance was compared to two existing state-of-the-art deformable image registration methods that use handcrafted features. To demonstrate the scalability of the proposed image registration framework image registration experiments were conducted on 7.0-tesla brain MR images. In all experiments, the results showed the new image registration framework consistently demonstrated more accurate registration results when compared to state-of-the-art. PMID:26552069

Scalable High-Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning.

PubMed

Wu, Guorong; Kim, Minjeong; Wang, Qian; Munsell, Brent C; Shen, Dinggang

2016-07-01

Feature selection is a critical step in deformable image registration. In particular, selecting the most discriminative features that accurately and concisely describe complex morphological patterns in image patches improves correspondence detection, which in turn improves image registration accuracy. Furthermore, since more and more imaging modalities are being invented to better identify morphological changes in medical imaging data, the development of deformable image registration method that scales well to new image modalities or new image applications with little to no human intervention would have a significant impact on the medical image analysis community. To address these concerns, a learning-based image registration framework is proposed that uses deep learning to discover compact and highly discriminative features upon observed imaging data. Specifically, the proposed feature selection method uses a convolutional stacked autoencoder to identify intrinsic deep feature representations in image patches. Since deep learning is an unsupervised learning method, no ground truth label knowledge is required. This makes the proposed feature selection method more flexible to new imaging modalities since feature representations can be directly learned from the observed imaging data in a very short amount of time. Using the LONI and ADNI imaging datasets, image registration performance was compared to two existing state-of-the-art deformable image registration methods that use handcrafted features. To demonstrate the scalability of the proposed image registration framework, image registration experiments were conducted on 7.0-T brain MR images. In all experiments, the results showed that the new image registration framework consistently demonstrated more accurate registration results when compared to state of the art.

Applications of deep convolutional neural networks to digitized natural history collections

PubMed Central

Frandsen, Paul B.; Dikow, Rebecca B.; Brown, Abel; Orli, Sylvia; Peters, Melinda; Metallo, Adam; Funk, Vicki A.; Dorr, Laurence J.

2017-01-01

Abstract Natural history collections contain data that are critical for many scientific endeavors. Recent efforts in mass digitization are generating large datasets from these collections that can provide unprecedented insight. Here, we present examples of how deep convolutional neural networks can be applied in analyses of imaged herbarium specimens. We first demonstrate that a convolutional neural network can detect mercury-stained specimens across a collection with 90% accuracy. We then show that such a network can correctly distinguish two morphologically similar plant families 96% of the time. Discarding the most challenging specimen images increases accuracy to 94% and 99%, respectively. These results highlight the importance of mass digitization and deep learning approaches and reveal how they can together deliver powerful new investigative tools. PMID:29200929

Image quality assessment using deep convolutional networks

NASA Astrophysics Data System (ADS)

Li, Yezhou; Ye, Xiang; Li, Yong

2017-12-01

This paper proposes a method of accurately assessing image quality without a reference image by using a deep convolutional neural network. Existing training based methods usually utilize a compact set of linear filters for learning features of images captured by different sensors to assess their quality. These methods may not be able to learn the semantic features that are intimately related with the features used in human subject assessment. Observing this drawback, this work proposes training a deep convolutional neural network (CNN) with labelled images for image quality assessment. The ReLU in the CNN allows non-linear transformations for extracting high-level image features, providing a more reliable assessment of image quality than linear filters. To enable the neural network to take images of any arbitrary size as input, the spatial pyramid pooling (SPP) is introduced connecting the top convolutional layer and the fully-connected layer. In addition, the SPP makes the CNN robust to object deformations to a certain extent. The proposed method taking an image as input carries out an end-to-end learning process, and outputs the quality of the image. It is tested on public datasets. Experimental results show that it outperforms existing methods by a large margin and can accurately assess the image quality on images taken by different sensors of varying sizes.

Retinal fundus images for glaucoma analysis: the RIGA dataset

NASA Astrophysics Data System (ADS)

Almazroa, Ahmed; Alodhayb, Sami; Osman, Essameldin; Ramadan, Eslam; Hummadi, Mohammed; Dlaim, Mohammed; Alkatee, Muhannad; Raahemifar, Kaamran; Lakshminarayanan, Vasudevan

2018-03-01

Glaucoma neuropathy is a major cause of irreversible blindness worldwide. Current models of chronic care will not be able to close the gap of growing prevalence of glaucoma and challenges for access to healthcare services. Teleophthalmology is being developed to close this gap. In order to develop automated techniques for glaucoma detection which can be used in tele-ophthalmology we have developed a large retinal fundus dataset. A de-identified dataset of retinal fundus images for glaucoma analysis (RIGA) was derived from three sources for a total of 750 images. The optic cup and disc boundaries for each image was marked and annotated manually by six experienced ophthalmologists and included the cup to disc (CDR) estimates. Six parameters were extracted and assessed (the disc area and centroid, cup area and centroid, horizontal and vertical cup to disc ratios) among the ophthalmologists. The inter-observer annotations were compared by calculating the standard deviation (SD) for every image between the six ophthalmologists in order to determine if the outliers amongst the six and was used to filter the corresponding images. The data set will be made available to the research community in order to crowd source other analysis from other research groups in order to develop, validate and implement analysis algorithms appropriate for tele-glaucoma assessment. The RIGA dataset can be freely accessed online through University of Michigan, Deep Blue website (doi:10.7302/Z23R0R29).

DeepSAT: A Deep Learning Approach to Tree-Cover Delineation in 1-m NAIP Imagery for the Continental United States

NASA Technical Reports Server (NTRS)

Ganguly, Sangram; Basu, Saikat; Nemani, Ramakrishna R.; Mukhopadhyay, Supratik; Michaelis, Andrew; Votava, Petr

2016-01-01

High resolution tree cover classification maps are needed to increase the accuracy of current land ecosystem and climate model outputs. Limited studies are in place that demonstrates the state-of-the-art in deriving very high resolution (VHR) tree cover products. In addition, most methods heavily rely on commercial softwares that are difficult to scale given the region of study (e.g. continents to globe). Complexities in present approaches relate to (a) scalability of the algorithm, (b) large image data processing (compute and memory intensive), (c) computational cost, (d) massively parallel architecture, and (e) machine learning automation. In addition, VHR satellite datasets are of the order of terabytes and features extracted from these datasets are of the order of petabytes. In our present study, we have acquired the National Agriculture Imagery Program (NAIP) dataset for the Continental United States at a spatial resolution of 1-m. This data comes as image tiles (a total of quarter million image scenes with 60 million pixels) and has a total size of 65 terabytes for a single acquisition. Features extracted from the entire dataset would amount to 8-10 petabytes. In our proposed approach, we have implemented a novel semi-automated machine learning algorithm rooted on the principles of "deep learning" to delineate the percentage of tree cover. Using the NASA Earth Exchange (NEX) initiative, we have developed an end-to-end architecture by integrating a segmentation module based on Statistical Region Merging, a classification algorithm using Deep Belief Network and a structured prediction algorithm using Conditional Random Fields to integrate the results from the segmentation and classification modules to create per-pixel class labels. The training process is scaled up using the power of GPUs and the prediction is scaled to quarter million NAIP tiles spanning the whole of Continental United States using the NEX HPC supercomputing cluster. An initial pilot over the state of California spanning a total of 11,095 NAIP tiles covering a total geographical area of 163,696 sq. miles has produced true positive rates of around 88 percent for fragmented forests and 74 percent for urban tree cover areas, with false positive rates lower than 2 percent for both landscapes.

DeepSAT: A Deep Learning Approach to Tree-cover Delineation in 1-m NAIP Imagery for the Continental United States

NASA Astrophysics Data System (ADS)

Ganguly, S.; Basu, S.; Nemani, R. R.; Mukhopadhyay, S.; Michaelis, A.; Votava, P.

2016-12-01

High resolution tree cover classification maps are needed to increase the accuracy of current land ecosystem and climate model outputs. Limited studies are in place that demonstrates the state-of-the-art in deriving very high resolution (VHR) tree cover products. In addition, most methods heavily rely on commercial softwares that are difficult to scale given the region of study (e.g. continents to globe). Complexities in present approaches relate to (a) scalability of the algorithm, (b) large image data processing (compute and memory intensive), (c) computational cost, (d) massively parallel architecture, and (e) machine learning automation. In addition, VHR satellite datasets are of the order of terabytes and features extracted from these datasets are of the order of petabytes. In our present study, we have acquired the National Agriculture Imagery Program (NAIP) dataset for the Continental United States at a spatial resolution of 1-m. This data comes as image tiles (a total of quarter million image scenes with 60 million pixels) and has a total size of 65 terabytes for a single acquisition. Features extracted from the entire dataset would amount to 8-10 petabytes. In our proposed approach, we have implemented a novel semi-automated machine learning algorithm rooted on the principles of "deep learning" to delineate the percentage of tree cover. Using the NASA Earth Exchange (NEX) initiative, we have developed an end-to-end architecture by integrating a segmentation module based on Statistical Region Merging, a classification algorithm using Deep Belief Network and a structured prediction algorithm using Conditional Random Fields to integrate the results from the segmentation and classification modules to create per-pixel class labels. The training process is scaled up using the power of GPUs and the prediction is scaled to quarter million NAIP tiles spanning the whole of Continental United States using the NEX HPC supercomputing cluster. An initial pilot over the state of California spanning a total of 11,095 NAIP tiles covering a total geographical area of 163,696 sq. miles has produced true positive rates of around 88% for fragmented forests and 74% for urban tree cover areas, with false positive rates lower than 2% for both landscapes.

Deep classification hashing for person re-identification

NASA Astrophysics Data System (ADS)

Wang, Jiabao; Li, Yang; Zhang, Xiancai; Miao, Zhuang; Tao, Gang

2018-04-01

As the development of surveillance in public, person re-identification becomes more and more important. The largescale databases call for efficient computation and storage, hashing technique is one of the most important methods. In this paper, we proposed a new deep classification hashing network by introducing a new binary appropriation layer in the traditional ImageNet pre-trained CNN models. It outputs binary appropriate features, which can be easily quantized into binary hash-codes for hamming similarity comparison. Experiments show that our deep hashing method can outperform the state-of-the-art methods on the public CUHK03 and Market1501 datasets.

Deep learning based state recognition of substation switches

NASA Astrophysics Data System (ADS)

Wang, Jin

2018-06-01

Different from the traditional method which recognize the state of substation switches based on the running rules of electrical power system, this work proposes a novel convolutional neuron network-based state recognition approach of substation switches. Inspired by the theory of transfer learning, we first establish a convolutional neuron network model trained on the large-scale image set ILSVRC2012, then the restricted Boltzmann machine is employed to replace the full connected layer of the convolutional neuron network and trained on our small image dataset of 110kV substation switches to get a stronger model. Experiments conducted on our image dataset of 110kV substation switches show that, the proposed approach can be applicable to the substation to reduce the running cost and implement the real unattended operation.

Three-Category Classification of Magnetic Resonance Hearing Loss Images Based on Deep Autoencoder.

PubMed

Jia, Wenjuan; Yang, Ming; Wang, Shui-Hua

2017-09-11

Hearing loss, a partial or total inability to hear, is known as hearing impairment. Untreated hearing loss can have a bad effect on normal social communication, and it can cause psychological problems in patients. Therefore, we design a three-category classification system to detect the specific category of hearing loss, which is beneficial to be treated in time for patients. Before the training and test stages, we use the technology of data augmentation to produce a balanced dataset. Then we use deep autoencoder neural network to classify the magnetic resonance brain images. In the stage of deep autoencoder, we use stacked sparse autoencoder to generate visual features, and softmax layer to classify the different brain images into three categories of hearing loss. Our method can obtain good experimental results. The overall accuracy of our method is 99.5%, and the time consuming is 0.078 s per brain image. Our proposed method based on stacked sparse autoencoder works well in classification of hearing loss images. The overall accuracy of our method is 4% higher than the best of state-of-the-art approaches.

Combination of rs-fMRI and sMRI Data to Discriminate Autism Spectrum Disorders in Young Children Using Deep Belief Network.

PubMed

Akhavan Aghdam, Maryam; Sharifi, Arash; Pedram, Mir Mohsen

2018-05-07

In recent years, the use of advanced magnetic resonance (MR) imaging methods such as functional magnetic resonance imaging (fMRI) and structural magnetic resonance imaging (sMRI) has recorded a great increase in neuropsychiatric disorders. Deep learning is a branch of machine learning that is increasingly being used for applications of medical image analysis such as computer-aided diagnosis. In a bid to classify and represent learning tasks, this study utilized one of the most powerful deep learning algorithms (deep belief network (DBN)) for the combination of data from Autism Brain Imaging Data Exchange I and II (ABIDE I and ABIDE II) datasets. The DBN was employed so as to focus on the combination of resting-state fMRI (rs-fMRI), gray matter (GM), and white matter (WM) data. This was done based on the brain regions that were defined using the automated anatomical labeling (AAL), in order to classify autism spectrum disorders (ASDs) from typical controls (TCs). Since the diagnosis of ASD is much more effective at an early age, only 185 individuals (116 ASD and 69 TC) ranging in age from 5 to 10 years were included in this analysis. In contrast, the proposed method is used to exploit the latent or abstract high-level features inside rs-fMRI and sMRI data while the old methods consider only the simple low-level features extracted from neuroimages. Moreover, combining multiple data types and increasing the depth of DBN can improve classification accuracy. In this study, the best combination comprised rs-fMRI, GM, and WM for DBN of depth 3 with 65.56% accuracy (sensitivity = 84%, specificity = 32.96%, F1 score = 74.76%) obtained via 10-fold cross-validation. This result outperforms previously presented methods on ABIDE I dataset.

Validation and Uncertainty Estimates for MODIS Collection 6 "Deep Blue" Aerosol Data

NASA Technical Reports Server (NTRS)

Sayer, A. M.; Hsu, N. C.; Bettenhausen, C.; Jeong, M.-J.

2013-01-01

The "Deep Blue" aerosol optical depth (AOD) retrieval algorithm was introduced in Collection 5 of the Moderate Resolution Imaging Spectroradiometer (MODIS) product suite, and complemented the existing "Dark Target" land and ocean algorithms by retrieving AOD over bright arid land surfaces, such as deserts. The forthcoming Collection 6 of MODIS products will include a "second generation" Deep Blue algorithm, expanding coverage to all cloud-free and snow-free land surfaces. The Deep Blue dataset will also provide an estimate of the absolute uncertainty on AOD at 550 nm for each retrieval. This study describes the validation of Deep Blue Collection 6 AOD at 550 nm (Tau(sub M)) from MODIS Aqua against Aerosol Robotic Network (AERONET) data from 60 sites to quantify these uncertainties. The highest quality (denoted quality assurance flag value 3) data are shown to have an absolute uncertainty of approximately (0.086+0.56Tau(sub M))/AMF, where AMF is the geometric air mass factor. For a typical AMF of 2.8, this is approximately 0.03+0.20Tau(sub M), comparable in quality to other satellite AOD datasets. Regional variability of retrieval performance and comparisons against Collection 5 results are also discussed.

Semantic labeling of high-resolution aerial images using an ensemble of fully convolutional networks

NASA Astrophysics Data System (ADS)

Sun, Xiaofeng; Shen, Shuhan; Lin, Xiangguo; Hu, Zhanyi

2017-10-01

High-resolution remote sensing data classification has been a challenging and promising research topic in the community of remote sensing. In recent years, with the rapid advances of deep learning, remarkable progress has been made in this field, which facilitates a transition from hand-crafted features designing to an automatic end-to-end learning. A deep fully convolutional networks (FCNs) based ensemble learning method is proposed to label the high-resolution aerial images. To fully tap the potentials of FCNs, both the Visual Geometry Group network and a deeper residual network, ResNet, are employed. Furthermore, to enlarge training samples with diversity and gain better generalization, in addition to the commonly used data augmentation methods (e.g., rotation, multiscale, and aspect ratio) in the literature, aerial images from other datasets are also collected for cross-scene learning. Finally, we combine these learned models to form an effective FCN ensemble and refine the results using a fully connected conditional random field graph model. Experiments on the ISPRS 2-D Semantic Labeling Contest dataset show that our proposed end-to-end classification method achieves an overall accuracy of 90.7%, a state-of-the-art in the field.

Identifying Corresponding Patches in SAR and Optical Images With a Pseudo-Siamese CNN

NASA Astrophysics Data System (ADS)

Hughes, Lloyd H.; Schmitt, Michael; Mou, Lichao; Wang, Yuanyuan; Zhu, Xiao Xiang

2018-05-01

In this letter, we propose a pseudo-siamese convolutional neural network (CNN) architecture that enables to solve the task of identifying corresponding patches in very-high-resolution (VHR) optical and synthetic aperture radar (SAR) remote sensing imagery. Using eight convolutional layers each in two parallel network streams, a fully connected layer for the fusion of the features learned in each stream, and a loss function based on binary cross-entropy, we achieve a one-hot indication if two patches correspond or not. The network is trained and tested on an automatically generated dataset that is based on a deterministic alignment of SAR and optical imagery via previously reconstructed and subsequently co-registered 3D point clouds. The satellite images, from which the patches comprising our dataset are extracted, show a complex urban scene containing many elevated objects (i.e. buildings), thus providing one of the most difficult experimental environments. The achieved results show that the network is able to predict corresponding patches with high accuracy, thus indicating great potential for further development towards a generalized multi-sensor key-point matching procedure. Index Terms-synthetic aperture radar (SAR), optical imagery, data fusion, deep learning, convolutional neural networks (CNN), image matching, deep matching

Semantic Segmentation and Difference Extraction via Time Series Aerial Video Camera and its Application

NASA Astrophysics Data System (ADS)

Amit, S. N. K.; Saito, S.; Sasaki, S.; Kiyoki, Y.; Aoki, Y.

2015-04-01

Google earth with high-resolution imagery basically takes months to process new images before online updates. It is a time consuming and slow process especially for post-disaster application. The objective of this research is to develop a fast and effective method of updating maps by detecting local differences occurred over different time series; where only region with differences will be updated. In our system, aerial images from Massachusetts's road and building open datasets, Saitama district datasets are used as input images. Semantic segmentation is then applied to input images. Semantic segmentation is a pixel-wise classification of images by implementing deep neural network technique. Deep neural network technique is implemented due to being not only efficient in learning highly discriminative image features such as road, buildings etc., but also partially robust to incomplete and poorly registered target maps. Then, aerial images which contain semantic information are stored as database in 5D world map is set as ground truth images. This system is developed to visualise multimedia data in 5 dimensions; 3 dimensions as spatial dimensions, 1 dimension as temporal dimension, and 1 dimension as degenerated dimensions of semantic and colour combination dimension. Next, ground truth images chosen from database in 5D world map and a new aerial image with same spatial information but different time series are compared via difference extraction method. The map will only update where local changes had occurred. Hence, map updating will be cheaper, faster and more effective especially post-disaster application, by leaving unchanged region and only update changed region.

Deep Learning Nuclei Detection in Digitized Histology Images by Superpixels

PubMed Central

Sornapudi, Sudhir; Stanley, Ronald Joe; Stoecker, William V.; Almubarak, Haidar; Long, Rodney; Antani, Sameer; Thoma, George; Zuna, Rosemary; Frazier, Shelliane R.

2018-01-01

Background: Advances in image analysis and computational techniques have facilitated automatic detection of critical features in histopathology images. Detection of nuclei is critical for squamous epithelium cervical intraepithelial neoplasia (CIN) classification into normal, CIN1, CIN2, and CIN3 grades. Methods: In this study, a deep learning (DL)-based nuclei segmentation approach is investigated based on gathering localized information through the generation of superpixels using a simple linear iterative clustering algorithm and training with a convolutional neural network. Results: The proposed approach was evaluated on a dataset of 133 digitized histology images and achieved an overall nuclei detection (object-based) accuracy of 95.97%, with demonstrated improvement over imaging-based and clustering-based benchmark techniques. Conclusions: The proposed DL-based nuclei segmentation Method with superpixel analysis has shown improved segmentation results in comparison to state-of-the-art methods. PMID:29619277

Exploring the complementarity of THz pulse imaging and DCE-MRIs: Toward a unified multi-channel classification and a deep learning framework.

PubMed

Yin, X-X; Zhang, Y; Cao, J; Wu, J-L; Hadjiloucas, S

2016-12-01

We provide a comprehensive account of recent advances in biomedical image analysis and classification from two complementary imaging modalities: terahertz (THz) pulse imaging and dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). The work aims to highlight underlining commonalities in both data structures so that a common multi-channel data fusion framework can be developed. Signal pre-processing in both datasets is discussed briefly taking into consideration advances in multi-resolution analysis and model based fractional order calculus system identification. Developments in statistical signal processing using principal component and independent component analysis are also considered. These algorithms have been developed independently by the THz-pulse imaging and DCE-MRI communities, and there is scope to place them in a common multi-channel framework to provide better software standardization at the pre-processing de-noising stage. A comprehensive discussion of feature selection strategies is also provided and the importance of preserving textural information is highlighted. Feature extraction and classification methods taking into consideration recent advances in support vector machine (SVM) and extreme learning machine (ELM) classifiers and their complex extensions are presented. An outlook on Clifford algebra classifiers and deep learning techniques suitable to both types of datasets is also provided. The work points toward the direction of developing a new unified multi-channel signal processing framework for biomedical image analysis that will explore synergies from both sensing modalities for inferring disease proliferation. Copyright © 2016 Elsevier Ireland Ltd. All rights reserved.

Shakeout: A New Approach to Regularized Deep Neural Network Training.

PubMed

Kang, Guoliang; Li, Jun; Tao, Dacheng

2018-05-01

Recent years have witnessed the success of deep neural networks in dealing with a plenty of practical problems. Dropout has played an essential role in many successful deep neural networks, by inducing regularization in the model training. In this paper, we present a new regularized training approach: Shakeout. Instead of randomly discarding units as Dropout does at the training stage, Shakeout randomly chooses to enhance or reverse each unit's contribution to the next layer. This minor modification of Dropout has the statistical trait: the regularizer induced by Shakeout adaptively combines , and regularization terms. Our classification experiments with representative deep architectures on image datasets MNIST, CIFAR-10 and ImageNet show that Shakeout deals with over-fitting effectively and outperforms Dropout. We empirically demonstrate that Shakeout leads to sparser weights under both unsupervised and supervised settings. Shakeout also leads to the grouping effect of the input units in a layer. Considering the weights in reflecting the importance of connections, Shakeout is superior to Dropout, which is valuable for the deep model compression. Moreover, we demonstrate that Shakeout can effectively reduce the instability of the training process of the deep architecture.

Deep SOMs for automated feature extraction and classification from big data streaming

NASA Astrophysics Data System (ADS)

Sakkari, Mohamed; Ejbali, Ridha; Zaied, Mourad

2017-03-01

In this paper, we proposed a deep self-organizing map model (Deep-SOMs) for automated features extracting and learning from big data streaming which we benefit from the framework Spark for real time streams and highly parallel data processing. The SOMs deep architecture is based on the notion of abstraction (patterns automatically extract from the raw data, from the less to more abstract). The proposed model consists of three hidden self-organizing layers, an input and an output layer. Each layer is made up of a multitude of SOMs, each map only focusing at local headmistress sub-region from the input image. Then, each layer trains the local information to generate more overall information in the higher layer. The proposed Deep-SOMs model is unique in terms of the layers architecture, the SOMs sampling method and learning. During the learning stage we use a set of unsupervised SOMs for feature extraction. We validate the effectiveness of our approach on large data sets such as Leukemia dataset and SRBCT. Results of comparison have shown that the Deep-SOMs model performs better than many existing algorithms for images classification.

A Deep Learning Approach to Neuroanatomical Characterisation of Alzheimer's Disease.

PubMed

Ambastha, Abhinit Kumar; Leong, Tze-Yun

2017-01-01

Alzheimer's disease (AD) is a neurological degenerative disorder that leads to progressive mental deterioration. This work introduces a computational approach to improve our understanding of the progression of AD. We use ensemble learning methods and deep neural networks to identify salient structural correlations among brain regions that degenerate together in AD; this provides an understanding of how AD progresses in the brain. The proposed technique has a classification accuracy of 81.79% for AD against healthy subjects using a single modality imaging dataset.

Deep learning with domain adaptation for accelerated projection-reconstruction MR.

PubMed

Han, Yoseob; Yoo, Jaejun; Kim, Hak Hee; Shin, Hee Jung; Sung, Kyunghyun; Ye, Jong Chul

2018-09-01

The radial k-space trajectory is a well-established sampling trajectory used in conjunction with magnetic resonance imaging. However, the radial k-space trajectory requires a large number of radial lines for high-resolution reconstruction. Increasing the number of radial lines causes longer acquisition time, making it more difficult for routine clinical use. On the other hand, if we reduce the number of radial lines, streaking artifact patterns are unavoidable. To solve this problem, we propose a novel deep learning approach with domain adaptation to restore high-resolution MR images from under-sampled k-space data. The proposed deep network removes the streaking artifacts from the artifact corrupted images. To address the situation given the limited available data, we propose a domain adaptation scheme that employs a pre-trained network using a large number of X-ray computed tomography (CT) or synthesized radial MR datasets, which is then fine-tuned with only a few radial MR datasets. The proposed method outperforms existing compressed sensing algorithms, such as the total variation and PR-FOCUSS methods. In addition, the calculation time is several orders of magnitude faster than the total variation and PR-FOCUSS methods. Moreover, we found that pre-training using CT or MR data from similar organ data is more important than pre-training using data from the same modality for different organ. We demonstrate the possibility of a domain-adaptation when only a limited amount of MR data is available. The proposed method surpasses the existing compressed sensing algorithms in terms of the image quality and computation time. © 2018 International Society for Magnetic Resonance in Medicine.

A deep learning method for classifying mammographic breast density categories.

PubMed

Mohamed, Aly A; Berg, Wendie A; Peng, Hong; Luo, Yahong; Jankowitz, Rachel C; Wu, Shandong

2018-01-01

Mammographic breast density is an established risk marker for breast cancer and is visually assessed by radiologists in routine mammogram image reading, using four qualitative Breast Imaging and Reporting Data System (BI-RADS) breast density categories. It is particularly difficult for radiologists to consistently distinguish the two most common and most variably assigned BI-RADS categories, i.e., "scattered density" and "heterogeneously dense". The aim of this work was to investigate a deep learning-based breast density classifier to consistently distinguish these two categories, aiming at providing a potential computerized tool to assist radiologists in assigning a BI-RADS category in current clinical workflow. In this study, we constructed a convolutional neural network (CNN)-based model coupled with a large (i.e., 22,000 images) digital mammogram imaging dataset to evaluate the classification performance between the two aforementioned breast density categories. All images were collected from a cohort of 1,427 women who underwent standard digital mammography screening from 2005 to 2016 at our institution. The truths of the density categories were based on standard clinical assessment made by board-certified breast imaging radiologists. Effects of direct training from scratch solely using digital mammogram images and transfer learning of a pretrained model on a large nonmedical imaging dataset were evaluated for the specific task of breast density classification. In order to measure the classification performance, the CNN classifier was also tested on a refined version of the mammogram image dataset by removing some potentially inaccurately labeled images. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were used to measure the accuracy of the classifier. The AUC was 0.9421 when the CNN-model was trained from scratch on our own mammogram images, and the accuracy increased gradually along with an increased size of training samples. Using the pretrained model followed by a fine-tuning process with as few as 500 mammogram images led to an AUC of 0.9265. After removing the potentially inaccurately labeled images, AUC was increased to 0.9882 and 0.9857 for without and with the pretrained model, respectively, both significantly higher (P < 0.001) than when using the full imaging dataset. Our study demonstrated high classification accuracies between two difficult to distinguish breast density categories that are routinely assessed by radiologists. We anticipate that our approach will help enhance current clinical assessment of breast density and better support consistent density notification to patients in breast cancer screening. © 2017 American Association of Physicists in Medicine.

An open, multi-vendor, multi-field-strength brain MR dataset and analysis of publicly available skull stripping methods agreement.

PubMed

Souza, Roberto; Lucena, Oeslle; Garrafa, Julia; Gobbi, David; Saluzzi, Marina; Appenzeller, Simone; Rittner, Letícia; Frayne, Richard; Lotufo, Roberto

2018-04-15

This paper presents an open, multi-vendor, multi-field strength magnetic resonance (MR) T1-weighted volumetric brain imaging dataset, named Calgary-Campinas-359 (CC-359). The dataset is composed of images of older healthy adults (29-80 years) acquired on scanners from three vendors (Siemens, Philips and General Electric) at both 1.5 T and 3 T. CC-359 is comprised of 359 datasets, approximately 60 subjects per vendor and magnetic field strength. The dataset is approximately age and gender balanced, subject to the constraints of the available images. It provides consensus brain extraction masks for all volumes generated using supervised classification. Manual segmentation results for twelve randomly selected subjects performed by an expert are also provided. The CC-359 dataset allows investigation of 1) the influences of both vendor and magnetic field strength on quantitative analysis of brain MR; 2) parameter optimization for automatic segmentation methods; and potentially 3) machine learning classifiers with big data, specifically those based on deep learning methods, as these approaches require a large amount of data. To illustrate the utility of this dataset, we compared to the results of a supervised classifier, the results of eight publicly available skull stripping methods and one publicly available consensus algorithm. A linear mixed effects model analysis indicated that vendor (p-value<0.001) and magnetic field strength (p-value<0.001) have statistically significant impacts on skull stripping results. Copyright © 2017 Elsevier Inc. All rights reserved.

A web-based system for neural network based classification in temporomandibular joint osteoarthritis.

PubMed

de Dumast, Priscille; Mirabel, Clément; Cevidanes, Lucia; Ruellas, Antonio; Yatabe, Marilia; Ioshida, Marcos; Ribera, Nina Tubau; Michoud, Loic; Gomes, Liliane; Huang, Chao; Zhu, Hongtu; Muniz, Luciana; Shoukri, Brandon; Paniagua, Beatriz; Styner, Martin; Pieper, Steve; Budin, Francois; Vimort, Jean-Baptiste; Pascal, Laura; Prieto, Juan Carlos

2018-07-01

The purpose of this study is to describe the methodological innovations of a web-based system for storage, integration and computation of biomedical data, using a training imaging dataset to remotely compute a deep neural network classifier of temporomandibular joint osteoarthritis (TMJOA). This study imaging dataset consisted of three-dimensional (3D) surface meshes of mandibular condyles constructed from cone beam computed tomography (CBCT) scans. The training dataset consisted of 259 condyles, 105 from control subjects and 154 from patients with diagnosis of TMJ OA. For the image analysis classification, 34 right and left condyles from 17 patients (39.9 ± 11.7 years), who experienced signs and symptoms of the disease for less than 5 years, were included as the testing dataset. For the integrative statistical model of clinical, biological and imaging markers, the sample consisted of the same 17 test OA subjects and 17 age and sex matched control subjects (39.4 ± 15.4 years), who did not show any sign or symptom of OA. For these 34 subjects, a standardized clinical questionnaire, blood and saliva samples were also collected. The technological methodologies in this study include a deep neural network classifier of 3D condylar morphology (ShapeVariationAnalyzer, SVA), and a flexible web-based system for data storage, computation and integration (DSCI) of high dimensional imaging, clinical, and biological data. The DSCI system trained and tested the neural network, indicating 5 stages of structural degenerative changes in condylar morphology in the TMJ with 91% close agreement between the clinician consensus and the SVA classifier. The DSCI remotely ran with a novel application of a statistical analysis, the Multivariate Functional Shape Data Analysis, that computed high dimensional correlations between shape 3D coordinates, clinical pain levels and levels of biological markers, and then graphically displayed the computation results. The findings of this study demonstrate a comprehensive phenotypic characterization of TMJ health and disease at clinical, imaging and biological levels, using novel flexible and versatile open-source tools for a web-based system that provides advanced shape statistical analysis and a neural network based classification of temporomandibular joint osteoarthritis. Published by Elsevier Ltd.

Deep learning application: rubbish classification with aid of an android device

NASA Astrophysics Data System (ADS)

Liu, Sijiang; Jiang, Bo; Zhan, Jie

2017-06-01

Deep learning is a very hot topic currently in pattern recognition and artificial intelligence researches. Aiming at the practical problem that people usually don't know correct classifications some rubbish should belong to, based on the powerful image classification ability of the deep learning method, we have designed a prototype system to help users to classify kinds of rubbish. Firstly the CaffeNet Model was adopted for our classification network training on the ImageNet dataset, and the trained network was deployed on a web server. Secondly an android app was developed for users to capture images of unclassified rubbish, upload images to the web server for analyzing backstage and retrieve the feedback, so that users can obtain the classification guide by an android device conveniently. Tests on our prototype system of rubbish classification show that: an image of one single type of rubbish with origin shape can be better used to judge its classification, while an image containing kinds of rubbish or rubbish with changed shape may fail to help users to decide rubbish's classification. However, the system still shows promising auxiliary function for rubbish classification if the network training strategy can be optimized further.

Brain extraction from normal and pathological images: A joint PCA/Image-Reconstruction approach.

PubMed

Han, Xu; Kwitt, Roland; Aylward, Stephen; Bakas, Spyridon; Menze, Bjoern; Asturias, Alexander; Vespa, Paul; Van Horn, John; Niethammer, Marc

2018-08-01

Brain extraction from 3D medical images is a common pre-processing step. A variety of approaches exist, but they are frequently only designed to perform brain extraction from images without strong pathologies. Extracting the brain from images exhibiting strong pathologies, for example, the presence of a brain tumor or of a traumatic brain injury (TBI), is challenging. In such cases, tissue appearance may substantially deviate from normal tissue appearance and hence violates algorithmic assumptions for standard approaches to brain extraction; consequently, the brain may not be correctly extracted. This paper proposes a brain extraction approach which can explicitly account for pathologies by jointly modeling normal tissue appearance and pathologies. Specifically, our model uses a three-part image decomposition: (1) normal tissue appearance is captured by principal component analysis (PCA), (2) pathologies are captured via a total variation term, and (3) the skull and surrounding tissue is captured by a sparsity term. Due to its convexity, the resulting decomposition model allows for efficient optimization. Decomposition and image registration steps are alternated to allow statistical modeling of normal tissue appearance in a fixed atlas coordinate system. As a beneficial side effect, the decomposition model allows for the identification of potentially pathological areas and the reconstruction of a quasi-normal image in atlas space. We demonstrate the effectiveness of our approach on four datasets: the publicly available IBSR and LPBA40 datasets which show normal image appearance, the BRATS dataset containing images with brain tumors, and a dataset containing clinical TBI images. We compare the performance with other popular brain extraction models: ROBEX, BEaST, MASS, BET, BSE and a recently proposed deep learning approach. Our model performs better than these competing approaches on all four datasets. Specifically, our model achieves the best median (97.11) and mean (96.88) Dice scores over all datasets. The two best performing competitors, ROBEX and MASS, achieve scores of 96.23/95.62 and 96.67/94.25 respectively. Hence, our approach is an effective method for high quality brain extraction for a wide variety of images. Copyright © 2018 Elsevier Inc. All rights reserved.

Investigating the Use of Deep Convective Clouds (DCCT) to Monitor On-orbit Performance of the Geostationary Lightning Mapper (GLM) using Lightning Imaging Sensor (LIS) Measurements

NASA Technical Reports Server (NTRS)

Buechler, Dennis E.; Christian, Hugh J.; Koshak, William J.; Goodman, Steven J.

2013-01-01

There is a need to monitor the on-orbit performance of the Geostationary Lightning Mapper (GLM) on the Geostationary Operational Environmental Satellite R (GOES-R) for changes in instrument calibration that will affect GLM's lightning detection efficiency. GLM has no onboard calibration so GLM background radiance observations (available every 2.5 min) of Deep Convective Clouds (DCCs) are investigated as invariant targets to monitor GLM performance. Observations from the Lightning Imaging Sensor (LIS) and the Visible and Infrared Scanner (VIRS) onboard the Tropical Rainfall Measuring Mission (TRMM) satellite are used as proxy datasets for GLM and ABI 11 m measurements.

Classification of brain MRI with big data and deep 3D convolutional neural networks

NASA Astrophysics Data System (ADS)

Wegmayr, Viktor; Aitharaju, Sai; Buhmann, Joachim

2018-02-01

Our ever-aging society faces the growing problem of neurodegenerative diseases, in particular dementia. Magnetic Resonance Imaging provides a unique tool for non-invasive investigation of these brain diseases. However, it is extremely difficult for neurologists to identify complex disease patterns from large amounts of three-dimensional images. In contrast, machine learning excels at automatic pattern recognition from large amounts of data. In particular, deep learning has achieved impressive results in image classification. Unfortunately, its application to medical image classification remains difficult. We consider two reasons for this difficulty: First, volumetric medical image data is considerably scarcer than natural images. Second, the complexity of 3D medical images is much higher compared to common 2D images. To address the problem of small data set size, we assemble the largest dataset ever used for training a deep 3D convolutional neural network to classify brain images as healthy (HC), mild cognitive impairment (MCI) or Alzheimers disease (AD). We use more than 20.000 images from subjects of these three classes, which is almost 9x the size of the previously largest data set. The problem of high dimensionality is addressed by using a deep 3D convolutional neural network, which is state-of-the-art in large-scale image classification. We exploit its ability to process the images directly, only with standard preprocessing, but without the need for elaborate feature engineering. Compared to other work, our workflow is considerably simpler, which increases clinical applicability. Accuracy is measured on the ADNI+AIBL data sets, and the independent CADDementia benchmark.

Human fatigue expression recognition through image-based dynamic multi-information and bimodal deep learning

NASA Astrophysics Data System (ADS)

Zhao, Lei; Wang, Zengcai; Wang, Xiaojin; Qi, Yazhou; Liu, Qing; Zhang, Guoxin

2016-09-01

Human fatigue is an important cause of traffic accidents. To improve the safety of transportation, we propose, in this paper, a framework for fatigue expression recognition using image-based facial dynamic multi-information and a bimodal deep neural network. First, the landmark of face region and the texture of eye region, which complement each other in fatigue expression recognition, are extracted from facial image sequences captured by a single camera. Then, two stacked autoencoder neural networks are trained for landmark and texture, respectively. Finally, the two trained neural networks are combined by learning a joint layer on top of them to construct a bimodal deep neural network. The model can be used to extract a unified representation that fuses landmark and texture modalities together and classify fatigue expressions accurately. The proposed system is tested on a human fatigue dataset obtained from an actual driving environment. The experimental results demonstrate that the proposed method performs stably and robustly, and that the average accuracy achieves 96.2%.

Brain MRI analysis for Alzheimer's disease diagnosis using an ensemble system of deep convolutional neural networks.

PubMed

Islam, Jyoti; Zhang, Yanqing

2018-05-31

Alzheimer's disease is an incurable, progressive neurological brain disorder. Earlier detection of Alzheimer's disease can help with proper treatment and prevent brain tissue damage. Several statistical and machine learning models have been exploited by researchers for Alzheimer's disease diagnosis. Analyzing magnetic resonance imaging (MRI) is a common practice for Alzheimer's disease diagnosis in clinical research. Detection of Alzheimer's disease is exacting due to the similarity in Alzheimer's disease MRI data and standard healthy MRI data of older people. Recently, advanced deep learning techniques have successfully demonstrated human-level performance in numerous fields including medical image analysis. We propose a deep convolutional neural network for Alzheimer's disease diagnosis using brain MRI data analysis. While most of the existing approaches perform binary classification, our model can identify different stages of Alzheimer's disease and obtains superior performance for early-stage diagnosis. We conducted ample experiments to demonstrate that our proposed model outperformed comparative baselines on the Open Access Series of Imaging Studies dataset.

Deep neural networks for texture classification-A theoretical analysis.

PubMed

Basu, Saikat; Mukhopadhyay, Supratik; Karki, Manohar; DiBiano, Robert; Ganguly, Sangram; Nemani, Ramakrishna; Gayaka, Shreekant

2018-01-01

We investigate the use of Deep Neural Networks for the classification of image datasets where texture features are important for generating class-conditional discriminative representations. To this end, we first derive the size of the feature space for some standard textural features extracted from the input dataset and then use the theory of Vapnik-Chervonenkis dimension to show that hand-crafted feature extraction creates low-dimensional representations which help in reducing the overall excess error rate. As a corollary to this analysis, we derive for the first time upper bounds on the VC dimension of Convolutional Neural Network as well as Dropout and Dropconnect networks and the relation between excess error rate of Dropout and Dropconnect networks. The concept of intrinsic dimension is used to validate the intuition that texture-based datasets are inherently higher dimensional as compared to handwritten digits or other object recognition datasets and hence more difficult to be shattered by neural networks. We then derive the mean distance from the centroid to the nearest and farthest sampling points in an n-dimensional manifold and show that the Relative Contrast of the sample data vanishes as dimensionality of the underlying vector space tends to infinity. Copyright © 2017 Elsevier Ltd. All rights reserved.

Recurrent neural networks for breast lesion classification based on DCE-MRIs

NASA Astrophysics Data System (ADS)

Antropova, Natasha; Huynh, Benjamin; Giger, Maryellen

2018-02-01

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) plays a significant role in breast cancer screening, cancer staging, and monitoring response to therapy. Recently, deep learning methods are being rapidly incorporated in image-based breast cancer diagnosis and prognosis. However, most of the current deep learning methods make clinical decisions based on 2-dimentional (2D) or 3D images and are not well suited for temporal image data. In this study, we develop a deep learning methodology that enables integration of clinically valuable temporal components of DCE-MRIs into deep learning-based lesion classification. Our work is performed on a database of 703 DCE-MRI cases for the task of distinguishing benign and malignant lesions, and uses the area under the ROC curve (AUC) as the performance metric in conducting that task. We train a recurrent neural network, specifically a long short-term memory network (LSTM), on sequences of image features extracted from the dynamic MRI sequences. These features are extracted with VGGNet, a convolutional neural network pre-trained on a large dataset of natural images ImageNet. The features are obtained from various levels of the network, to capture low-, mid-, and high-level information about the lesion. Compared to a classification method that takes as input only images at a single time-point (yielding an AUC = 0.81 (se = 0.04)), our LSTM method improves lesion classification with an AUC of 0.85 (se = 0.03).

Image annotation by deep neural networks with attention shaping

NASA Astrophysics Data System (ADS)

Zheng, Kexin; Lv, Shaohe; Ma, Fang; Chen, Fei; Jin, Chi; Dou, Yong

2017-07-01

Image annotation is a task of assigning semantic labels to an image. Recently, deep neural networks with visual attention have been utilized successfully in many computer vision tasks. In this paper, we show that conventional attention mechanism is easily misled by the salient class, i.e., the attended region always contains part of the image area describing the content of salient class at different attention iterations. To this end, we propose a novel attention shaping mechanism, which aims to maximize the non-overlapping area between consecutive attention processes by taking into account the history of previous attention vectors. Several weighting polices are studied to utilize the history information in different manners. In two benchmark datasets, i.e., PASCAL VOC2012 and MIRFlickr-25k, the average precision is improved by up to 10% in comparison with the state-of-the-art annotation methods.

Deep Learning Methods for Quantifying Invasive Benthic Species in the Great Lakes

NASA Astrophysics Data System (ADS)

Billings, G.; Skinner, K.; Johnson-Roberson, M.

2017-12-01

In recent decades, invasive species such as the round goby and dreissenid mussels have greatly impacted the Great Lakes ecosystem. It is critical to monitor these species, model their distribution, and quantify the impacts on the native fisheries and surrounding ecosystem in order to develop an effective management response. However, data collection in underwater environments is challenging and expensive. Furthermore, the round goby is typically found in rocky habitats, which are inaccessible to standard survey techniques such as bottom trawling. In this work we propose a robotic system for visual data collection to automatically detect and quantify invasive round gobies and mussels in the Great Lakes. Robotic platforms equipped with cameras can perform efficient, cost-effective, low-bias benthic surveys. This data collection can be further optimized through automatic detection and annotation of the target species. Deep learning methods have shown success in image recognition tasks. However, these methods often rely on a labelled training dataset, with up to millions of labelled images. Hand labeling large numbers of images is expensive and often impracticable. Furthermore, data collected in the field may be sparse when only considering images that contain the objects of interest. It is easier to collect dense, clean data in controlled lab settings, but this data is not a realistic representation of real field environments. In this work, we propose a deep learning approach to generate a large set of labelled training data realistic of underwater environments in the field. To generate these images, first we draw random sample images of individual fish and mussels from a library of images captured in a controlled lab environment. Next, these randomly drawn samples will be automatically merged into natural background images. Finally, we will use a generative adversarial network (GAN) that incorporates constraints of the physical model of underwater light propagation to simulate the process of underwater image formation in various water conditions. The output of the GAN will be realistic looking annotated underwater images. This generated dataset of images will be used to train a classifier to identify round gobies and mussels in order to measure the biomass and abundance of these invasive species in the Great Lakes.

3D Deep Learning Angiography (3D-DLA) from C-arm Conebeam CT.

PubMed

Montoya, J C; Li, Y; Strother, C; Chen, G-H

2018-05-01

Deep learning is a branch of artificial intelligence that has demonstrated unprecedented performance in many medical imaging applications. Our purpose was to develop a deep learning angiography method to generate 3D cerebral angiograms from a single contrast-enhanced C-arm conebeam CT acquisition in order to reduce image artifacts and radiation dose. A set of 105 3D rotational angiography examinations were randomly selected from an internal data base. All were acquired using a clinical system in conjunction with a standard injection protocol. More than 150 million labeled voxels from 35 subjects were used for training. A deep convolutional neural network was trained to classify each image voxel into 3 tissue types (vasculature, bone, and soft tissue). The trained deep learning angiography model was then applied for tissue classification into a validation cohort of 8 subjects and a final testing cohort of the remaining 62 subjects. The final vasculature tissue class was used to generate the 3D deep learning angiography images. To quantify the generalization error of the trained model, we calculated the accuracy, sensitivity, precision, and Dice similarity coefficients for vasculature classification in relevant anatomy. The 3D deep learning angiography and clinical 3D rotational angiography images were subjected to a qualitative assessment for the presence of intersweep motion artifacts. Vasculature classification accuracy and 95% CI in the testing dataset were 98.7% (98.3%-99.1%). No residual signal from osseous structures was observed for any 3D deep learning angiography testing cases except for small regions in the otic capsule and nasal cavity compared with 37% (23/62) of the 3D rotational angiographies. Deep learning angiography accurately recreated the vascular anatomy of the 3D rotational angiography reconstructions without a mask. Deep learning angiography reduced misregistration artifacts induced by intersweep motion, and it reduced radiation exposure required to obtain clinically useful 3D rotational angiography. © 2018 by American Journal of Neuroradiology.

Forecast Modelling via Variations in Binary Image-Encoded Information Exploited by Deep Learning Neural Networks.

PubMed

Liu, Da; Xu, Ming; Niu, Dongxiao; Wang, Shoukai; Liang, Sai

2016-01-01

Traditional forecasting models fit a function approximation from dependent invariables to independent variables. However, they usually get into trouble when date are presented in various formats, such as text, voice and image. This study proposes a novel image-encoded forecasting method that input and output binary digital two-dimensional (2D) images are transformed from decimal data. Omitting any data analysis or cleansing steps for simplicity, all raw variables were selected and converted to binary digital images as the input of a deep learning model, convolutional neural network (CNN). Using shared weights, pooling and multiple-layer back-propagation techniques, the CNN was adopted to locate the nexus among variations in local binary digital images. Due to the computing capability that was originally developed for binary digital bitmap manipulation, this model has significant potential for forecasting with vast volume of data. The model was validated by a power loads predicting dataset from the Global Energy Forecasting Competition 2012.

Forecast Modelling via Variations in Binary Image-Encoded Information Exploited by Deep Learning Neural Networks

PubMed Central

Xu, Ming; Niu, Dongxiao; Wang, Shoukai; Liang, Sai

2016-01-01

Traditional forecasting models fit a function approximation from dependent invariables to independent variables. However, they usually get into trouble when date are presented in various formats, such as text, voice and image. This study proposes a novel image-encoded forecasting method that input and output binary digital two-dimensional (2D) images are transformed from decimal data. Omitting any data analysis or cleansing steps for simplicity, all raw variables were selected and converted to binary digital images as the input of a deep learning model, convolutional neural network (CNN). Using shared weights, pooling and multiple-layer back-propagation techniques, the CNN was adopted to locate the nexus among variations in local binary digital images. Due to the computing capability that was originally developed for binary digital bitmap manipulation, this model has significant potential for forecasting with vast volume of data. The model was validated by a power loads predicting dataset from the Global Energy Forecasting Competition 2012. PMID:27281032

Deep Learning at Chest Radiography: Automated Classification of Pulmonary Tuberculosis by Using Convolutional Neural Networks.

PubMed

Lakhani, Paras; Sundaram, Baskaran

2017-08-01

Purpose To evaluate the efficacy of deep convolutional neural networks (DCNNs) for detecting tuberculosis (TB) on chest radiographs. Materials and Methods Four deidentified HIPAA-compliant datasets were used in this study that were exempted from review by the institutional review board, which consisted of 1007 posteroanterior chest radiographs. The datasets were split into training (68.0%), validation (17.1%), and test (14.9%). Two different DCNNs, AlexNet and GoogLeNet, were used to classify the images as having manifestations of pulmonary TB or as healthy. Both untrained and pretrained networks on ImageNet were used, and augmentation with multiple preprocessing techniques. Ensembles were performed on the best-performing algorithms. For cases where the classifiers were in disagreement, an independent board-certified cardiothoracic radiologist blindly interpreted the images to evaluate a potential radiologist-augmented workflow. Receiver operating characteristic curves and areas under the curve (AUCs) were used to assess model performance by using the DeLong method for statistical comparison of receiver operating characteristic curves. Results The best-performing classifier had an AUC of 0.99, which was an ensemble of the AlexNet and GoogLeNet DCNNs. The AUCs of the pretrained models were greater than that of the untrained models (P < .001). Augmenting the dataset further increased accuracy (P values for AlexNet and GoogLeNet were .03 and .02, respectively). The DCNNs had disagreement in 13 of the 150 test cases, which were blindly reviewed by a cardiothoracic radiologist, who correctly interpreted all 13 cases (100%). This radiologist-augmented approach resulted in a sensitivity of 97.3% and specificity 100%. Conclusion Deep learning with DCNNs can accurately classify TB at chest radiography with an AUC of 0.99. A radiologist-augmented approach for cases where there was disagreement among the classifiers further improved accuracy. © RSNA, 2017.

Fabric defect detection based on visual saliency using deep feature and low-rank recovery

NASA Astrophysics Data System (ADS)

Liu, Zhoufeng; Wang, Baorui; Li, Chunlei; Li, Bicao; Dong, Yan

2018-04-01

Fabric defect detection plays an important role in improving the quality of fabric product. In this paper, a novel fabric defect detection method based on visual saliency using deep feature and low-rank recovery was proposed. First, unsupervised training is carried out by the initial network parameters based on MNIST large datasets. The supervised fine-tuning of fabric image library based on Convolutional Neural Networks (CNNs) is implemented, and then more accurate deep neural network model is generated. Second, the fabric images are uniformly divided into the image block with the same size, then we extract their multi-layer deep features using the trained deep network. Thereafter, all the extracted features are concentrated into a feature matrix. Third, low-rank matrix recovery is adopted to divide the feature matrix into the low-rank matrix which indicates the background and the sparse matrix which indicates the salient defect. In the end, the iterative optimal threshold segmentation algorithm is utilized to segment the saliency maps generated by the sparse matrix to locate the fabric defect area. Experimental results demonstrate that the feature extracted by CNN is more suitable for characterizing the fabric texture than the traditional LBP, HOG and other hand-crafted features extraction method, and the proposed method can accurately detect the defect regions of various fabric defects, even for the image with complex texture.

Deep Learning for Image-Based Cassava Disease Detection.

PubMed

Ramcharan, Amanda; Baranowski, Kelsee; McCloskey, Peter; Ahmed, Babuali; Legg, James; Hughes, David P

2017-01-01

Cassava is the third largest source of carbohydrates for human food in the world but is vulnerable to virus diseases, which threaten to destabilize food security in sub-Saharan Africa. Novel methods of cassava disease detection are needed to support improved control which will prevent this crisis. Image recognition offers both a cost effective and scalable technology for disease detection. New deep learning models offer an avenue for this technology to be easily deployed on mobile devices. Using a dataset of cassava disease images taken in the field in Tanzania, we applied transfer learning to train a deep convolutional neural network to identify three diseases and two types of pest damage (or lack thereof). The best trained model accuracies were 98% for brown leaf spot (BLS), 96% for red mite damage (RMD), 95% for green mite damage (GMD), 98% for cassava brown streak disease (CBSD), and 96% for cassava mosaic disease (CMD). The best model achieved an overall accuracy of 93% for data not used in the training process. Our results show that the transfer learning approach for image recognition of field images offers a fast, affordable, and easily deployable strategy for digital plant disease detection.

A deep learning framework for supporting the classification of breast lesions in ultrasound images.

PubMed

Han, Seokmin; Kang, Ho-Kyung; Jeong, Ja-Yeon; Park, Moon-Ho; Kim, Wonsik; Bang, Won-Chul; Seong, Yeong-Kyeong

2017-09-15

In this research, we exploited the deep learning framework to differentiate the distinctive types of lesions and nodules in breast acquired with ultrasound imaging. A biopsy-proven benchmarking dataset was built from 5151 patients cases containing a total of 7408 ultrasound breast images, representative of semi-automatically segmented lesions associated with masses. The dataset comprised 4254 benign and 3154 malignant lesions. The developed method includes histogram equalization, image cropping and margin augmentation. The GoogLeNet convolutionary neural network was trained to the database to differentiate benign and malignant tumors. The networks were trained on the data with augmentation and the data without augmentation. Both of them showed an area under the curve of over 0.9. The networks showed an accuracy of about 0.9 (90%), a sensitivity of 0.86 and a specificity of 0.96. Although target regions of interest (ROIs) were selected by radiologists, meaning that radiologists still have to point out the location of the ROI, the classification of malignant lesions showed promising results. If this method is used by radiologists in clinical situations it can classify malignant lesions in a short time and support the diagnosis of radiologists in discriminating malignant lesions. Therefore, the proposed method can work in tandem with human radiologists to improve performance, which is a fundamental purpose of computer-aided diagnosis.

Robust Vehicle Detection in Aerial Images Based on Cascaded Convolutional Neural Networks.

PubMed

Zhong, Jiandan; Lei, Tao; Yao, Guangle

2017-11-24

Vehicle detection in aerial images is an important and challenging task. Traditionally, many target detection models based on sliding-window fashion were developed and achieved acceptable performance, but these models are time-consuming in the detection phase. Recently, with the great success of convolutional neural networks (CNNs) in computer vision, many state-of-the-art detectors have been designed based on deep CNNs. However, these CNN-based detectors are inefficient when applied in aerial image data due to the fact that the existing CNN-based models struggle with small-size object detection and precise localization. To improve the detection accuracy without decreasing speed, we propose a CNN-based detection model combining two independent convolutional neural networks, where the first network is applied to generate a set of vehicle-like regions from multi-feature maps of different hierarchies and scales. Because the multi-feature maps combine the advantage of the deep and shallow convolutional layer, the first network performs well on locating the small targets in aerial image data. Then, the generated candidate regions are fed into the second network for feature extraction and decision making. Comprehensive experiments are conducted on the Vehicle Detection in Aerial Imagery (VEDAI) dataset and Munich vehicle dataset. The proposed cascaded detection model yields high performance, not only in detection accuracy but also in detection speed.

A deep learning framework for supporting the classification of breast lesions in ultrasound images

NASA Astrophysics Data System (ADS)

Han, Seokmin; Kang, Ho-Kyung; Jeong, Ja-Yeon; Park, Moon-Ho; Kim, Wonsik; Bang, Won-Chul; Seong, Yeong-Kyeong

2017-10-01

In this research, we exploited the deep learning framework to differentiate the distinctive types of lesions and nodules in breast acquired with ultrasound imaging. A biopsy-proven benchmarking dataset was built from 5151 patients cases containing a total of 7408 ultrasound breast images, representative of semi-automatically segmented lesions associated with masses. The dataset comprised 4254 benign and 3154 malignant lesions. The developed method includes histogram equalization, image cropping and margin augmentation. The GoogLeNet convolutionary neural network was trained to the database to differentiate benign and malignant tumors. The networks were trained on the data with augmentation and the data without augmentation. Both of them showed an area under the curve of over 0.9. The networks showed an accuracy of about 0.9 (90%), a sensitivity of 0.86 and a specificity of 0.96. Although target regions of interest (ROIs) were selected by radiologists, meaning that radiologists still have to point out the location of the ROI, the classification of malignant lesions showed promising results. If this method is used by radiologists in clinical situations it can classify malignant lesions in a short time and support the diagnosis of radiologists in discriminating malignant lesions. Therefore, the proposed method can work in tandem with human radiologists to improve performance, which is a fundamental purpose of computer-aided diagnosis.

Robust Vehicle Detection in Aerial Images Based on Cascaded Convolutional Neural Networks

PubMed Central

Zhong, Jiandan; Lei, Tao; Yao, Guangle

2017-01-01

Vehicle detection in aerial images is an important and challenging task. Traditionally, many target detection models based on sliding-window fashion were developed and achieved acceptable performance, but these models are time-consuming in the detection phase. Recently, with the great success of convolutional neural networks (CNNs) in computer vision, many state-of-the-art detectors have been designed based on deep CNNs. However, these CNN-based detectors are inefficient when applied in aerial image data due to the fact that the existing CNN-based models struggle with small-size object detection and precise localization. To improve the detection accuracy without decreasing speed, we propose a CNN-based detection model combining two independent convolutional neural networks, where the first network is applied to generate a set of vehicle-like regions from multi-feature maps of different hierarchies and scales. Because the multi-feature maps combine the advantage of the deep and shallow convolutional layer, the first network performs well on locating the small targets in aerial image data. Then, the generated candidate regions are fed into the second network for feature extraction and decision making. Comprehensive experiments are conducted on the Vehicle Detection in Aerial Imagery (VEDAI) dataset and Munich vehicle dataset. The proposed cascaded detection model yields high performance, not only in detection accuracy but also in detection speed. PMID:29186756

Residual Shuffling Convolutional Neural Networks for Deep Semantic Image Segmentation Using Multi-Modal Data

NASA Astrophysics Data System (ADS)

Chen, K.; Weinmann, M.; Gao, X.; Yan, M.; Hinz, S.; Jutzi, B.; Weinmann, M.

2018-05-01

In this paper, we address the deep semantic segmentation of aerial imagery based on multi-modal data. Given multi-modal data composed of true orthophotos and the corresponding Digital Surface Models (DSMs), we extract a variety of hand-crafted radiometric and geometric features which are provided separately and in different combinations as input to a modern deep learning framework. The latter is represented by a Residual Shuffling Convolutional Neural Network (RSCNN) combining the characteristics of a Residual Network with the advantages of atrous convolution and a shuffling operator to achieve a dense semantic labeling. Via performance evaluation on a benchmark dataset, we analyze the value of different feature sets for the semantic segmentation task. The derived results reveal that the use of radiometric features yields better classification results than the use of geometric features for the considered dataset. Furthermore, the consideration of data on both modalities leads to an improvement of the classification results. However, the derived results also indicate that the use of all defined features is less favorable than the use of selected features. Consequently, data representations derived via feature extraction and feature selection techniques still provide a gain if used as the basis for deep semantic segmentation.

Adding the missing piece: Spitzer imaging of the HSC-Deep/PFS fields

NASA Astrophysics Data System (ADS)

Sajina, Anna; Bezanson, Rachel; Capak, Peter; Egami, Eiichi; Fan, Xiaohui; Farrah, Duncan; Greene, Jenny; Goulding, Andy; Lacy, Mark; Lin, Yen-Ting; Liu, Xin; Marchesini, Danilo; Moutard, Thibaud; Ono, Yoshiaki; Ouchi, Masami; Sawicki, Marcin; Strauss, Michael; Surace, Jason; Whitaker, Katherine

2018-05-01

We propose to observe a total of 7sq.deg. to complete the Spitzer-IRAC coverage of the HSC-Deep survey fields. These fields are the sites of the PrimeFocusSpectrograph (PFS) galaxy evolution survey which will provide spectra of wide wavelength range and resolution for almost all M* galaxies at z 0.7-1.7, and extend out to z 7 for targeted samples. Our fields already have deep broadband and narrowband photometry in 12 bands spanning from u through K and a wealth of other ancillary data. We propose completing the matching depth IRAC observations in the extended COSMOS, ELAIS-N1 and Deep2-3 fields. By complementing existing Spitzer coverage, this program will lead to an unprecedended in spectro-photometric coverage dataset across a total of 15 sq.deg. This dataset will have significant legacy value as it samples a large enough cosmic volume to be representative of the full range of environments, but also doing so with sufficient information content per galaxy to confidently derive stellar population characteristics. This enables detailed studies of the growth and quenching of galaxies and their supermassive black holes in the context of a galaxy's local and large scale environment.

Observing the Earliest Galaxies: Looking for the Sources of Reionization

NASA Astrophysics Data System (ADS)

Illingworth, Garth

2015-04-01

Systematic searches for the earliest galaxies in the reionization epoch finally became possible in 2009 when the Hubble Space Telescope was updated with a powerful new infrared camera during the final Shuttle servicing mission SM4 to Hubble. The reionization epoch represents the last major phase transition of the universe and was a major event in cosmic history. The intense ultraviolet radiation from young star-forming galaxies is increasingly considered to be the source of the photons that reionized intergalactic hydrogen in the period between the ``dark ages'' (the time before the first stars and galaxies at about 100-200 million years after the Big Bang) and the end of reionization around 800-900 million years. Yet finding and measuring the earliest galaxies in this era of cosmic dawn has proven to a challenging task, even with Hubble's new infrared camera. I will discuss the deep imaging undertaken by Hubble and the remarkable insights that have accrued from the imaging datasets taken over the last decade on the Hubble Ultra-Deep Field (HUDF, HUDF09/12) and other regions. The HUDF datasets are central to the story and have been assembled into the eXtreme Deep Field (XDF), the deepest image ever from Hubble data. The XDF, when combined with results from shallower wide-area imaging surveys (e.g., GOODS, CANDELS) and with detections of galaxies from the Frontier Fields, has provided significant insights into the role of galaxies in reionization. Yet many questions remain. The puzzle is far from being fully solved and, while much will done over the next few years, the solution likely awaits the launch of JWST. NASA/STScI Grant HST-GO-11563.

Dense deconvolution net: Multi path fusion and dense deconvolution for high resolution skin lesion segmentation.

PubMed

He, Xinzi; Yu, Zhen; Wang, Tianfu; Lei, Baiying; Shi, Yiyan

2018-01-01

Dermoscopy imaging has been a routine examination approach for skin lesion diagnosis. Accurate segmentation is the first step for automatic dermoscopy image assessment. The main challenges for skin lesion segmentation are numerous variations in viewpoint and scale of skin lesion region. To handle these challenges, we propose a novel skin lesion segmentation network via a very deep dense deconvolution network based on dermoscopic images. Specifically, the deep dense layer and generic multi-path Deep RefineNet are combined to improve the segmentation performance. The deep representation of all available layers is aggregated to form the global feature maps using skip connection. Also, the dense deconvolution layer is leveraged to capture diverse appearance features via the contextual information. Finally, we apply the dense deconvolution layer to smooth segmentation maps and obtain final high-resolution output. Our proposed method shows the superiority over the state-of-the-art approaches based on the public available 2016 and 2017 skin lesion challenge dataset and achieves the accuracy of 96.0% and 93.9%, which obtained a 6.0% and 1.2% increase over the traditional method, respectively. By utilizing Dense Deconvolution Net, the average time for processing one testing images with our proposed framework was 0.253 s.

Using deep learning for detecting gender in adult chest radiographs

NASA Astrophysics Data System (ADS)

Xue, Zhiyun; Antani, Sameer; Long, L. Rodney; Thoma, George R.

2018-03-01

In this paper, we present a method for automatically identifying the gender of an imaged person using their frontal chest x-ray images. Our work is motivated by the need to determine missing gender information in some datasets. The proposed method employs the technique of convolutional neural network (CNN) based deep learning and transfer learning to overcome the challenge of developing handcrafted features in limited data. Specifically, the method consists of four main steps: pre-processing, CNN feature extractor, feature selection, and classifier. The method is tested on a combined dataset obtained from several sources with varying acquisition quality resulting in different pre-processing steps that are applied for each. For feature extraction, we tested and compared four CNN architectures, viz., AlexNet, VggNet, GoogLeNet, and ResNet. We applied a feature selection technique, since the feature length is larger than the number of images. Two popular classifiers: SVM and Random Forest, are used and compared. We evaluated the classification performance by cross-validation and used seven performance measures. The best performer is the VggNet-16 feature extractor with the SVM classifier, with accuracy of 86.6% and ROC Area being 0.932 for 5-fold cross validation. We also discuss several misclassified cases and describe future work for performance improvement.

WFIRST: Science from the Guest Investigator and Parallel Observation Programs

NASA Astrophysics Data System (ADS)

Postman, Marc; Nataf, David; Furlanetto, Steve; Milam, Stephanie; Robertson, Brant; Williams, Ben; Teplitz, Harry; Moustakas, Leonidas; Geha, Marla; Gilbert, Karoline; Dickinson, Mark; Scolnic, Daniel; Ravindranath, Swara; Strolger, Louis; Peek, Joshua; Marc Postman

2018-01-01

The Wide Field InfraRed Survey Telescope (WFIRST) mission will provide an extremely rich archival dataset that will enable a broad range of scientific investigations beyond the initial objectives of the proposed key survey programs. The scientific impact of WFIRST will thus be significantly expanded by a robust Guest Investigator (GI) archival research program. We will present examples of GI research opportunities ranging from studies of the properties of a variety of Solar System objects, surveys of the outer Milky Way halo, comprehensive studies of cluster galaxies, to unique and new constraints on the epoch of cosmic re-ionization and the assembly of galaxies in the early universe.WFIRST will also support the acquisition of deep wide-field imaging and slitless spectroscopic data obtained in parallel during campaigns with the coronagraphic instrument (CGI). These parallel wide-field imager (WFI) datasets can provide deep imaging data covering several square degrees at no impact to the scheduling of the CGI program. A competitively selected program of well-designed parallel WFI observation programs will, like the GI science above, maximize the overall scientific impact of WFIRST. We will give two examples of parallel observations that could be conducted during a proposed CGI program centered on a dozen nearby stars.

Deep Learning: A Primer for Radiologists.

PubMed

Chartrand, Gabriel; Cheng, Phillip M; Vorontsov, Eugene; Drozdzal, Michal; Turcotte, Simon; Pal, Christopher J; Kadoury, Samuel; Tang, An

2017-01-01

Deep learning is a class of machine learning methods that are gaining success and attracting interest in many domains, including computer vision, speech recognition, natural language processing, and playing games. Deep learning methods produce a mapping from raw inputs to desired outputs (eg, image classes). Unlike traditional machine learning methods, which require hand-engineered feature extraction from inputs, deep learning methods learn these features directly from data. With the advent of large datasets and increased computing power, these methods can produce models with exceptional performance. These models are multilayer artificial neural networks, loosely inspired by biologic neural systems. Weighted connections between nodes (neurons) in the network are iteratively adjusted based on example pairs of inputs and target outputs by back-propagating a corrective error signal through the network. For computer vision tasks, convolutional neural networks (CNNs) have proven to be effective. Recently, several clinical applications of CNNs have been proposed and studied in radiology for classification, detection, and segmentation tasks. This article reviews the key concepts of deep learning for clinical radiologists, discusses technical requirements, describes emerging applications in clinical radiology, and outlines limitations and future directions in this field. Radiologists should become familiar with the principles and potential applications of deep learning in medical imaging. © RSNA, 2017.

Automatic feature learning using multichannel ROI based on deep structured algorithms for computerized lung cancer diagnosis.

PubMed

Sun, Wenqing; Zheng, Bin; Qian, Wei

2017-10-01

This study aimed to analyze the ability of extracting automatically generated features using deep structured algorithms in lung nodule CT image diagnosis, and compare its performance with traditional computer aided diagnosis (CADx) systems using hand-crafted features. All of the 1018 cases were acquired from Lung Image Database Consortium (LIDC) public lung cancer database. The nodules were segmented according to four radiologists' markings, and 13,668 samples were generated by rotating every slice of nodule images. Three multichannel ROI based deep structured algorithms were designed and implemented in this study: convolutional neural network (CNN), deep belief network (DBN), and stacked denoising autoencoder (SDAE). For the comparison purpose, we also implemented a CADx system using hand-crafted features including density features, texture features and morphological features. The performance of every scheme was evaluated by using a 10-fold cross-validation method and an assessment index of the area under the receiver operating characteristic curve (AUC). The observed highest area under the curve (AUC) was 0.899±0.018 achieved by CNN, which was significantly higher than traditional CADx with the AUC=0.848±0.026. The results from DBN was also slightly higher than CADx, while SDAE was slightly lower. By visualizing the automatic generated features, we found some meaningful detectors like curvy stroke detectors from deep structured schemes. The study results showed the deep structured algorithms with automatically generated features can achieve desirable performance in lung nodule diagnosis. With well-tuned parameters and large enough dataset, the deep learning algorithms can have better performance than current popular CADx. We believe the deep learning algorithms with similar data preprocessing procedure can be used in other medical image analysis areas as well. Copyright © 2017. Published by Elsevier Ltd.

Wide-angle Marine Seismic Refraction Imaging of Vertical Faults: Pre-Stack Turning Wave Migrations of Synthetic Data and Implications for Survey Design

NASA Astrophysics Data System (ADS)

Miller, N. C.; Lizarralde, D.; McGuire, J.; Hole, J. A.

2006-12-01

We consider methodologies, including survey design and processing algorithms, which are best suited to imaging vertical reflectors in oceanic crust using marine seismic techniques. The ability to image the reflectivity structure of transform faults as a function of depth, for example, may provide new insights into what controls seismicity along these plate boundaries. Turning-wave migration has been used with success to image vertical faults on land. With synthetic datasets we find that this approach has unique difficulties in the deep ocean. The fault-reflected crustal refraction phase (Pg-r) typically used in pre-stack migrations is difficult to isolate in marine seismic data. An "imagable" Pg-r is only observed in a time window between the first arrivals and arrivals from the sediments and the thick, slow water layer at offsets beyond ~25 km. Ocean- bottom seismometers (OBSs), as opposed to a long surface streamer, must be used to acquire data suitable for crustal-scale vertical imaging. The critical distance for Moho reflections (PmP) in oceanic crust is also ~25 km, thus Pg-r and PmP-r are observed with very little separation, and the fault-reflected mantle refraction (Pn-r) arrives prior to Pg-r as the observation window opens with increased OBS-to-fault distance. This situation presents difficulties for "first-arrival" based Kirchoff migration approaches and suggests that wave- equation approaches, which in theory can image all three phases simultaneously, may be more suitable for vertical imaging in oceanic crust. We will present a comparison of these approaches as applied to a synthetic dataset generated from realistic, stochastic velocity models. We will assess their suitability, the migration artifacts unique to the deep ocean, and the ideal instrument layout for such an experiment.

a Novel Deep Convolutional Neural Network for Spectral-Spatial Classification of Hyperspectral Data

NASA Astrophysics Data System (ADS)

Li, N.; Wang, C.; Zhao, H.; Gong, X.; Wang, D.

2018-04-01

Spatial and spectral information are obtained simultaneously by hyperspectral remote sensing. Joint extraction of these information of hyperspectral image is one of most import methods for hyperspectral image classification. In this paper, a novel deep convolutional neural network (CNN) is proposed, which extracts spectral-spatial information of hyperspectral images correctly. The proposed model not only learns sufficient knowledge from the limited number of samples, but also has powerful generalization ability. The proposed framework based on three-dimensional convolution can extract spectral-spatial features of labeled samples effectively. Though CNN has shown its robustness to distortion, it cannot extract features of different scales through the traditional pooling layer that only have one size of pooling window. Hence, spatial pyramid pooling (SPP) is introduced into three-dimensional local convolutional filters for hyperspectral classification. Experimental results with a widely used hyperspectral remote sensing dataset show that the proposed model provides competitive performance.

MassImager: A software for interactive and in-depth analysis of mass spectrometry imaging data.

PubMed

He, Jiuming; Huang, Luojiao; Tian, Runtao; Li, Tiegang; Sun, Chenglong; Song, Xiaowei; Lv, Yiwei; Luo, Zhigang; Li, Xin; Abliz, Zeper

2018-07-26

Mass spectrometry imaging (MSI) has become a powerful tool to probe molecule events in biological tissue. However, it is a widely held viewpoint that one of the biggest challenges is an easy-to-use data processing software for discovering the underlying biological information from complicated and huge MSI dataset. Here, a user-friendly and full-featured MSI software including three subsystems, Solution, Visualization and Intelligence, named MassImager, is developed focusing on interactive visualization, in-situ biomarker discovery and artificial intelligent pathological diagnosis. Simplified data preprocessing and high-throughput MSI data exchange, serialization jointly guarantee the quick reconstruction of ion image and rapid analysis of dozens of gigabytes datasets. It also offers diverse self-defined operations for visual processing, including multiple ion visualization, multiple channel superposition, image normalization, visual resolution enhancement and image filter. Regions-of-interest analysis can be performed precisely through the interactive visualization between the ion images and mass spectra, also the overlaid optical image guide, to directly find out the region-specific biomarkers. Moreover, automatic pattern recognition can be achieved immediately upon the supervised or unsupervised multivariate statistical modeling. Clear discrimination between cancer tissue and adjacent tissue within a MSI dataset can be seen in the generated pattern image, which shows great potential in visually in-situ biomarker discovery and artificial intelligent pathological diagnosis of cancer. All the features are integrated together in MassImager to provide a deep MSI processing solution at the in-situ metabolomics level for biomarker discovery and future clinical pathological diagnosis. Copyright © 2018 The Authors. Published by Elsevier B.V. All rights reserved.

Deep neural network using color and synthesized three-dimensional shape for face recognition

NASA Astrophysics Data System (ADS)

Rhee, Seon-Min; Yoo, ByungIn; Han, Jae-Joon; Hwang, Wonjun

2017-03-01

We present an approach for face recognition using synthesized three-dimensional (3-D) shape information together with two-dimensional (2-D) color in a deep convolutional neural network (DCNN). As 3-D facial shape is hardly affected by the extrinsic 2-D texture changes caused by illumination, make-up, and occlusions, it could provide more reliable complementary features in harmony with the 2-D color feature in face recognition. Unlike other approaches that use 3-D shape information with the help of an additional depth sensor, our approach generates a personalized 3-D face model by using only face landmarks in the 2-D input image. Using the personalized 3-D face model, we generate a frontalized 2-D color facial image as well as 3-D facial images (e.g., a depth image and a normal image). In our DCNN, we first feed 2-D and 3-D facial images into independent convolutional layers, where the low-level kernels are successfully learned according to their own characteristics. Then, we merge them and feed into higher-level layers under a single deep neural network. Our proposed approach is evaluated with labeled faces in the wild dataset and the results show that the error rate of the verification rate at false acceptance rate 1% is improved by up to 32.1% compared with the baseline where only a 2-D color image is used.

Identifying images of handwritten digits using deep learning in H2O

NASA Astrophysics Data System (ADS)

Sadhasivam, Jayakumar; Charanya, R.; Kumar, S. Harish; Srinivasan, A.

2017-11-01

Automatic digit recognition is of popular interest today. Deep learning techniques make it possible for object recognition in image data. Perceiving the digit has turned into a fundamental part as far as certifiable applications. Since, digits are composed in various styles in this way to distinguish the digit it is important to perceive and arrange it with the assistance of machine learning methods. This exploration depends on supervised learning vector quantization neural system arranged under counterfeit artificial neural network. The pictures of digits are perceived, prepared and tried. After the system is made digits are prepared utilizing preparing dataset vectors and testing is connected to the pictures of digits which are separated to each other by fragmenting the picture and resizing the digit picture as needs be for better precision.

A DBN based anomaly targets detector for HSI

NASA Astrophysics Data System (ADS)

Ma, Ning; Wang, Shaojun; Yu, Jinxiang; Peng, Yu

2017-10-01

Due to the assumption that Hyperspectral image (HSI) should conform to Gaussian distribution, traditional Mahalanobis distance-based anomaly targets detectors perform poor because the assumption may not always hold. In order to solve those problems, a deep learning based detector, Deep Belief Network(DBN) anomaly detector(DBN-AD), was proposed to fit the unknown distribution of HSI by energy modeling, the reconstruction errors of this encode-decode processing are used for discriminating the anomaly targets. Experiments are implemented on real and synthesized HSI dataset which collection by Airborne Visible Infra-Red Imaging Spectrometer (AVIRIS). Comparing to classic anomaly detector, the proposed method shows better performance, it performs about 0.17 higher in Area Under ROC Curve (AUC) than that of Reed-Xiaoli detector(RXD) and Kernel-RXD (K-RXD).

Identifying Medical Diagnoses and Treatable Diseases by Image-Based Deep Learning.

PubMed

Kermany, Daniel S; Goldbaum, Michael; Cai, Wenjia; Valentim, Carolina C S; Liang, Huiying; Baxter, Sally L; McKeown, Alex; Yang, Ge; Wu, Xiaokang; Yan, Fangbing; Dong, Justin; Prasadha, Made K; Pei, Jacqueline; Ting, Magdalene Y L; Zhu, Jie; Li, Christina; Hewett, Sierra; Dong, Jason; Ziyar, Ian; Shi, Alexander; Zhang, Runze; Zheng, Lianghong; Hou, Rui; Shi, William; Fu, Xin; Duan, Yaou; Huu, Viet A N; Wen, Cindy; Zhang, Edward D; Zhang, Charlotte L; Li, Oulan; Wang, Xiaobo; Singer, Michael A; Sun, Xiaodong; Xu, Jie; Tafreshi, Ali; Lewis, M Anthony; Xia, Huimin; Zhang, Kang

2018-02-22

The implementation of clinical-decision support algorithms for medical imaging faces challenges with reliability and interpretability. Here, we establish a diagnostic tool based on a deep-learning framework for the screening of patients with common treatable blinding retinal diseases. Our framework utilizes transfer learning, which trains a neural network with a fraction of the data of conventional approaches. Applying this approach to a dataset of optical coherence tomography images, we demonstrate performance comparable to that of human experts in classifying age-related macular degeneration and diabetic macular edema. We also provide a more transparent and interpretable diagnosis by highlighting the regions recognized by the neural network. We further demonstrate the general applicability of our AI system for diagnosis of pediatric pneumonia using chest X-ray images. This tool may ultimately aid in expediting the diagnosis and referral of these treatable conditions, thereby facilitating earlier treatment, resulting in improved clinical outcomes. VIDEO ABSTRACT. Copyright © 2018 Elsevier Inc. All rights reserved.

Super-resolution reconstruction of MR image with a novel residual learning network algorithm

NASA Astrophysics Data System (ADS)

Shi, Jun; Liu, Qingping; Wang, Chaofeng; Zhang, Qi; Ying, Shihui; Xu, Haoyu

2018-04-01

Spatial resolution is one of the key parameters of magnetic resonance imaging (MRI). The image super-resolution (SR) technique offers an alternative approach to improve the spatial resolution of MRI due to its simplicity. Convolutional neural networks (CNN)-based SR algorithms have achieved state-of-the-art performance, in which the global residual learning (GRL) strategy is now commonly used due to its effectiveness for learning image details for SR. However, the partial loss of image details usually happens in a very deep network due to the degradation problem. In this work, we propose a novel residual learning-based SR algorithm for MRI, which combines both multi-scale GRL and shallow network block-based local residual learning (LRL). The proposed LRL module works effectively in capturing high-frequency details by learning local residuals. One simulated MRI dataset and two real MRI datasets have been used to evaluate our algorithm. The experimental results show that the proposed SR algorithm achieves superior performance to all of the other compared CNN-based SR algorithms in this work.

Bilinear Convolutional Neural Networks for Fine-grained Visual Recognition.

PubMed

Lin, Tsung-Yu; RoyChowdhury, Aruni; Maji, Subhransu

2017-07-04

We present a simple and effective architecture for fine-grained recognition called Bilinear Convolutional Neural Networks (B-CNNs). These networks represent an image as a pooled outer product of features derived from two CNNs and capture localized feature interactions in a translationally invariant manner. B-CNNs are related to orderless texture representations built on deep features but can be trained in an end-to-end manner. Our most accurate model obtains 84.1%, 79.4%, 84.5% and 91.3% per-image accuracy on the Caltech-UCSD birds [66], NABirds [63], FGVC aircraft [42], and Stanford cars [33] dataset respectively and runs at 30 frames-per-second on a NVIDIA Titan X GPU. We then present a systematic analysis of these networks and show that (1) the bilinear features are highly redundant and can be reduced by an order of magnitude in size without significant loss in accuracy, (2) are also effective for other image classification tasks such as texture and scene recognition, and (3) can be trained from scratch on the ImageNet dataset offering consistent improvements over the baseline architecture. Finally, we present visualizations of these models on various datasets using top activations of neural units and gradient-based inversion techniques. The source code for the complete system is available at http://vis-www.cs.umass.edu/bcnn.

Rock images classification by using deep convolution neural network

NASA Astrophysics Data System (ADS)

Cheng, Guojian; Guo, Wenhui

2017-08-01

Granularity analysis is one of the most essential issues in authenticate under microscope. To improve the efficiency and accuracy of traditional manual work, an convolutional neural network based method is proposed for granularity analysis from thin section image, which chooses and extracts features from image samples while build classifier to recognize granularity of input image samples. 4800 samples from Ordos basin are used for experiments under colour spaces of HSV, YCbCr and RGB respectively. On the test dataset, the correct rate in RGB colour space is 98.5%, and it is believable in HSV and YCbCr colour space. The results show that the convolution neural network can classify the rock images with high reliability.

Single-shot T2 mapping using overlapping-echo detachment planar imaging and a deep convolutional neural network.

PubMed

Cai, Congbo; Wang, Chao; Zeng, Yiqing; Cai, Shuhui; Liang, Dong; Wu, Yawen; Chen, Zhong; Ding, Xinghao; Zhong, Jianhui

2018-04-24

An end-to-end deep convolutional neural network (CNN) based on deep residual network (ResNet) was proposed to efficiently reconstruct reliable T 2 mapping from single-shot overlapping-echo detachment (OLED) planar imaging. The training dataset was obtained from simulations that were carried out on SPROM (Simulation with PRoduct Operator Matrix) software developed by our group. The relationship between the original OLED image containing two echo signals and the corresponding T 2 mapping was learned by ResNet training. After the ResNet was trained, it was applied to reconstruct the T 2 mapping from simulation and in vivo human brain data. Although the ResNet was trained entirely on simulated data, the trained network was generalized well to real human brain data. The results from simulation and in vivo human brain experiments show that the proposed method significantly outperforms the echo-detachment-based method. Reliable T 2 mapping with higher accuracy is achieved within 30 ms after the network has been trained, while the echo-detachment-based OLED reconstruction method took approximately 2 min. The proposed method will facilitate real-time dynamic and quantitative MR imaging via OLED sequence, and deep convolutional neural network has the potential to reconstruct maps from complex MRI sequences efficiently. © 2018 International Society for Magnetic Resonance in Medicine.

Single-Image Super Resolution for Multispectral Remote Sensing Data Using Convolutional Neural Networks

NASA Astrophysics Data System (ADS)

Liebel, L.; Körner, M.

2016-06-01

In optical remote sensing, spatial resolution of images is crucial for numerous applications. Space-borne systems are most likely to be affected by a lack of spatial resolution, due to their natural disadvantage of a large distance between the sensor and the sensed object. Thus, methods for single-image super resolution are desirable to exceed the limits of the sensor. Apart from assisting visual inspection of datasets, post-processing operations—e.g., segmentation or feature extraction—can benefit from detailed and distinguishable structures. In this paper, we show that recently introduced state-of-the-art approaches for single-image super resolution of conventional photographs, making use of deep learning techniques, such as convolutional neural networks (CNN), can successfully be applied to remote sensing data. With a huge amount of training data available, end-to-end learning is reasonably easy to apply and can achieve results unattainable using conventional handcrafted algorithms. We trained our CNN on a specifically designed, domain-specific dataset, in order to take into account the special characteristics of multispectral remote sensing data. This dataset consists of publicly available SENTINEL-2 images featuring 13 spectral bands, a ground resolution of up to 10m, and a high radiometric resolution and thus satisfying our requirements in terms of quality and quantity. In experiments, we obtained results superior compared to competing approaches trained on generic image sets, which failed to reasonably scale satellite images with a high radiometric resolution, as well as conventional interpolation methods.

Fully Convolutional Networks for Ground Classification from LIDAR Point Clouds

NASA Astrophysics Data System (ADS)

Rizaldy, A.; Persello, C.; Gevaert, C. M.; Oude Elberink, S. J.

2018-05-01

Deep Learning has been massively used for image classification in recent years. The use of deep learning for ground classification from LIDAR point clouds has also been recently studied. However, point clouds need to be converted into an image in order to use Convolutional Neural Networks (CNNs). In state-of-the-art techniques, this conversion is slow because each point is converted into a separate image. This approach leads to highly redundant computation during conversion and classification. The goal of this study is to design a more efficient data conversion and ground classification. This goal is achieved by first converting the whole point cloud into a single image. The classification is then performed by a Fully Convolutional Network (FCN), a modified version of CNN designed for pixel-wise image classification. The proposed method is significantly faster than state-of-the-art techniques. On the ISPRS Filter Test dataset, it is 78 times faster for conversion and 16 times faster for classification. Our experimental analysis on the same dataset shows that the proposed method results in 5.22 % of total error, 4.10 % of type I error, and 15.07 % of type II error. Compared to the previous CNN-based technique and LAStools software, the proposed method reduces the total error and type I error (while type II error is slightly higher). The method was also tested on a very high point density LIDAR point clouds resulting in 4.02 % of total error, 2.15 % of type I error and 6.14 % of type II error.

Through the Looking GLASS: A JWST Exploration of Galaxy Formation and Evolution from Cosmic Dawn to Present Day

NASA Astrophysics Data System (ADS)

Treu, Tommaso; Abramson, L.; Bradac, M.; Brammer, G.; Fontana, A.; Henry, A.; Hoag, A.; Huang, K.; Mason, C.; Morishita, T.; Pentericci, L.; Wang, X.

2017-11-01

We propose a carefully designed set of observations of the lensing cluster Abell 2744 to study intrinsically faint magnified galaxies from the epoch of reionization to redshift of 1, demonstrating and characterizing complementary spectroscopic modes with NIRSPEC and NIRISS. The observations are designed to address the questions: 1) when did reionization happen and what were the sources of reionizing photons? 2) How do baryons cycle in and out of galaxies? This dataset with deep spectroscopy on the cluster and deep multiband NIRCAM imaging in parallel will enable a wealth of investigations and will thus be of interest to a broad section of the astronomical community. The dataset will illustrate the power and challenges of: 1) combining rest frame UV and optical NIRSPEC spectroscopy for galaxies at the epoch of reionization, 2) obtaining spatially resolved emission line maps with NIRISS, 3) combining NIRISS and NIRSPEC spectroscopy. Building on our extensive experience with HST slitless spectroscopy and imaging in clusters of galaxies as part of the GLASS, WISP, SURFSUP, and ASTRODEEP projects, we will provide the following science-enabling products to the community: 1)quantitative comparison of spatially resolved (NIRISS) and spectrally resolved (NIRSPEC) spectroscopy, 2) Object based interactive exploration tools for multi-instrument datasets, 3) Interface for easy forced extractionof slitless spectra based on coordinates, 4) UV-optical spectroscopic templates of highredshift galaxies, 5) NIRCAM parallel catalogs and a list of 26 z>=9 dropouts for spectroscopic follow-up in Cycle-2.

AggNet: Deep Learning From Crowds for Mitosis Detection in Breast Cancer Histology Images.

PubMed

Albarqouni, Shadi; Baur, Christoph; Achilles, Felix; Belagiannis, Vasileios; Demirci, Stefanie; Navab, Nassir

2016-05-01

The lack of publicly available ground-truth data has been identified as the major challenge for transferring recent developments in deep learning to the biomedical imaging domain. Though crowdsourcing has enabled annotation of large scale databases for real world images, its application for biomedical purposes requires a deeper understanding and hence, more precise definition of the actual annotation task. The fact that expert tasks are being outsourced to non-expert users may lead to noisy annotations introducing disagreement between users. Despite being a valuable resource for learning annotation models from crowdsourcing, conventional machine-learning methods may have difficulties dealing with noisy annotations during training. In this manuscript, we present a new concept for learning from crowds that handle data aggregation directly as part of the learning process of the convolutional neural network (CNN) via additional crowdsourcing layer (AggNet). Besides, we present an experimental study on learning from crowds designed to answer the following questions. 1) Can deep CNN be trained with data collected from crowdsourcing? 2) How to adapt the CNN to train on multiple types of annotation datasets (ground truth and crowd-based)? 3) How does the choice of annotation and aggregation affect the accuracy? Our experimental setup involved Annot8, a self-implemented web-platform based on Crowdflower API realizing image annotation tasks for a publicly available biomedical image database. Our results give valuable insights into the functionality of deep CNN learning from crowd annotations and prove the necessity of data aggregation integration.

The impact of privacy protection filters on gender recognition

NASA Astrophysics Data System (ADS)

Ruchaud, Natacha; Antipov, Grigory; Korshunov, Pavel; Dugelay, Jean-Luc; Ebrahimi, Touradj; Berrani, Sid-Ahmed

2015-09-01

Deep learning-based algorithms have become increasingly efficient in recognition and detection tasks, especially when they are trained on large-scale datasets. Such recent success has led to a speculation that deep learning methods are comparable to or even outperform human visual system in its ability to detect and recognize objects and their features. In this paper, we focus on the specific task of gender recognition in images when they have been processed by privacy protection filters (e.g., blurring, masking, and pixelization) applied at different strengths. Assuming a privacy protection scenario, we compare the performance of state of the art deep learning algorithms with a subjective evaluation obtained via crowdsourcing to understand how privacy protection filters affect both machine and human vision.

Saliency detection by conditional generative adversarial network

NASA Astrophysics Data System (ADS)

Cai, Xiaoxu; Yu, Hui

2018-04-01

Detecting salient objects in images has been a fundamental problem in computer vision. In recent years, deep learning has shown its impressive performance in dealing with many kinds of vision tasks. In this paper, we propose a new method to detect salient objects by using Conditional Generative Adversarial Network (GAN). This type of network not only learns the mapping from RGB images to salient regions, but also learns a loss function for training the mapping. To the best of our knowledge, this is the first time that Conditional GAN has been used in salient object detection. We evaluate our saliency detection method on 2 large publicly available datasets with pixel accurate annotations. The experimental results have shown the significant and consistent improvements over the state-of-the-art method on a challenging dataset, and the testing speed is much faster.

Image-Based Multi-Target Tracking through Multi-Bernoulli Filtering with Interactive Likelihoods.

PubMed

Hoak, Anthony; Medeiros, Henry; Povinelli, Richard J

2017-03-03

We develop an interactive likelihood (ILH) for sequential Monte Carlo (SMC) methods for image-based multiple target tracking applications. The purpose of the ILH is to improve tracking accuracy by reducing the need for data association. In addition, we integrate a recently developed deep neural network for pedestrian detection along with the ILH with a multi-Bernoulli filter. We evaluate the performance of the multi-Bernoulli filter with the ILH and the pedestrian detector in a number of publicly available datasets (2003 PETS INMOVE, Australian Rules Football League (AFL) and TUD-Stadtmitte) using standard, well-known multi-target tracking metrics (optimal sub-pattern assignment (OSPA) and classification of events, activities and relationships for multi-object trackers (CLEAR MOT)). In all datasets, the ILH term increases the tracking accuracy of the multi-Bernoulli filter.

Image-Based Multi-Target Tracking through Multi-Bernoulli Filtering with Interactive Likelihoods

PubMed Central

Hoak, Anthony; Medeiros, Henry; Povinelli, Richard J.

2017-01-01

We develop an interactive likelihood (ILH) for sequential Monte Carlo (SMC) methods for image-based multiple target tracking applications. The purpose of the ILH is to improve tracking accuracy by reducing the need for data association. In addition, we integrate a recently developed deep neural network for pedestrian detection along with the ILH with a multi-Bernoulli filter. We evaluate the performance of the multi-Bernoulli filter with the ILH and the pedestrian detector in a number of publicly available datasets (2003 PETS INMOVE, Australian Rules Football League (AFL) and TUD-Stadtmitte) using standard, well-known multi-target tracking metrics (optimal sub-pattern assignment (OSPA) and classification of events, activities and relationships for multi-object trackers (CLEAR MOT)). In all datasets, the ILH term increases the tracking accuracy of the multi-Bernoulli filter. PMID:28273796

Accurate Natural Trail Detection Using a Combination of a Deep Neural Network and Dynamic Programming.

PubMed

Adhikari, Shyam Prasad; Yang, Changju; Slot, Krzysztof; Kim, Hyongsuk

2018-01-10

This paper presents a vision sensor-based solution to the challenging problem of detecting and following trails in highly unstructured natural environments like forests, rural areas and mountains, using a combination of a deep neural network and dynamic programming. The deep neural network (DNN) concept has recently emerged as a very effective tool for processing vision sensor signals. A patch-based DNN is trained with supervised data to classify fixed-size image patches into "trail" and "non-trail" categories, and reshaped to a fully convolutional architecture to produce trail segmentation map for arbitrary-sized input images. As trail and non-trail patches do not exhibit clearly defined shapes or forms, the patch-based classifier is prone to misclassification, and produces sub-optimal trail segmentation maps. Dynamic programming is introduced to find an optimal trail on the sub-optimal DNN output map. Experimental results showing accurate trail detection for real-world trail datasets captured with a head mounted vision system are presented.

Improving resolution of MR images with an adversarial network incorporating images with different contrast.

PubMed

Kim, Ki Hwan; Do, Won-Joon; Park, Sung-Hong

2018-05-04

The routine MRI scan protocol consists of multiple pulse sequences that acquire images of varying contrast. Since high frequency contents such as edges are not significantly affected by image contrast, down-sampled images in one contrast may be improved by high resolution (HR) images acquired in another contrast, reducing the total scan time. In this study, we propose a new deep learning framework that uses HR MR images in one contrast to generate HR MR images from highly down-sampled MR images in another contrast. The proposed convolutional neural network (CNN) framework consists of two CNNs: (a) a reconstruction CNN for generating HR images from the down-sampled images using HR images acquired with a different MRI sequence and (b) a discriminator CNN for improving the perceptual quality of the generated HR images. The proposed method was evaluated using a public brain tumor database and in vivo datasets. The performance of the proposed method was assessed in tumor and no-tumor cases separately, with perceptual image quality being judged by a radiologist. To overcome the challenge of training the network with a small number of available in vivo datasets, the network was pretrained using the public database and then fine-tuned using the small number of in vivo datasets. The performance of the proposed method was also compared to that of several compressed sensing (CS) algorithms. Incorporating HR images of another contrast improved the quantitative assessments of the generated HR image in reference to ground truth. Also, incorporating a discriminator CNN yielded perceptually higher image quality. These results were verified in regions of normal tissue as well as tumors for various MRI sequences from pseudo k-space data generated from the public database. The combination of pretraining with the public database and fine-tuning with the small number of real k-space datasets enhanced the performance of CNNs in in vivo application compared to training CNNs from scratch. The proposed method outperformed the compressed sensing methods. The proposed method can be a good strategy for accelerating routine MRI scanning. © 2018 American Association of Physicists in Medicine.

Automatic Gleason grading of H and E stained microscopic prostate images using deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Gummeson, Anna; Arvidsson, Ida; Ohlsson, Mattias; Overgaard, Niels C.; Krzyzanowska, Agnieszka; Heyden, Anders; Bjartell, Anders; Aström, Kalle

2017-03-01

Prostate cancer is the most diagnosed cancer in men. The diagnosis is confirmed by pathologists based on ocular inspection of prostate biopsies in order to classify them according to Gleason score. The main goal of this paper is to automate the classification using convolutional neural networks (CNNs). The introduction of CNNs has broadened the field of pattern recognition. It replaces the classical way of designing and extracting hand-made features used for classification with the substantially different strategy of letting the computer itself decide which features are of importance. For automated prostate cancer classification into the classes: Benign, Gleason grade 3, 4 and 5 we propose a CNN with small convolutional filters that has been trained from scratch using stochastic gradient descent with momentum. The input consists of microscopic images of haematoxylin and eosin stained tissue, the output is a coarse segmentation into regions of the four different classes. The dataset used consists of 213 images, each considered to be of one class only. Using four-fold cross-validation we obtained an error rate of 7.3%, which is significantly better than previous state of the art using the same dataset. Although the dataset was rather small, good results were obtained. From this we conclude that CNN is a promising method for this problem. Future work includes obtaining a larger dataset, which potentially could diminish the error margin.

The Hubble Deep UV Legacy Survey (HDUV)

NASA Astrophysics Data System (ADS)

Montes, Mireia; Oesch, Pascal

2015-08-01

Deep HST imaging has shown that the overall star formation density and UV light density at z>3 is dominated by faint, blue galaxies. Remarkably, very little is known about the equivalent galaxy population at lower redshifts. Understanding how these galaxies evolve across the epoch of peak cosmic star-formation is key to a complete picture of galaxy evolution. Here, we present a new HST WFC3/UVIS program, the Hubble Deep UV (HDUV) legacy survey. The HDUV is a 132 orbit program to obtain deep imaging in two filters (F275W and F336W) over the two CANDELS Deep fields. We will cover ~100 arcmin2 sampling the rest-frame far-UV at z>~0.5, this will provide a unique legacy dataset with exquisite HST multi-wavelength imaging as well as ancillary HST grism NIR spectroscopy for a detailed study of faint, star-forming galaxies at z~0.5-2. The HDUV will enable a wealth of research by the community, which includes tracing the evolution of the FUV luminosity function over the peak of the star formation rate density from z~3 down to z~0.5, measuring the physical properties of sub-L* galaxies, and characterizing resolved stellar populations to decipher the build-up of the Hubble sequence from sub-galactic clumps. This poster provides an overview of the HDUV survey and presents the reduced data products and catalogs which will be released to the community, reaching down to 27.5-28.0 mag at 5 sigma. By directly sampling the rest-frame far-UV at z>~0.5, this will provide a unique legacy dataset with exquisite HST multi-wavelength imaging as well as ancillary HST grism NIR spectroscopy for a detailed study of faint, star-forming galaxies at z~0.5-2. The HDUV will enable a wealth of research by the community, which includes tracing the evolution of the FUV luminosity function over the peak of the star formation rate density from z~3 down to z~0.5, measuring the physical properties of sub-L* galaxies, and characterizing resolved stellar populations to decipher the build-up of the Hubble sequence from sub-galactic clumps. This poster provides an overview of the HDUV survey and presents reduced data products and catalogs which will be released to the community.

S-CNN: Subcategory-aware convolutional networks for object detection.

PubMed

Chen, Tao; Lu, Shijian; Fan, Jiayuan

2017-09-26

The marriage between the deep convolutional neural network (CNN) and region proposals has made breakthroughs for object detection in recent years. While the discriminative object features are learned via a deep CNN for classification, the large intra-class variation and deformation still limit the performance of the CNN based object detection. We propose a subcategory-aware CNN (S-CNN) to solve the object intra-class variation problem. In the proposed technique, the training samples are first grouped into multiple subcategories automatically through a novel instance sharing maximum margin clustering process. A multi-component Aggregated Channel Feature (ACF) detector is then trained to produce more latent training samples, where each ACF component corresponds to one clustered subcategory. The produced latent samples together with their subcategory labels are further fed into a CNN classifier to filter out false proposals for object detection. An iterative learning algorithm is designed for the joint optimization of image subcategorization, multi-component ACF detector, and subcategory-aware CNN classifier. Experiments on INRIA Person dataset, Pascal VOC 2007 dataset and MS COCO dataset show that the proposed technique clearly outperforms the state-of-the-art methods for generic object detection.

Discovery radiomics via evolutionary deep radiomic sequencer discovery for pathologically proven lung cancer detection.

PubMed

Shafiee, Mohammad Javad; Chung, Audrey G; Khalvati, Farzad; Haider, Masoom A; Wong, Alexander

2017-10-01

While lung cancer is the second most diagnosed form of cancer in men and women, a sufficiently early diagnosis can be pivotal in patient survival rates. Imaging-based, or radiomics-driven, detection methods have been developed to aid diagnosticians, but largely rely on hand-crafted features that may not fully encapsulate the differences between cancerous and healthy tissue. Recently, the concept of discovery radiomics was introduced, where custom abstract features are discovered from readily available imaging data. We propose an evolutionary deep radiomic sequencer discovery approach based on evolutionary deep intelligence. Motivated by patient privacy concerns and the idea of operational artificial intelligence, the evolutionary deep radiomic sequencer discovery approach organically evolves increasingly more efficient deep radiomic sequencers that produce significantly more compact yet similarly descriptive radiomic sequences over multiple generations. As a result, this framework improves operational efficiency and enables diagnosis to be run locally at the radiologist's computer while maintaining detection accuracy. We evaluated the evolved deep radiomic sequencer (EDRS) discovered via the proposed evolutionary deep radiomic sequencer discovery framework against state-of-the-art radiomics-driven and discovery radiomics methods using clinical lung CT data with pathologically proven diagnostic data from the LIDC-IDRI dataset. The EDRS shows improved sensitivity (93.42%), specificity (82.39%), and diagnostic accuracy (88.78%) relative to previous radiomics approaches.

Anomaly detection for medical images based on a one-class classification

NASA Astrophysics Data System (ADS)

Wei, Qi; Ren, Yinhao; Hou, Rui; Shi, Bibo; Lo, Joseph Y.; Carin, Lawrence

2018-02-01

Detecting an anomaly such as a malignant tumor or a nodule from medical images including mammogram, CT or PET images is still an ongoing research problem drawing a lot of attention with applications in medical diagnosis. A conventional way to address this is to learn a discriminative model using training datasets of negative and positive samples. The learned model can be used to classify a testing sample into a positive or negative class. However, in medical applications, the high unbalance between negative and positive samples poses a difficulty for learning algorithms, as they will be biased towards the majority group, i.e., the negative one. To address this imbalanced data issue as well as leverage the huge amount of negative samples, i.e., normal medical images, we propose to learn an unsupervised model to characterize the negative class. To make the learned model more flexible and extendable for medical images of different scales, we have designed an autoencoder based on a deep neural network to characterize the negative patches decomposed from large medical images. A testing image is decomposed into patches and then fed into the learned autoencoder to reconstruct these patches themselves. The reconstruction error of one patch is used to classify this patch into a binary class, i.e., a positive or a negative one, leading to a one-class classifier. The positive patches highlight the suspicious areas containing anomalies in a large medical image. The proposed method has been tested on InBreast dataset and achieves an AUC of 0.84. The main contribution of our work can be summarized as follows. 1) The proposed one-class learning requires only data from one class, i.e., the negative data; 2) The patch-based learning makes the proposed method scalable to images of different sizes and helps avoid the large scale problem for medical images; 3) The training of the proposed deep convolutional neural network (DCNN) based auto-encoder is fast and stable.

MJO Signals in Latent Heating: Results from TRMM Retrievals

NASA Technical Reports Server (NTRS)

Zhang, Chidong; Ling, Jian; Hagos, Samson; Tao, Wei-Kuo; Lang, Steve; Takayabu, Yukari N.; Shige, Shoichi; Katsumata, Masaki; Olson, William S.; L'Ecuyer, Tristan

2010-01-01

The Madden-Julian Oscillation (MJO) is the dominant intraseasonal signal in the global tropical atmosphere. Almost all numerical climate models have difficulty to simulate realistic MJO. Four TRMM datasets of latent heating were diagnosed for signals in the MJO. In all four datasets, vertical structures of latent heating are dominated by two components, one deep with its peak above the melting level and one shallow with its peak below. Profiles of the two components are nearly ubiquitous in longitude, allowing a separation of the vertical and zonal/temporal variations when the latitudinal dependence is not considered. All four datasets exhibit robust MJO spectral signals in the deep component as eastward propagating spectral peaks centered at period of 50 days and zonal wavenumber 1, well distinguished from lower- and higher-frequency power and much stronger than the corresponding westward power. The shallow component shows similar but slightly less robust MJO spectral peaks. MJO signals were further extracted from a combination of band-pass (30 - 90 day) filtered deep and shallow components. Largest amplitudes of both deep and shallow components of the MJO are confined to the Indian and western Pacific Oceans. There is a local minimum in the deep components over the Maritime Continent. The shallow components of the MJO differ substantially among the four TRMM datasets in their detailed zonal distributions in the eastern hemisphere. In composites of the heating evolution through the life cycle of the MJO, the shallow components lead the deep ones in some datasets and at certain longitudes. In many respects, the four TRMM datasets agree well in their deep components, but not in their shallow components and the phase relations between the deep and shallow components. These results indicate that caution must be exercised in applications of these latent heating data.

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs.

PubMed

Chen, Liang-Chieh; Papandreou, George; Kokkinos, Iasonas; Murphy, Kevin; Yuille, Alan L

2018-04-01

In this work we address the task of semantic image segmentation with Deep Learning and make three main contributions that are experimentally shown to have substantial practical merit. First, we highlight convolution with upsampled filters, or 'atrous convolution', as a powerful tool in dense prediction tasks. Atrous convolution allows us to explicitly control the resolution at which feature responses are computed within Deep Convolutional Neural Networks. It also allows us to effectively enlarge the field of view of filters to incorporate larger context without increasing the number of parameters or the amount of computation. Second, we propose atrous spatial pyramid pooling (ASPP) to robustly segment objects at multiple scales. ASPP probes an incoming convolutional feature layer with filters at multiple sampling rates and effective fields-of-views, thus capturing objects as well as image context at multiple scales. Third, we improve the localization of object boundaries by combining methods from DCNNs and probabilistic graphical models. The commonly deployed combination of max-pooling and downsampling in DCNNs achieves invariance but has a toll on localization accuracy. We overcome this by combining the responses at the final DCNN layer with a fully connected Conditional Random Field (CRF), which is shown both qualitatively and quantitatively to improve localization performance. Our proposed "DeepLab" system sets the new state-of-art at the PASCAL VOC-2012 semantic image segmentation task, reaching 79.7 percent mIOU in the test set, and advances the results on three other datasets: PASCAL-Context, PASCAL-Person-Part, and Cityscapes. All of our code is made publicly available online.

The Chandra Deepest Fields in the Infrared: Making the Connection between Normal Galaxies and AGN

NASA Astrophysics Data System (ADS)

Grogin, N. A.; Ferguson, H. C.; Dickinson, M. E.; Giavalisco, M.; Mobasher, B.; Padovani, P.; Williams, R. E.; Chary, R.; Gilli, R.; Heckman, T. M.; Stern, D.; Winge, C.

2001-12-01

Within each of the two Chandra Deepest Fields (CDFs), there are ~10'x15' regions targeted for non-proprietary, deep SIRTF 3.6--24μ m imaging as part of the Great Observatories Origins Deep Survey (GOODS) Legacy program. In advance of the SIRTF observations, the GOODS team has recently begun obtaining non-proprietary, deep ground-based optical and near-IR imaging and spectroscopy over these regions, which contain virtually all of the current ≈1 Msec CXO coverage in the CDF North and much of the ≈1 Msec coverage in the CDF South. In particular, the planned depth of the near-IR imaging (JAB ~ 25.3; HAB ~ 24.8; KAB ~ 24.4) combined with the deep Chandra data can allow us to trace the evolutionary connection between normal galaxies, starbursts, and AGN out to z ~ 1 and beyond. We describe our CDF Archival program, which is integrating these GOODS-supporting observations together with the CDF archival data and other publicly-available datasets in these regions to create a multi-wavelength deep imaging and spectroscpic database available to the entire community. We highlight progress toward near-term science goals of this program, including: (a) pushing constraints on the redshift distribution and spectral-energy distributions of the faintest X-ray sources to the deepest possible levels via photometric redshifts; and (b) better characterizing the heavily-obscured and the high-redshift populations via both a near-IR search for optically-undetected CDF X-ray sources and also X-ray stacking analyses on the CXO-undetected EROs in these fields.

A Model-Based Approach for Microvasculature Structure Distortion Correction in Two-Photon Fluorescence Microscopy Images

PubMed Central

Dao, Lam; Glancy, Brian; Lucotte, Bertrand; Chang, Lin-Ching; Balaban, Robert S; Hsu, Li-Yueh

2015-01-01

SUMMARY This paper investigates a post-processing approach to correct spatial distortion in two-photon fluorescence microscopy images for vascular network reconstruction. It is aimed at in vivo imaging of large field-of-view, deep-tissue studies of vascular structures. Based on simple geometric modeling of the object-of-interest, a distortion function is directly estimated from the image volume by deconvolution analysis. Such distortion function is then applied to sub volumes of the image stack to adaptively adjust for spatially varying distortion and reduce the image blurring through blind deconvolution. The proposed technique was first evaluated in phantom imaging of fluorescent microspheres that are comparable in size to the underlying capillary vascular structures. The effectiveness of restoring three-dimensional spherical geometry of the microspheres using the estimated distortion function was compared with empirically measured point-spread function. Next, the proposed approach was applied to in vivo vascular imaging of mouse skeletal muscle to reduce the image distortion of the capillary structures. We show that the proposed method effectively improve the image quality and reduce spatially varying distortion that occurs in large field-of-view deep-tissue vascular dataset. The proposed method will help in qualitative interpretation and quantitative analysis of vascular structures from fluorescence microscopy images. PMID:26224257

Rotation invariant deep binary hashing for fast image retrieval

NASA Astrophysics Data System (ADS)

Dai, Lai; Liu, Jianming; Jiang, Aiwen

2017-07-01

In this paper, we study how to compactly represent image's characteristics for fast image retrieval. We propose supervised rotation invariant compact discriminative binary descriptors through combining convolutional neural network with hashing. In the proposed network, binary codes are learned by employing a hidden layer for representing latent concepts that dominate on class labels. A loss function is proposed to minimize the difference between binary descriptors that describe reference image and the rotated one. Compared with some other supervised methods, the proposed network doesn't have to require pair-wised inputs for binary code learning. Experimental results show that our method is effective and achieves state-of-the-art results on the CIFAR-10 and MNIST datasets.

A Physics-Based Deep Learning Approach to Shadow Invariant Representations of Hyperspectral Images.

PubMed

Windrim, Lloyd; Ramakrishnan, Rishi; Melkumyan, Arman; Murphy, Richard J

2018-02-01

This paper proposes the Relit Spectral Angle-Stacked Autoencoder, a novel unsupervised feature learning approach for mapping pixel reflectances to illumination invariant encodings. This work extends the Spectral Angle-Stacked Autoencoder so that it can learn a shadow-invariant mapping. The method is inspired by a deep learning technique, Denoising Autoencoders, with the incorporation of a physics-based model for illumination such that the algorithm learns a shadow invariant mapping without the need for any labelled training data, additional sensors, a priori knowledge of the scene or the assumption of Planckian illumination. The method is evaluated using datasets captured from several different cameras, with experiments to demonstrate the illumination invariance of the features and how they can be used practically to improve the performance of high-level perception algorithms that operate on images acquired outdoors.

Automatic detection of kidney in 3D pediatric ultrasound images using deep neural networks

NASA Astrophysics Data System (ADS)

Tabrizi, Pooneh R.; Mansoor, Awais; Biggs, Elijah; Jago, James; Linguraru, Marius George

2018-02-01

Ultrasound (US) imaging is the routine and safe diagnostic modality for detecting pediatric urology problems, such as hydronephrosis in the kidney. Hydronephrosis is the swelling of one or both kidneys because of the build-up of urine. Early detection of hydronephrosis can lead to a substantial improvement in kidney health outcomes. Generally, US imaging is a challenging modality for the evaluation of pediatric kidneys with different shape, size, and texture characteristics. The aim of this study is to present an automatic detection method to help kidney analysis in pediatric 3DUS images. The method localizes the kidney based on its minimum volume oriented bounding box) using deep neural networks. Separate deep neural networks are trained to estimate the kidney position, orientation, and scale, making the method computationally efficient by avoiding full parameter training. The performance of the method was evaluated using a dataset of 45 kidneys (18 normal and 27 diseased kidneys diagnosed with hydronephrosis) through the leave-one-out cross validation method. Quantitative results show the proposed detection method could extract the kidney position, orientation, and scale ratio with root mean square values of 1.3 +/- 0.9 mm, 6.34 +/- 4.32 degrees, and 1.73 +/- 0.04, respectively. This method could be helpful in automating kidney segmentation for routine clinical evaluation.

Adaptive Neuron Apoptosis for Accelerating Deep Learning on Large Scale Systems

DOE Office of Scientific and Technical Information (OSTI.GOV)

Siegel, Charles M.; Daily, Jeffrey A.; Vishnu, Abhinav

Machine Learning and Data Mining (MLDM) algorithms are becoming ubiquitous in {\\em model learning} from the large volume of data generated using simulations, experiments and handheld devices. Deep Learning algorithms -- a class of MLDM algorithms -- are applied for automatic feature extraction, and learning non-linear models for unsupervised and supervised algorithms. Naturally, several libraries which support large scale Deep Learning -- such as TensorFlow and Caffe -- have become popular. In this paper, we present novel techniques to accelerate the convergence of Deep Learning algorithms by conducting low overhead removal of redundant neurons -- {\\em apoptosis} of neurons --more » which do not contribute to model learning, during the training phase itself. We provide in-depth theoretical underpinnings of our heuristics (bounding accuracy loss and handling apoptosis of several neuron types), and present the methods to conduct adaptive neuron apoptosis. We implement our proposed heuristics with the recently introduced TensorFlow and using its recently proposed extension with MPI. Our performance evaluation on two difference clusters -- one connected with Intel Haswell multi-core systems, and other with nVIDIA GPUs -- using InfiniBand, indicates the efficacy of the proposed heuristics and implementations. Specifically, we are able to improve the training time for several datasets by 2-3x, while reducing the number of parameters by 30x (4-5x on average) on datasets such as ImageNet classification. For the Higgs Boson dataset, our implementation improves the accuracy (measured by Area Under Curve (AUC)) for classification from 0.88/1 to 0.94/1, while reducing the number of parameters by 3x in comparison to existing literature, while achieving a 2.44x speedup in comparison to the default (no apoptosis) algorithm.« less

Deep multi-spectral ensemble learning for electronic cleansing in dual-energy CT colonography

NASA Astrophysics Data System (ADS)

Tachibana, Rie; Näppi, Janne J.; Hironaka, Toru; Kim, Se Hyung; Yoshida, Hiroyuki

2017-03-01

We developed a novel electronic cleansing (EC) method for dual-energy CT colonography (DE-CTC) based on an ensemble deep convolution neural network (DCNN) and multi-spectral multi-slice image patches. In the method, an ensemble DCNN is used to classify each voxel of a DE-CTC image volume into five classes: luminal air, soft tissue, tagged fecal materials, and partial-volume boundaries between air and tagging and those between soft tissue and tagging. Each DCNN acts as a voxel classifier, where an input image patch centered at the voxel is generated as input to the DCNNs. An image patch has three channels that are mapped from a region-of-interest containing the image plane of the voxel and the two adjacent image planes. Six different types of spectral input image datasets were derived using two dual-energy CT images, two virtual monochromatic images, and two material images. An ensemble DCNN was constructed by use of a meta-classifier that combines the output of multiple DCNNs, each of which was trained with a different type of multi-spectral image patches. The electronically cleansed CTC images were calculated by removal of regions classified as other than soft tissue, followed by a colon surface reconstruction. For pilot evaluation, 359 volumes of interest (VOIs) representing sources of subtraction artifacts observed in current EC schemes were sampled from 30 clinical CTC cases. Preliminary results showed that the ensemble DCNN can yield high accuracy in labeling of the VOIs, indicating that deep learning of multi-spectral EC with multi-slice imaging could accurately remove residual fecal materials from CTC images without generating major EC artifacts.

Improving protein disorder prediction by deep bidirectional long short-term memory recurrent neural networks.

PubMed

Hanson, Jack; Yang, Yuedong; Paliwal, Kuldip; Zhou, Yaoqi

2017-03-01

Capturing long-range interactions between structural but not sequence neighbors of proteins is a long-standing challenging problem in bioinformatics. Recently, long short-term memory (LSTM) networks have significantly improved the accuracy of speech and image classification problems by remembering useful past information in long sequential events. Here, we have implemented deep bidirectional LSTM recurrent neural networks in the problem of protein intrinsic disorder prediction. The new method, named SPOT-Disorder, has steadily improved over a similar method using a traditional, window-based neural network (SPINE-D) in all datasets tested without separate training on short and long disordered regions. Independent tests on four other datasets including the datasets from critical assessment of structure prediction (CASP) techniques and >10 000 annotated proteins from MobiDB, confirmed SPOT-Disorder as one of the best methods in disorder prediction. Moreover, initial studies indicate that the method is more accurate in predicting functional sites in disordered regions. These results highlight the usefulness combining LSTM with deep bidirectional recurrent neural networks in capturing non-local, long-range interactions for bioinformatics applications. SPOT-disorder is available as a web server and as a standalone program at: http://sparks-lab.org/server/SPOT-disorder/index.php . j.hanson@griffith.edu.au or yuedong.yang@griffith.edu.au or yaoqi.zhou@griffith.edu.au. Supplementary data is available at Bioinformatics online. © The Author 2016. Published by Oxford University Press. All rights reserved. For Permissions, please e-mail: journals.permissions@oup.com

Candidate-penetrative-fracture mapping of the Grand Canyon area, Arizona, from spatial correlation of deep geophysical features and surficial lineaments

USGS Publications Warehouse

Gettings, Mark E.; Bultman, Mark W.

2005-01-01

Some aquifers of the southwestern Colorado Plateaus Province are deeply buried and overlain by several impermeable shale layers, and so recharge to the aquifer probably is mainly by seepage down penetrative-fracture systems. The purpose of this 2-year study, sponsored by the U.S. National Park Service, was to map candidate deep penetrative fractures over a 120,000-km2 area, using gravity and aeromagnetic-anomaly data together with surficial-fracture data. The study area was on the Colorado Plateau south of the Grand Canyon and west of Black Mesa; mapping was carried out at a scale of 1:250,000. The resulting database constitutes a spatially registered estimate of deep-fracture locations. Candidate penetrative fractures were located by spatial correlation of horizontal- gradient and analytic-signal maximums of gravity and magnetic anomalies with major surficial lineaments obtained from geologic, topographic, side-looking-airborne-radar, and satellite imagery. The maps define a subset of candidate penetrative fractures because of limitations in the data coverage and the analytical technique. In particular, the data and analytical technique used cannot predict whether the fractures are open or closed. Correlations were carried out by using image-processing software, such that every pixel on the resulting images was coded to uniquely identify which datasets are correlated. The technique correctly identified known and many new deep fracture systems. The resulting penetrative-fracture-distribution maps constitute an objectively obtained, repeatable dataset and a benchmark from which additional studies can begin. The maps also define in detail the tectonic fabrics of the southwestern Colorado Plateaus Province. Overlaying the correlated lineaments on the normalized-density-of-vegetation-index image reveals that many of these lineaments correlate with the boundaries of vegetation zones in drainages and canyons and so may be controlling near-surface water availability in some places. Many derivative products can be produced from the database, such as fracture-density-estimate maps, and maps with the number of correlations color-coded to estimate the possible quality of correlation. The database contained in this report is designed to be used in a geographic information system and image-processing systems, and most data layers are in georeferenced tagged image format (Geotiff) or ARC grids. The report includes 163 map plates and various metadata, supporting, and statistical diagram files.

Bag-of-visual-phrases and hierarchical deep models for traffic sign detection and recognition in mobile laser scanning data

NASA Astrophysics Data System (ADS)

Yu, Yongtao; Li, Jonathan; Wen, Chenglu; Guan, Haiyan; Luo, Huan; Wang, Cheng

2016-03-01

This paper presents a novel algorithm for detection and recognition of traffic signs in mobile laser scanning (MLS) data for intelligent transportation-related applications. The traffic sign detection task is accomplished based on 3-D point clouds by using bag-of-visual-phrases representations; whereas the recognition task is achieved based on 2-D images by using a Gaussian-Bernoulli deep Boltzmann machine-based hierarchical classifier. To exploit high-order feature encodings of feature regions, a deep Boltzmann machine-based feature encoder is constructed. For detecting traffic signs in 3-D point clouds, the proposed algorithm achieves an average recall, precision, quality, and F-score of 0.956, 0.946, 0.907, and 0.951, respectively, on the four selected MLS datasets. For on-image traffic sign recognition, a recognition accuracy of 97.54% is achieved by using the proposed hierarchical classifier. Comparative studies with the existing traffic sign detection and recognition methods demonstrate that our algorithm obtains promising, reliable, and high performance in both detecting traffic signs in 3-D point clouds and recognizing traffic signs on 2-D images.

Using transfer learning to detect galaxy mergers

NASA Astrophysics Data System (ADS)

Ackermann, Sandro; Schawinksi, Kevin; Zhang, Ce; Weigel, Anna K.; Turp, M. Dennis

2018-05-01

We investigate the use of deep convolutional neural networks (deep CNNs) for automatic visual detection of galaxy mergers. Moreover, we investigate the use of transfer learning in conjunction with CNNs, by retraining networks first trained on pictures of everyday objects. We test the hypothesis that transfer learning is useful for improving classification performance for small training sets. This would make transfer learning useful for finding rare objects in astronomical imaging datasets. We find that these deep learning methods perform significantly better than current state-of-the-art merger detection methods based on nonparametric systems like CAS and GM20. Our method is end-to-end and robust to image noise and distortions; it can be applied directly without image preprocessing. We also find that transfer learning can act as a regulariser in some cases, leading to better overall classification accuracy (p = 0.02). Transfer learning on our full training set leads to a lowered error rate from 0.0381 down to 0.0321, a relative improvement of 15%. Finally, we perform a basic sanity-check by creating a merger sample with our method, and comparing with an already existing, manually created merger catalogue in terms of colour-mass distribution and stellar mass function.

An adaptive deep Q-learning strategy for handwritten digit recognition.

PubMed

Qiao, Junfei; Wang, Gongming; Li, Wenjing; Chen, Min

2018-02-22

Handwritten digits recognition is a challenging problem in recent years. Although many deep learning-based classification algorithms are studied for handwritten digits recognition, the recognition accuracy and running time still need to be further improved. In this paper, an adaptive deep Q-learning strategy is proposed to improve accuracy and shorten running time for handwritten digit recognition. The adaptive deep Q-learning strategy combines the feature-extracting capability of deep learning and the decision-making of reinforcement learning to form an adaptive Q-learning deep belief network (Q-ADBN). First, Q-ADBN extracts the features of original images using an adaptive deep auto-encoder (ADAE), and the extracted features are considered as the current states of Q-learning algorithm. Second, Q-ADBN receives Q-function (reward signal) during recognition of the current states, and the final handwritten digits recognition is implemented by maximizing the Q-function using Q-learning algorithm. Finally, experimental results from the well-known MNIST dataset show that the proposed Q-ADBN has a superiority to other similar methods in terms of accuracy and running time. Copyright © 2018 Elsevier Ltd. All rights reserved.

An Efficient Implementation of Deep Convolutional Neural Networks for MRI Segmentation.

PubMed

Hoseini, Farnaz; Shahbahrami, Asadollah; Bayat, Peyman

2018-02-27

Image segmentation is one of the most common steps in digital image processing, classifying a digital image into different segments. The main goal of this paper is to segment brain tumors in magnetic resonance images (MRI) using deep learning. Tumors having different shapes, sizes, brightness and textures can appear anywhere in the brain. These complexities are the reasons to choose a high-capacity Deep Convolutional Neural Network (DCNN) containing more than one layer. The proposed DCNN contains two parts: architecture and learning algorithms. The architecture and the learning algorithms are used to design a network model and to optimize parameters for the network training phase, respectively. The architecture contains five convolutional layers, all using 3 × 3 kernels, and one fully connected layer. Due to the advantage of using small kernels with fold, it allows making the effect of larger kernels with smaller number of parameters and fewer computations. Using the Dice Similarity Coefficient metric, we report accuracy results on the BRATS 2016, brain tumor segmentation challenge dataset, for the complete, core, and enhancing regions as 0.90, 0.85, and 0.84 respectively. The learning algorithm includes the task-level parallelism. All the pixels of an MR image are classified using a patch-based approach for segmentation. We attain a good performance and the experimental results show that the proposed DCNN increases the segmentation accuracy compared to previous techniques.

Neural Network for Nanoscience Scanning Electron Microscope Image Recognition.

PubMed

Modarres, Mohammad Hadi; Aversa, Rossella; Cozzini, Stefano; Ciancio, Regina; Leto, Angelo; Brandino, Giuseppe Piero

2017-10-16

In this paper we applied transfer learning techniques for image recognition, automatic categorization, and labeling of nanoscience images obtained by scanning electron microscope (SEM). Roughly 20,000 SEM images were manually classified into 10 categories to form a labeled training set, which can be used as a reference set for future applications of deep learning enhanced algorithms in the nanoscience domain. The categories chosen spanned the range of 0-Dimensional (0D) objects such as particles, 1D nanowires and fibres, 2D films and coated surfaces, and 3D patterned surfaces such as pillars. The training set was used to retrain on the SEM dataset and to compare many convolutional neural network models (Inception-v3, Inception-v4, ResNet). We obtained compatible results by performing a feature extraction of the different models on the same dataset. We performed additional analysis of the classifier on a second test set to further investigate the results both on particular cases and from a statistical point of view. Our algorithm was able to successfully classify around 90% of a test dataset consisting of SEM images, while reduced accuracy was found in the case of images at the boundary between two categories or containing elements of multiple categories. In these cases, the image classification did not identify a predominant category with a high score. We used the statistical outcomes from testing to deploy a semi-automatic workflow able to classify and label images generated by the SEM. Finally, a separate training was performed to determine the volume fraction of coherently aligned nanowires in SEM images. The results were compared with what was obtained using the Local Gradient Orientation method. This example demonstrates the versatility and the potential of transfer learning to address specific tasks of interest in nanoscience applications.

Do deep convolutional neural networks really need to be deep when applied for remote scene classification?

NASA Astrophysics Data System (ADS)

Luo, Chang; Wang, Jie; Feng, Gang; Xu, Suhui; Wang, Shiqiang

2017-10-01

Deep convolutional neural networks (CNNs) have been widely used to obtain high-level representation in various computer vision tasks. However, for remote scene classification, there are not sufficient images to train a very deep CNN from scratch. From two viewpoints of generalization power, we propose two promising kinds of deep CNNs for remote scenes and try to find whether deep CNNs need to be deep for remote scene classification. First, we transfer successful pretrained deep CNNs to remote scenes based on the theory that depth of CNNs brings the generalization power by learning available hypothesis for finite data samples. Second, according to the opposite viewpoint that generalization power of deep CNNs comes from massive memorization and shallow CNNs with enough neural nodes have perfect finite sample expressivity, we design a lightweight deep CNN (LDCNN) for remote scene classification. With five well-known pretrained deep CNNs, experimental results on two independent remote-sensing datasets demonstrate that transferred deep CNNs can achieve state-of-the-art results in an unsupervised setting. However, because of its shallow architecture, LDCNN cannot obtain satisfactory performance, regardless of whether in an unsupervised, semisupervised, or supervised setting. CNNs really need depth to obtain general features for remote scenes. This paper also provides baseline for applying deep CNNs to other remote sensing tasks.

Skin Lesion Analysis towards Melanoma Detection Using Deep Learning Network.

PubMed

Li, Yuexiang; Shen, Linlin

2018-02-11

Skin lesions are a severe disease globally. Early detection of melanoma in dermoscopy images significantly increases the survival rate. However, the accurate recognition of melanoma is extremely challenging due to the following reasons: low contrast between lesions and skin, visual similarity between melanoma and non-melanoma lesions, etc. Hence, reliable automatic detection of skin tumors is very useful to increase the accuracy and efficiency of pathologists. In this paper, we proposed two deep learning methods to address three main tasks emerging in the area of skin lesion image processing, i.e., lesion segmentation (task 1), lesion dermoscopic feature extraction (task 2) and lesion classification (task 3). A deep learning framework consisting of two fully convolutional residual networks (FCRN) is proposed to simultaneously produce the segmentation result and the coarse classification result. A lesion index calculation unit (LICU) is developed to refine the coarse classification results by calculating the distance heat-map. A straight-forward CNN is proposed for the dermoscopic feature extraction task. The proposed deep learning frameworks were evaluated on the ISIC 2017 dataset. Experimental results show the promising accuracies of our frameworks, i.e., 0.753 for task 1, 0.848 for task 2 and 0.912 for task 3 were achieved.

Fusion of Deep Learning and Compressed Domain features for Content Based Image Retrieval.

PubMed

Liu, Peizhong; Guo, Jing-Ming; Wu, Chi-Yi; Cai, Danlin

2017-08-29

This paper presents an effective image retrieval method by combining high-level features from Convolutional Neural Network (CNN) model and low-level features from Dot-Diffused Block Truncation Coding (DDBTC). The low-level features, e.g., texture and color, are constructed by VQ-indexed histogram from DDBTC bitmap, maximum, and minimum quantizers. Conversely, high-level features from CNN can effectively capture human perception. With the fusion of the DDBTC and CNN features, the extended deep learning two-layer codebook features (DL-TLCF) is generated using the proposed two-layer codebook, dimension reduction, and similarity reweighting to improve the overall retrieval rate. Two metrics, average precision rate (APR) and average recall rate (ARR), are employed to examine various datasets. As documented in the experimental results, the proposed schemes can achieve superior performance compared to the state-of-the-art methods with either low- or high-level features in terms of the retrieval rate. Thus, it can be a strong candidate for various image retrieval related applications.

Fine-tuning convolutional deep features for MRI based brain tumor classification

NASA Astrophysics Data System (ADS)

Ahmed, Kaoutar B.; Hall, Lawrence O.; Goldgof, Dmitry B.; Liu, Renhao; Gatenby, Robert A.

2017-03-01

Prediction of survival time from brain tumor magnetic resonance images (MRI) is not commonly performed and would ordinarily be a time consuming process. However, current cross-sectional imaging techniques, particularly MRI, can be used to generate many features that may provide information on the patient's prognosis, including survival. This information can potentially be used to identify individuals who would benefit from more aggressive therapy. Rather than using pre-defined and hand-engineered features as with current radiomics methods, we investigated the use of deep features extracted from pre-trained convolutional neural networks (CNNs) in predicting survival time. We also provide evidence for the power of domain specific fine-tuning in improving the performance of a pre-trained CNN's, even though our data set is small. We fine-tuned a CNN initially trained on a large natural image recognition dataset (Imagenet ILSVRC) and transferred the learned feature representations to the survival time prediction task, obtaining over 81% accuracy in a leave one out cross validation.

A deep learning approach for the analysis of masses in mammograms with minimal user intervention.

PubMed

Dhungel, Neeraj; Carneiro, Gustavo; Bradley, Andrew P

2017-04-01

We present an integrated methodology for detecting, segmenting and classifying breast masses from mammograms with minimal user intervention. This is a long standing problem due to low signal-to-noise ratio in the visualisation of breast masses, combined with their large variability in terms of shape, size, appearance and location. We break the problem down into three stages: mass detection, mass segmentation, and mass classification. For the detection, we propose a cascade of deep learning methods to select hypotheses that are refined based on Bayesian optimisation. For the segmentation, we propose the use of deep structured output learning that is subsequently refined by a level set method. Finally, for the classification, we propose the use of a deep learning classifier, which is pre-trained with a regression to hand-crafted feature values and fine-tuned based on the annotations of the breast mass classification dataset. We test our proposed system on the publicly available INbreast dataset and compare the results with the current state-of-the-art methodologies. This evaluation shows that our system detects 90% of masses at 1 false positive per image, has a segmentation accuracy of around 0.85 (Dice index) on the correctly detected masses, and overall classifies masses as malignant or benign with sensitivity (Se) of 0.98 and specificity (Sp) of 0.7. Copyright © 2017 Elsevier B.V. All rights reserved.

Deep learning and three-compartment breast imaging in breast cancer diagnosis

NASA Astrophysics Data System (ADS)

Drukker, Karen; Huynh, Benjamin Q.; Giger, Maryellen L.; Malkov, Serghei; Avila, Jesus I.; Fan, Bo; Joe, Bonnie; Kerlikowske, Karla; Drukteinis, Jennifer S.; Kazemi, Leila; Pereira, Malesa M.; Shepherd, John

2017-03-01

We investigated whether deep learning has potential to aid in the diagnosis of breast cancer when applied to mammograms and biologic tissue composition images derived from three-compartment (3CB) imaging. The dataset contained diagnostic mammograms and 3CB images (water, lipid, and protein content) of biopsy-sampled BIRADS 4 and 5 lesions in 195 patients. In 58 patients, the lesion manifested as a mass (13 malignant vs. 45 benign), in 87 as microcalcifications (19 vs. 68), and in 56 as (focal) asymmetry or architectural distortion (11 vs. 45). Six patients had both a mass and calcifications. For each mammogram and corresponding 3CB images, a 128x128 region of interest containing the lesion was selected by an expert radiologist and used directly as input to a deep learning method pretrained on a very large independent set of non-medical images. We used a nested leave-one-out-by-case (patient) model selection and classification protocol. The area under the ROC curve (AUC) for the task of distinguishing between benign and malignant lesions was used as performance metric. For the cases with mammographic masses, the AUC increased from 0.83 (mammograms alone) to 0.89 (mammograms+3CB, p=.162). For the microcalcification and asymmetry/architectural distortion cases the AUC increased from 0.84 to 0.91 (p=.116) and from 0.61 to 0.87 (p=.006), respectively. Our results indicate great potential for the application of deep learning methods in the diagnosis of breast cancer and additional knowledge of the biologic tissue composition appeared to improve performance, especially for lesions mammographically manifesting as asymmetries or architectural distortions.

Evaluation of the impact of deep learning architectural components selection and dataset size on a medical imaging task

NASA Astrophysics Data System (ADS)

Dutta, Sandeep; Gros, Eric

2018-03-01

Deep Learning (DL) has been successfully applied in numerous fields fueled by increasing computational power and access to data. However, for medical imaging tasks, limited training set size is a common challenge when applying DL. This paper explores the applicability of DL to the task of classifying a single axial slice from a CT exam into one of six anatomy regions. A total of 29000 images selected from 223 CT exams were manually labeled for ground truth. An additional 54 exams were labeled and used as an independent test set. The network architecture developed for this application is composed of 6 convolutional layers and 2 fully connected layers with RELU non-linear activations between each layer. Max-pooling was used after every second convolutional layer, and a softmax layer was used at the end. Given this base architecture, the effect of inclusion of network architecture components such as Dropout and Batch Normalization on network performance and training is explored. The network performance as a function of training and validation set size is characterized by training each network architecture variation using 5,10,20,40,50 and 100% of the available training data. The performance comparison of the various network architectures was done for anatomy classification as well as two computer vision datasets. The anatomy classifier accuracy varied from 74.1% to 92.3% in this study depending on the training size and network layout used. Dropout layers improved the model accuracy for all training sizes.

Phytoplankton off the West Coast of Africa

NASA Technical Reports Server (NTRS)

2002-01-01

Just off the coast of West Africa, persistent northeasterly trade winds often churn up deep ocean water. When the nutrients in these deep waters reach the ocean's surface, they often give rise to large blooms of phytoplankton. This image of the Mauritanian coast shows swirls of phytoplankton fed by the upwelling of nutrient-rich water. The scene was acquired by the Medium Resolution Imaging Spectrometer (MERIS) aboard the European Space Agency's ENVISAT. MERIS will monitor changes in phytoplankton across Earth's oceans and seas, both for the purpose of managing fisheries and conducting global change research. NASA scientists will use data from this European instrument in the Sensor Intercomparison and Merger for Biological and Interdisciplinary Oceanic Studies (SIMBIOS) program. The mission of SIMBIOS is to construct a consistent long-term dataset of ocean color (phytoplankton abundance) measurements made by multiple satellite instruments, including the Sea-viewing Wide Field-of-view Sensor (SeaWiFS) and the Moderate-Resolution Imaging Spectroradiometer (MODIS). For more information about MERIS and ENVISAT, visit the ENVISAT home page. Image copyright European Space Agency

Classification of C2C12 cells at differentiation by convolutional neural network of deep learning using phase contrast images.

PubMed

Niioka, Hirohiko; Asatani, Satoshi; Yoshimura, Aina; Ohigashi, Hironori; Tagawa, Seiichi; Miyake, Jun

2018-01-01

In the field of regenerative medicine, tremendous numbers of cells are necessary for tissue/organ regeneration. Today automatic cell-culturing system has been developed. The next step is constructing a non-invasive method to monitor the conditions of cells automatically. As an image analysis method, convolutional neural network (CNN), one of the deep learning method, is approaching human recognition level. We constructed and applied the CNN algorithm for automatic cellular differentiation recognition of myogenic C2C12 cell line. Phase-contrast images of cultured C2C12 are prepared as input dataset. In differentiation process from myoblasts to myotubes, cellular morphology changes from round shape to elongated tubular shape due to fusion of the cells. CNN abstract the features of the shape of the cells and classify the cells depending on the culturing days from when differentiation is induced. Changes in cellular shape depending on the number of days of culture (Day 0, Day 3, Day 6) are classified with 91.3% accuracy. Image analysis with CNN has a potential to realize regenerative medicine industry.

The 3D Reference Earth Model: Status and Preliminary Results

NASA Astrophysics Data System (ADS)

Moulik, P.; Lekic, V.; Romanowicz, B. A.

2017-12-01

In the 20th century, seismologists constructed models of how average physical properties (e.g. density, rigidity, compressibility, anisotropy) vary with depth in the Earth's interior. These one-dimensional (1D) reference Earth models (e.g. PREM) have proven indispensable in earthquake location, imaging of interior structure, understanding material properties under extreme conditions, and as a reference in other fields, such as particle physics and astronomy. Over the past three decades, new datasets motivated more sophisticated efforts that yielded models of how properties vary both laterally and with depth in the Earth's interior. Though these three-dimensional (3D) models exhibit compelling similarities at large scales, differences in the methodology, representation of structure, and dataset upon which they are based, have prevented the creation of 3D community reference models. As part of the REM-3D project, we are compiling and reconciling reference seismic datasets of body wave travel-time measurements, fundamental mode and overtone surface wave dispersion measurements, and normal mode frequencies and splitting functions. These reference datasets are being inverted for a long-wavelength, 3D reference Earth model that describes the robust long-wavelength features of mantle heterogeneity. As a community reference model with fully quantified uncertainties and tradeoffs and an associated publically available dataset, REM-3D will facilitate Earth imaging studies, earthquake characterization, inferences on temperature and composition in the deep interior, and be of improved utility to emerging scientific endeavors, such as neutrino geoscience. Here, we summarize progress made in the construction of the reference long period dataset and present a preliminary version of REM-3D in the upper-mantle. In order to determine the level of detail warranted for inclusion in REM-3D, we analyze the spectrum of discrepancies between models inverted with different subsets of the reference dataset. This procedure allows us to evaluate the extent of consistency in imaging heterogeneity at various depths and between spatial scales.

Quicksilver: Fast predictive image registration - A deep learning approach.

PubMed

Yang, Xiao; Kwitt, Roland; Styner, Martin; Niethammer, Marc

2017-09-01

This paper introduces Quicksilver, a fast deformable image registration method. Quicksilver registration for image-pairs works by patch-wise prediction of a deformation model based directly on image appearance. A deep encoder-decoder network is used as the prediction model. While the prediction strategy is general, we focus on predictions for the Large Deformation Diffeomorphic Metric Mapping (LDDMM) model. Specifically, we predict the momentum-parameterization of LDDMM, which facilitates a patch-wise prediction strategy while maintaining the theoretical properties of LDDMM, such as guaranteed diffeomorphic mappings for sufficiently strong regularization. We also provide a probabilistic version of our prediction network which can be sampled during the testing time to calculate uncertainties in the predicted deformations. Finally, we introduce a new correction network which greatly increases the prediction accuracy of an already existing prediction network. We show experimental results for uni-modal atlas-to-image as well as uni-/multi-modal image-to-image registrations. These experiments demonstrate that our method accurately predicts registrations obtained by numerical optimization, is very fast, achieves state-of-the-art registration results on four standard validation datasets, and can jointly learn an image similarity measure. Quicksilver is freely available as an open-source software. Copyright © 2017 Elsevier Inc. All rights reserved.

Accurate CT-MR image registration for deep brain stimulation: a multi-observer evaluation study

NASA Astrophysics Data System (ADS)

Rühaak, Jan; Derksen, Alexander; Heldmann, Stefan; Hallmann, Marc; Meine, Hans

2015-03-01

Since the first clinical interventions in the late 1980s, Deep Brain Stimulation (DBS) of the subthalamic nucleus has evolved into a very effective treatment option for patients with severe Parkinson's disease. DBS entails the implantation of an electrode that performs high frequency stimulations to a target area deep inside the brain. A very accurate placement of the electrode is a prerequisite for positive therapy outcome. The assessment of the intervention result is of central importance in DBS treatment and involves the registration of pre- and postinterventional scans. In this paper, we present an image processing pipeline for highly accurate registration of postoperative CT to preoperative MR. Our method consists of two steps: a fully automatic pre-alignment using a detection of the skull tip in the CT based on fuzzy connectedness, and an intensity-based rigid registration. The registration uses the Normalized Gradient Fields distance measure in a multilevel Gauss-Newton optimization framework and focuses on a region around the subthalamic nucleus in the MR. The accuracy of our method was extensively evaluated on 20 DBS datasets from clinical routine and compared with manual expert registrations. For each dataset, three independent registrations were available, thus allowing to relate algorithmic with expert performance. Our method achieved an average registration error of 0.95mm in the target region around the subthalamic nucleus as compared to an inter-observer variability of 1.12 mm. Together with the short registration time of about five seconds on average, our method forms a very attractive package that can be considered ready for clinical use.

Acoustic Reverse Time Migration of the Cascadia Subduction Zone Dataset

NASA Astrophysics Data System (ADS)

Jia, L.; Mallick, S.

2017-12-01

Reverse time migration (RTM) is a wave-equation based migration method, which provides more accurate images than ray-based migration methods, especially for the structures in deep areas, making it an effective tool for imaging the subduction plate boundary. In this work, we extend the work of Fortin (2015) and applied acoustic finite-element RTM on the Cascadia Subduction Zone (CSZ) dataset. The dataset was acquired by Cascadia Open-Access Seismic Transects (COAST) program, targeting the megathrust in the central Cascadia subduction zone (Figure 1). The data on a 2D seismic reflection line that crosses the Juan de Fuca/North American subduction boundary off Washington (Line 5) were pre-processed and worked through Kirchhoff prestack depth migration (PSDM). Figure 2 compares the depth image of Line 5 of the CSZ data using Kirchhoff PSDM (top) and RTM (bottom). In both images, the subducting plate is indicated with yellow arrows. Notice that the RTM image is much superior to the PSDM image by several aspects. First, the plate boundary appears to be much more continuous in the RTM image than the PSDM image. Second, the RTM image indicates the subducting plate is relatively smooth on the seaward (west) side between 0-50 km. Within the deformation front of the accretionary prism (50-80 km), the RTM image shows substantial roughness in the subducting plate. These features are not clear in the PSDM image. Third, the RTM image shows a lot of fine structures below the subducting plate which are almost absent in the PSDM image. Finally, the RTM image indicates that the plate is gently dipping within the undeformed sediment (0-50 km) and becomes steeply dipping beyond 50 km as it enters the deformation front of the accretionary prism. Although the same conclusion could be drawn from the discontinuous plate boundary imaged by PSDM, RTM results are far more convincing than the PSDM.

DeepQA: improving the estimation of single protein model quality with deep belief networks.

PubMed

Cao, Renzhi; Bhattacharya, Debswapna; Hou, Jie; Cheng, Jianlin

2016-12-05

Protein quality assessment (QA) useful for ranking and selecting protein models has long been viewed as one of the major challenges for protein tertiary structure prediction. Especially, estimating the quality of a single protein model, which is important for selecting a few good models out of a large model pool consisting of mostly low-quality models, is still a largely unsolved problem. We introduce a novel single-model quality assessment method DeepQA based on deep belief network that utilizes a number of selected features describing the quality of a model from different perspectives, such as energy, physio-chemical characteristics, and structural information. The deep belief network is trained on several large datasets consisting of models from the Critical Assessment of Protein Structure Prediction (CASP) experiments, several publicly available datasets, and models generated by our in-house ab initio method. Our experiments demonstrate that deep belief network has better performance compared to Support Vector Machines and Neural Networks on the protein model quality assessment problem, and our method DeepQA achieves the state-of-the-art performance on CASP11 dataset. It also outperformed two well-established methods in selecting good outlier models from a large set of models of mostly low quality generated by ab initio modeling methods. DeepQA is a useful deep learning tool for protein single model quality assessment and protein structure prediction. The source code, executable, document and training/test datasets of DeepQA for Linux is freely available to non-commercial users at http://cactus.rnet.missouri.edu/DeepQA/ .

A Pilot Study of Biomedical Text Comprehension using an Attention-Based Deep Neural Reader: Design and Experimental Analysis

PubMed Central

Lee, Kyubum; Kim, Byounggun; Jeon, Minji; Kim, Jihye; Tan, Aik Choon

2018-01-01

Background With the development of artificial intelligence (AI) technology centered on deep-learning, the computer has evolved to a point where it can read a given text and answer a question based on the context of the text. Such a specific task is known as the task of machine comprehension. Existing machine comprehension tasks mostly use datasets of general texts, such as news articles or elementary school-level storybooks. However, no attempt has been made to determine whether an up-to-date deep learning-based machine comprehension model can also process scientific literature containing expert-level knowledge, especially in the biomedical domain. Objective This study aims to investigate whether a machine comprehension model can process biomedical articles as well as general texts. Since there is no dataset for the biomedical literature comprehension task, our work includes generating a large-scale question answering dataset using PubMed and manually evaluating the generated dataset. Methods We present an attention-based deep neural model tailored to the biomedical domain. To further enhance the performance of our model, we used a pretrained word vector and biomedical entity type embedding. We also developed an ensemble method of combining the results of several independent models to reduce the variance of the answers from the models. Results The experimental results showed that our proposed deep neural network model outperformed the baseline model by more than 7% on the new dataset. We also evaluated human performance on the new dataset. The human evaluation result showed that our deep neural model outperformed humans in comprehension by 22% on average. Conclusions In this work, we introduced a new task of machine comprehension in the biomedical domain using a deep neural model. Since there was no large-scale dataset for training deep neural models in the biomedical domain, we created the new cloze-style datasets Biomedical Knowledge Comprehension Title (BMKC_T) and Biomedical Knowledge Comprehension Last Sentence (BMKC_LS) (together referred to as BioMedical Knowledge Comprehension) using the PubMed corpus. The experimental results showed that the performance of our model is much higher than that of humans. We observed that our model performed consistently better regardless of the degree of difficulty of a text, whereas humans have difficulty when performing biomedical literature comprehension tasks that require expert level knowledge. PMID:29305341

Deep Learning for Classification of Colorectal Polyps on Whole-slide Images.

PubMed

Korbar, Bruno; Olofson, Andrea M; Miraflor, Allen P; Nicka, Catherine M; Suriawinata, Matthew A; Torresani, Lorenzo; Suriawinata, Arief A; Hassanpour, Saeed

2017-01-01

Histopathological characterization of colorectal polyps is critical for determining the risk of colorectal cancer and future rates of surveillance for patients. However, this characterization is a challenging task and suffers from significant inter- and intra-observer variability. We built an automatic image analysis method that can accurately classify different types of colorectal polyps on whole-slide images to help pathologists with this characterization and diagnosis. Our method is based on deep-learning techniques, which rely on numerous levels of abstraction for data representation and have shown state-of-the-art results for various image analysis tasks. Our method covers five common types of polyps (i.e., hyperplastic, sessile serrated, traditional serrated, tubular, and tubulovillous/villous) that are included in the US Multisociety Task Force guidelines for colorectal cancer risk assessment and surveillance. We developed multiple deep-learning approaches by leveraging a dataset of 2074 crop images, which were annotated by multiple domain expert pathologists as reference standards. We evaluated our method on an independent test set of 239 whole-slide images and measured standard machine-learning evaluation metrics of accuracy, precision, recall, and F1 score and their 95% confidence intervals. Our evaluation shows that our method with residual network architecture achieves the best performance for classification of colorectal polyps on whole-slide images (overall accuracy: 93.0%, 95% confidence interval: 89.0%-95.9%). Our method can reduce the cognitive burden on pathologists and improve their efficacy in histopathological characterization of colorectal polyps and in subsequent risk assessment and follow-up recommendations.

Deep Learning for Classification of Colorectal Polyps on Whole-slide Images

PubMed Central

Korbar, Bruno; Olofson, Andrea M.; Miraflor, Allen P.; Nicka, Catherine M.; Suriawinata, Matthew A.; Torresani, Lorenzo; Suriawinata, Arief A.; Hassanpour, Saeed

2017-01-01

Context: Histopathological characterization of colorectal polyps is critical for determining the risk of colorectal cancer and future rates of surveillance for patients. However, this characterization is a challenging task and suffers from significant inter- and intra-observer variability. Aims: We built an automatic image analysis method that can accurately classify different types of colorectal polyps on whole-slide images to help pathologists with this characterization and diagnosis. Setting and Design: Our method is based on deep-learning techniques, which rely on numerous levels of abstraction for data representation and have shown state-of-the-art results for various image analysis tasks. Subjects and Methods: Our method covers five common types of polyps (i.e., hyperplastic, sessile serrated, traditional serrated, tubular, and tubulovillous/villous) that are included in the US Multisociety Task Force guidelines for colorectal cancer risk assessment and surveillance. We developed multiple deep-learning approaches by leveraging a dataset of 2074 crop images, which were annotated by multiple domain expert pathologists as reference standards. Statistical Analysis: We evaluated our method on an independent test set of 239 whole-slide images and measured standard machine-learning evaluation metrics of accuracy, precision, recall, and F1 score and their 95% confidence intervals. Results: Our evaluation shows that our method with residual network architecture achieves the best performance for classification of colorectal polyps on whole-slide images (overall accuracy: 93.0%, 95% confidence interval: 89.0%–95.9%). Conclusions: Our method can reduce the cognitive burden on pathologists and improve their efficacy in histopathological characterization of colorectal polyps and in subsequent risk assessment and follow-up recommendations. PMID:28828201

A warm Spitzer survey of the LSST/DES 'Deep drilling' fields

NASA Astrophysics Data System (ADS)

Lacy, Mark; Farrah, Duncan; Brandt, Niel; Sako, Masao; Richards, Gordon; Norris, Ray; Ridgway, Susan; Afonso, Jose; Brunner, Robert; Clements, Dave; Cooray, Asantha; Covone, Giovanni; D'Andrea, Chris; Dickinson, Mark; Ferguson, Harry; Frieman, Joshua; Gupta, Ravi; Hatziminaoglou, Evanthia; Jarvis, Matt; Kimball, Amy; Lubin, Lori; Mao, Minnie; Marchetti, Lucia; Mauduit, Jean-Christophe; Mei, Simona; Newman, Jeffrey; Nichol, Robert; Oliver, Seb; Perez-Fournon, Ismael; Pierre, Marguerite; Rottgering, Huub; Seymour, Nick; Smail, Ian; Surace, Jason; Thorman, Paul; Vaccari, Mattia; Verma, Aprajita; Wilson, Gillian; Wood-Vasey, Michael; Cane, Rachel; Wechsler, Risa; Martini, Paul; Evrard, August; McMahon, Richard; Borne, Kirk; Capozzi, Diego; Huang, Jiashang; Lagos, Claudia; Lidman, Chris; Maraston, Claudia; Pforr, Janine; Sajina, Anna; Somerville, Rachel; Strauss, Michael; Jones, Kristen; Barkhouse, Wayne; Cooper, Michael; Ballantyne, David; Jagannathan, Preshanth; Murphy, Eric; Pradoni, Isabella; Suntzeff, Nicholas; Covarrubias, Ricardo; Spitler, Lee

2014-12-01

We propose a warm Spitzer survey to microJy depth of the four predefined Deep Drilling Fields (DDFs) for the Large Synoptic Survey Telescope (LSST) (three of which are also deep drilling fields for the Dark Energy Survey (DES)). Imaging these fields with warm Spitzer is a key component of the overall success of these projects, that address the 'Physics of the Universe' theme of the Astro2010 decadal survey. With deep, accurate, near-infrared photometry from Spitzer in the DDFs, we will generate photometric redshift distributions to apply to the surveys as a whole. The DDFs are also the areas where the supernova searches of DES and LSST are concentrated, and deep Spitzer data is essential to obtain photometric redshifts, stellar masses and constraints on ages and metallicities for the >10000 supernova host galaxies these surveys will find. This 'DEEPDRILL' survey will also address the 'Cosmic Dawn' goal of Astro2010 through being deep enough to find all the >10^11 solar mass galaxies within the survey area out to z~6. DEEPDRILL will complete the final 24.4 square degrees of imaging in the DDFs, which, when added to the 14 square degrees already imaged to this depth, will map a volume of 1-Gpc^3 at z>2. It will find ~100 > 10^11 solar mass galaxies at z~5 and ~40 protoclusters at z>2, providing targets for JWST that can be found in no other way. The Spitzer data, in conjunction with the multiwavelength surveys in these fields, ranging from X-ray through far-infrared and cm-radio, will comprise a unique legacy dataset for studies of galaxy evolution.

Deep convolutional neural networks for pan-specific peptide-MHC class I binding prediction.

PubMed

Han, Youngmahn; Kim, Dongsup

2017-12-28

Computational scanning of peptide candidates that bind to a specific major histocompatibility complex (MHC) can speed up the peptide-based vaccine development process and therefore various methods are being actively developed. Recently, machine-learning-based methods have generated successful results by training large amounts of experimental data. However, many machine learning-based methods are generally less sensitive in recognizing locally-clustered interactions, which can synergistically stabilize peptide binding. Deep convolutional neural network (DCNN) is a deep learning method inspired by visual recognition process of animal brain and it is known to be able to capture meaningful local patterns from 2D images. Once the peptide-MHC interactions can be encoded into image-like array(ILA) data, DCNN can be employed to build a predictive model for peptide-MHC binding prediction. In this study, we demonstrated that DCNN is able to not only reliably predict peptide-MHC binding, but also sensitively detect locally-clustered interactions. Nonapeptide-HLA-A and -B binding data were encoded into ILA data. A DCNN, as a pan-specific prediction model, was trained on the ILA data. The DCNN showed higher performance than other prediction tools for the latest benchmark datasets, which consist of 43 datasets for 15 HLA-A alleles and 25 datasets for 10 HLA-B alleles. In particular, the DCNN outperformed other tools for alleles belonging to the HLA-A3 supertype. The F1 scores of the DCNN were 0.86, 0.94, and 0.67 for HLA-A*31:01, HLA-A*03:01, and HLA-A*68:01 alleles, respectively, which were significantly higher than those of other tools. We found that the DCNN was able to recognize locally-clustered interactions that could synergistically stabilize peptide binding. We developed ConvMHC, a web server to provide user-friendly web interfaces for peptide-MHC class I binding predictions using the DCNN. ConvMHC web server can be accessible via http://jumong.kaist.ac.kr:8080/convmhc . We developed a novel method for peptide-HLA-I binding predictions using DCNN trained on ILA data that encode peptide binding data and demonstrated the reliable performance of the DCNN in nonapeptide binding predictions through the independent evaluation on the latest IEDB benchmark datasets. Our approaches can be applied to characterize locally-clustered patterns in molecular interactions, such as protein/DNA, protein/RNA, and drug/protein interactions.

Image quality classification for DR screening using deep learning.

PubMed

FengLi Yu; Jing Sun; Annan Li; Jun Cheng; Cheng Wan; Jiang Liu

2017-07-01

The quality of input images significantly affects the outcome of automated diabetic retinopathy (DR) screening systems. Unlike the previous methods that only consider simple low-level features such as hand-crafted geometric and structural features, in this paper we propose a novel method for retinal image quality classification (IQC) that performs computational algorithms imitating the working of the human visual system. The proposed algorithm combines unsupervised features from saliency map and supervised features coming from convolutional neural networks (CNN), which are fed to an SVM to automatically detect high quality vs poor quality retinal fundus images. We demonstrate the superior performance of our proposed algorithm on a large retinal fundus image dataset and the method could achieve higher accuracy than other methods. Although retinal images are used in this study, the methodology is applicable to the image quality assessment and enhancement of other types of medical images.

Automated assessment of breast tissue density in non-contrast 3D CT images without image segmentation based on a deep CNN

NASA Astrophysics Data System (ADS)

Zhou, Xiangrong; Kano, Takuya; Koyasu, Hiromi; Li, Shuo; Zhou, Xinxin; Hara, Takeshi; Matsuo, Masayuki; Fujita, Hiroshi

2017-03-01

This paper describes a novel approach for the automatic assessment of breast density in non-contrast three-dimensional computed tomography (3D CT) images. The proposed approach trains and uses a deep convolutional neural network (CNN) from scratch to classify breast tissue density directly from CT images without segmenting the anatomical structures, which creates a bottleneck in conventional approaches. Our scheme determines breast density in a 3D breast region by decomposing the 3D region into several radial 2D-sections from the nipple, and measuring the distribution of breast tissue densities on each 2D section from different orientations. The whole scheme is designed as a compact network without the need for post-processing and provides high robustness and computational efficiency in clinical settings. We applied this scheme to a dataset of 463 non-contrast CT scans obtained from 30- to 45-year-old-women in Japan. The density of breast tissue in each CT scan was assigned to one of four categories (glandular tissue within the breast <25%, 25%-50%, 50%-75%, and >75%) by a radiologist as ground truth. We used 405 CT scans for training a deep CNN and the remaining 58 CT scans for testing the performance. The experimental results demonstrated that the findings of the proposed approach and those of the radiologist were the same in 72% of the CT scans among the training samples and 76% among the testing samples. These results demonstrate the potential use of deep CNN for assessing breast tissue density in non-contrast 3D CT images.

Performance of an Artificial Multi-observer Deep Neural Network for Fully Automated Segmentation of Polycystic Kidneys.

PubMed

Kline, Timothy L; Korfiatis, Panagiotis; Edwards, Marie E; Blais, Jaime D; Czerwiec, Frank S; Harris, Peter C; King, Bernard F; Torres, Vicente E; Erickson, Bradley J

2017-08-01

Deep learning techniques are being rapidly applied to medical imaging tasks-from organ and lesion segmentation to tissue and tumor classification. These techniques are becoming the leading algorithmic approaches to solve inherently difficult image processing tasks. Currently, the most critical requirement for successful implementation lies in the need for relatively large datasets that can be used for training the deep learning networks. Based on our initial studies of MR imaging examinations of the kidneys of patients affected by polycystic kidney disease (PKD), we have generated a unique database of imaging data and corresponding reference standard segmentations of polycystic kidneys. In the study of PKD, segmentation of the kidneys is needed in order to measure total kidney volume (TKV). Automated methods to segment the kidneys and measure TKV are needed to increase measurement throughput and alleviate the inherent variability of human-derived measurements. We hypothesize that deep learning techniques can be leveraged to perform fast, accurate, reproducible, and fully automated segmentation of polycystic kidneys. Here, we describe a fully automated approach for segmenting PKD kidneys within MR images that simulates a multi-observer approach in order to create an accurate and robust method for the task of segmentation and computation of TKV for PKD patients. A total of 2000 cases were used for training and validation, and 400 cases were used for testing. The multi-observer ensemble method had mean ± SD percent volume difference of 0.68 ± 2.2% compared with the reference standard segmentations. The complete framework performs fully automated segmentation at a level comparable with interobserver variability and could be considered as a replacement for the task of segmentation of PKD kidneys by a human.

The Hubble Deep UV Legacy Survey (HDUV): Survey Overview and First Results

NASA Astrophysics Data System (ADS)

Oesch, Pascal; Montes, Mireia; HDUV Survey Team

2015-08-01

Deep HST imaging has shown that the overall star formation density and UV light density at z>3 is dominated by faint, blue galaxies. Remarkably, very little is known about the equivalent galaxy population at lower redshifts. Understanding how these galaxies evolve across the epoch of peak cosmic star-formation is key to a complete picture of galaxy evolution. Here, we present a new HST WFC3/UVIS program, the Hubble Deep UV (HDUV) legacy survey. The HDUV is a 132 orbit program to obtain deep imaging in two filters (F275W and F336W) over the two CANDELS Deep fields. We will cover ~100 arcmin2, reaching down to 27.5-28.0 mag at 5 sigma. By directly sampling the rest-frame far-UV at z>~0.5, this will provide a unique legacy dataset with exquisite HST multi-wavelength imaging as well as ancillary HST grism NIR spectroscopy for a detailed study of faint, star-forming galaxies at z~0.5-2. The HDUV will enable a wealth of research by the community, which includes tracing the evolution of the FUV luminosity function over the peak of the star formation rate density from z~3 down to z~0.5, measuring the physical properties of sub-L* galaxies, and characterizing resolved stellar populations to decipher the build-up of the Hubble sequence from sub-galactic clumps. This poster provides an overview of the HDUV survey and presents the reduced data products and catalogs which will be released to the community.

Rapid Target Detection in High Resolution Remote Sensing Images Using Yolo Model

NASA Astrophysics Data System (ADS)

Wu, Z.; Chen, X.; Gao, Y.; Li, Y.

2018-04-01

Object detection in high resolution remote sensing images is a fundamental and challenging problem in the field of remote sensing imagery analysis for civil and military application due to the complex neighboring environments, which can cause the recognition algorithms to mistake irrelevant ground objects for target objects. Deep Convolution Neural Network(DCNN) is the hotspot in object detection for its powerful ability of feature extraction and has achieved state-of-the-art results in Computer Vision. Common pipeline of object detection based on DCNN consists of region proposal, CNN feature extraction, region classification and post processing. YOLO model frames object detection as a regression problem, using a single CNN predicts bounding boxes and class probabilities in an end-to-end way and make the predict faster. In this paper, a YOLO based model is used for object detection in high resolution sensing images. The experiments on NWPU VHR-10 dataset and our airport/airplane dataset gain from GoogleEarth show that, compare with the common pipeline, the proposed model speeds up the detection process and have good accuracy.

A high-throughput system for high-quality tomographic reconstruction of large datasets at Diamond Light Source

PubMed Central

Atwood, Robert C.; Bodey, Andrew J.; Price, Stephen W. T.; Basham, Mark; Drakopoulos, Michael

2015-01-01

Tomographic datasets collected at synchrotrons are becoming very large and complex, and, therefore, need to be managed efficiently. Raw images may have high pixel counts, and each pixel can be multidimensional and associated with additional data such as those derived from spectroscopy. In time-resolved studies, hundreds of tomographic datasets can be collected in sequence, yielding terabytes of data. Users of tomographic beamlines are drawn from various scientific disciplines, and many are keen to use tomographic reconstruction software that does not require a deep understanding of reconstruction principles. We have developed Savu, a reconstruction pipeline that enables users to rapidly reconstruct data to consistently create high-quality results. Savu is designed to work in an ‘orthogonal’ fashion, meaning that data can be converted between projection and sinogram space throughout the processing workflow as required. The Savu pipeline is modular and allows processing strategies to be optimized for users' purposes. In addition to the reconstruction algorithms themselves, it can include modules for identification of experimental problems, artefact correction, general image processing and data quality assessment. Savu is open source, open licensed and ‘facility-independent’: it can run on standard cluster infrastructure at any institution. PMID:25939626

Pre-Juno Optical Analysis of Jupiter's Atmosphere with the NMSU Acousto-optic Imaging Camera

NASA Astrophysics Data System (ADS)

Dahl, Emma; Chanover, Nancy J.; Voelz, David; Kuehn, David M.; Strycker, Paul D.

2016-10-01

Jupiter's upper atmosphere is a highly dynamic system in which clouds and storms change color, shape, and size on variable timescales. The exact mechanism by which the deep atmosphere affects these changes in the uppermost cloud deck is still unknown. With Juno's arrival at Jupiter in July 2016, the thermal radiation from the deep atmosphere will be measurable with the spacecraft's Microwave Radiometer. By taking detailed optical measurements of Jupiter's uppermost cloud deck in conjunction with Juno's microwave observations, we can provide a context in which to better understand these observations. This data will also provide a complement to the near-IR sensitivity of the Jovian InfraRed Auroral Mapper and will expand on the limited spectral coverage of JunoCam. Ultimately, we can utilize the two complementary datasets in order to thoroughly characterize Jupiter's atmosphere in terms of its vertical cloud structure, color distribution, and dynamical state throughout the Juno era. In order to obtain high spectral resolution images of Jupiter's atmosphere in the optical regime, we use the New Mexico State University Acousto-optic Imaging Camera (NAIC). NAIC contains an acousto-optic tunable filter, which allows us to take hyperspectral image cubes of Jupiter from 450-950 nm at an average spectral resolution (λ/dλ) of 242. We present an analysis of our pre-Juno dataset obtained with NAIC at the Apache Point Observatory 3.5-m telescope during the night of March 28, 2016. Under primarily photometric conditions, we obtained 6 hyperspectral image cubes of Jupiter over the course of the night, totaling approximately 2,960 images. From these data we derive low-resolution optical spectra of the Great Red Spot and a representative belt and zone to compare with previous work and laboratory measurements of candidate chromophore materials. Future work will focus on radiative transfer modeling to elucidate the Jovian cloud structure during the Juno era. This work was supported by NASA through award number NNX15AP34A.

Automated Melanoma Recognition in Dermoscopy Images via Very Deep Residual Networks.

PubMed

Yu, Lequan; Chen, Hao; Dou, Qi; Qin, Jing; Heng, Pheng-Ann

2017-04-01

Automated melanoma recognition in dermoscopy images is a very challenging task due to the low contrast of skin lesions, the huge intraclass variation of melanomas, the high degree of visual similarity between melanoma and non-melanoma lesions, and the existence of many artifacts in the image. In order to meet these challenges, we propose a novel method for melanoma recognition by leveraging very deep convolutional neural networks (CNNs). Compared with existing methods employing either low-level hand-crafted features or CNNs with shallower architectures, our substantially deeper networks (more than 50 layers) can acquire richer and more discriminative features for more accurate recognition. To take full advantage of very deep networks, we propose a set of schemes to ensure effective training and learning under limited training data. First, we apply the residual learning to cope with the degradation and overfitting problems when a network goes deeper. This technique can ensure that our networks benefit from the performance gains achieved by increasing network depth. Then, we construct a fully convolutional residual network (FCRN) for accurate skin lesion segmentation, and further enhance its capability by incorporating a multi-scale contextual information integration scheme. Finally, we seamlessly integrate the proposed FCRN (for segmentation) and other very deep residual networks (for classification) to form a two-stage framework. This framework enables the classification network to extract more representative and specific features based on segmented results instead of the whole dermoscopy images, further alleviating the insufficiency of training data. The proposed framework is extensively evaluated on ISBI 2016 Skin Lesion Analysis Towards Melanoma Detection Challenge dataset. Experimental results demonstrate the significant performance gains of the proposed framework, ranking the first in classification and the second in segmentation among 25 teams and 28 teams, respectively. This study corroborates that very deep CNNs with effective training mechanisms can be employed to solve complicated medical image analysis tasks, even with limited training data.

Landmark-based deep multi-instance learning for brain disease diagnosis.

PubMed

Liu, Mingxia; Zhang, Jun; Adeli, Ehsan; Shen, Dinggang

2018-01-01

In conventional Magnetic Resonance (MR) image based methods, two stages are often involved to capture brain structural information for disease diagnosis, i.e., 1) manually partitioning each MR image into a number of regions-of-interest (ROIs), and 2) extracting pre-defined features from each ROI for diagnosis with a certain classifier. However, these pre-defined features often limit the performance of the diagnosis, due to challenges in 1) defining the ROIs and 2) extracting effective disease-related features. In this paper, we propose a landmark-based deep multi-instance learning (LDMIL) framework for brain disease diagnosis. Specifically, we first adopt a data-driven learning approach to discover disease-related anatomical landmarks in the brain MR images, along with their nearby image patches. Then, our LDMIL framework learns an end-to-end MR image classifier for capturing both the local structural information conveyed by image patches located by landmarks and the global structural information derived from all detected landmarks. We have evaluated our proposed framework on 1526 subjects from three public datasets (i.e., ADNI-1, ADNI-2, and MIRIAD), and the experimental results show that our framework can achieve superior performance over state-of-the-art approaches. Copyright © 2017 Elsevier B.V. All rights reserved.

Skin Lesion Analysis towards Melanoma Detection Using Deep Learning Network

PubMed Central

2018-01-01

Skin lesions are a severe disease globally. Early detection of melanoma in dermoscopy images significantly increases the survival rate. However, the accurate recognition of melanoma is extremely challenging due to the following reasons: low contrast between lesions and skin, visual similarity between melanoma and non-melanoma lesions, etc. Hence, reliable automatic detection of skin tumors is very useful to increase the accuracy and efficiency of pathologists. In this paper, we proposed two deep learning methods to address three main tasks emerging in the area of skin lesion image processing, i.e., lesion segmentation (task 1), lesion dermoscopic feature extraction (task 2) and lesion classification (task 3). A deep learning framework consisting of two fully convolutional residual networks (FCRN) is proposed to simultaneously produce the segmentation result and the coarse classification result. A lesion index calculation unit (LICU) is developed to refine the coarse classification results by calculating the distance heat-map. A straight-forward CNN is proposed for the dermoscopic feature extraction task. The proposed deep learning frameworks were evaluated on the ISIC 2017 dataset. Experimental results show the promising accuracies of our frameworks, i.e., 0.753 for task 1, 0.848 for task 2 and 0.912 for task 3 were achieved. PMID:29439500

IrisDenseNet: Robust Iris Segmentation Using Densely Connected Fully Convolutional Networks in the Images by Visible Light and Near-Infrared Light Camera Sensors

PubMed Central

Arsalan, Muhammad; Naqvi, Rizwan Ali; Kim, Dong Seop; Nguyen, Phong Ha; Owais, Muhammad; Park, Kang Ryoung

2018-01-01

The recent advancements in computer vision have opened new horizons for deploying biometric recognition algorithms in mobile and handheld devices. Similarly, iris recognition is now much needed in unconstraint scenarios with accuracy. These environments make the acquired iris image exhibit occlusion, low resolution, blur, unusual glint, ghost effect, and off-angles. The prevailing segmentation algorithms cannot cope with these constraints. In addition, owing to the unavailability of near-infrared (NIR) light, iris recognition in visible light environment makes the iris segmentation challenging with the noise of visible light. Deep learning with convolutional neural networks (CNN) has brought a considerable breakthrough in various applications. To address the iris segmentation issues in challenging situations by visible light and near-infrared light camera sensors, this paper proposes a densely connected fully convolutional network (IrisDenseNet), which can determine the true iris boundary even with inferior-quality images by using better information gradient flow between the dense blocks. In the experiments conducted, five datasets of visible light and NIR environments were used. For visible light environment, noisy iris challenge evaluation part-II (NICE-II selected from UBIRIS.v2 database) and mobile iris challenge evaluation (MICHE-I) datasets were used. For NIR environment, the institute of automation, Chinese academy of sciences (CASIA) v4.0 interval, CASIA v4.0 distance, and IIT Delhi v1.0 iris datasets were used. Experimental results showed the optimal segmentation of the proposed IrisDenseNet and its excellent performance over existing algorithms for all five datasets. PMID:29748495

IrisDenseNet: Robust Iris Segmentation Using Densely Connected Fully Convolutional Networks in the Images by Visible Light and Near-Infrared Light Camera Sensors.

PubMed

Arsalan, Muhammad; Naqvi, Rizwan Ali; Kim, Dong Seop; Nguyen, Phong Ha; Owais, Muhammad; Park, Kang Ryoung

2018-05-10

The recent advancements in computer vision have opened new horizons for deploying biometric recognition algorithms in mobile and handheld devices. Similarly, iris recognition is now much needed in unconstraint scenarios with accuracy. These environments make the acquired iris image exhibit occlusion, low resolution, blur, unusual glint, ghost effect, and off-angles. The prevailing segmentation algorithms cannot cope with these constraints. In addition, owing to the unavailability of near-infrared (NIR) light, iris recognition in visible light environment makes the iris segmentation challenging with the noise of visible light. Deep learning with convolutional neural networks (CNN) has brought a considerable breakthrough in various applications. To address the iris segmentation issues in challenging situations by visible light and near-infrared light camera sensors, this paper proposes a densely connected fully convolutional network (IrisDenseNet), which can determine the true iris boundary even with inferior-quality images by using better information gradient flow between the dense blocks. In the experiments conducted, five datasets of visible light and NIR environments were used. For visible light environment, noisy iris challenge evaluation part-II (NICE-II selected from UBIRIS.v2 database) and mobile iris challenge evaluation (MICHE-I) datasets were used. For NIR environment, the institute of automation, Chinese academy of sciences (CASIA) v4.0 interval, CASIA v4.0 distance, and IIT Delhi v1.0 iris datasets were used. Experimental results showed the optimal segmentation of the proposed IrisDenseNet and its excellent performance over existing algorithms for all five datasets.

Segmentation and classification of colon glands with deep convolutional neural networks and total variation regularization.

PubMed

Kainz, Philipp; Pfeiffer, Michael; Urschler, Martin

2017-01-01

Segmentation of histopathology sections is a necessary preprocessing step for digital pathology. Due to the large variability of biological tissue, machine learning techniques have shown superior performance over conventional image processing methods. Here we present our deep neural network-based approach for segmentation and classification of glands in tissue of benign and malignant colorectal cancer, which was developed to participate in the GlaS@MICCAI2015 colon gland segmentation challenge. We use two distinct deep convolutional neural networks (CNN) for pixel-wise classification of Hematoxylin-Eosin stained images. While the first classifier separates glands from background, the second classifier identifies gland-separating structures. In a subsequent step, a figure-ground segmentation based on weighted total variation produces the final segmentation result by regularizing the CNN predictions. We present both quantitative and qualitative segmentation results on the recently released and publicly available Warwick-QU colon adenocarcinoma dataset associated with the GlaS@MICCAI2015 challenge and compare our approach to the simultaneously developed other approaches that participated in the same challenge. On two test sets, we demonstrate our segmentation performance and show that we achieve a tissue classification accuracy of 98% and 95%, making use of the inherent capability of our system to distinguish between benign and malignant tissue. Our results show that deep learning approaches can yield highly accurate and reproducible results for biomedical image analysis, with the potential to significantly improve the quality and speed of medical diagnoses.

Segmentation and classification of colon glands with deep convolutional neural networks and total variation regularization

PubMed Central

Kainz, Philipp; Pfeiffer, Michael

2017-01-01

Segmentation of histopathology sections is a necessary preprocessing step for digital pathology. Due to the large variability of biological tissue, machine learning techniques have shown superior performance over conventional image processing methods. Here we present our deep neural network-based approach for segmentation and classification of glands in tissue of benign and malignant colorectal cancer, which was developed to participate in the GlaS@MICCAI2015 colon gland segmentation challenge. We use two distinct deep convolutional neural networks (CNN) for pixel-wise classification of Hematoxylin-Eosin stained images. While the first classifier separates glands from background, the second classifier identifies gland-separating structures. In a subsequent step, a figure-ground segmentation based on weighted total variation produces the final segmentation result by regularizing the CNN predictions. We present both quantitative and qualitative segmentation results on the recently released and publicly available Warwick-QU colon adenocarcinoma dataset associated with the GlaS@MICCAI2015 challenge and compare our approach to the simultaneously developed other approaches that participated in the same challenge. On two test sets, we demonstrate our segmentation performance and show that we achieve a tissue classification accuracy of 98% and 95%, making use of the inherent capability of our system to distinguish between benign and malignant tissue. Our results show that deep learning approaches can yield highly accurate and reproducible results for biomedical image analysis, with the potential to significantly improve the quality and speed of medical diagnoses. PMID:29018612

A Pilot Study of Biomedical Text Comprehension using an Attention-Based Deep Neural Reader: Design and Experimental Analysis.

PubMed

Kim, Seongsoon; Park, Donghyeon; Choi, Yonghwa; Lee, Kyubum; Kim, Byounggun; Jeon, Minji; Kim, Jihye; Tan, Aik Choon; Kang, Jaewoo

2018-01-05

With the development of artificial intelligence (AI) technology centered on deep-learning, the computer has evolved to a point where it can read a given text and answer a question based on the context of the text. Such a specific task is known as the task of machine comprehension. Existing machine comprehension tasks mostly use datasets of general texts, such as news articles or elementary school-level storybooks. However, no attempt has been made to determine whether an up-to-date deep learning-based machine comprehension model can also process scientific literature containing expert-level knowledge, especially in the biomedical domain. This study aims to investigate whether a machine comprehension model can process biomedical articles as well as general texts. Since there is no dataset for the biomedical literature comprehension task, our work includes generating a large-scale question answering dataset using PubMed and manually evaluating the generated dataset. We present an attention-based deep neural model tailored to the biomedical domain. To further enhance the performance of our model, we used a pretrained word vector and biomedical entity type embedding. We also developed an ensemble method of combining the results of several independent models to reduce the variance of the answers from the models. The experimental results showed that our proposed deep neural network model outperformed the baseline model by more than 7% on the new dataset. We also evaluated human performance on the new dataset. The human evaluation result showed that our deep neural model outperformed humans in comprehension by 22% on average. In this work, we introduced a new task of machine comprehension in the biomedical domain using a deep neural model. Since there was no large-scale dataset for training deep neural models in the biomedical domain, we created the new cloze-style datasets Biomedical Knowledge Comprehension Title (BMKC_T) and Biomedical Knowledge Comprehension Last Sentence (BMKC_LS) (together referred to as BioMedical Knowledge Comprehension) using the PubMed corpus. The experimental results showed that the performance of our model is much higher than that of humans. We observed that our model performed consistently better regardless of the degree of difficulty of a text, whereas humans have difficulty when performing biomedical literature comprehension tasks that require expert level knowledge. ©Seongsoon Kim, Donghyeon Park, Yonghwa Choi, Kyubum Lee, Byounggun Kim, Minji Jeon, Jihye Kim, Aik Choon Tan, Jaewoo Kang. Originally published in JMIR Medical Informatics (http://medinform.jmir.org), 05.01.2018.

Automatic classification of cardioembolic and arteriosclerotic ischemic strokes from apparent diffusion coefficient datasets using texture analysis and deep learning

NASA Astrophysics Data System (ADS)

Villafruela, Javier; Crites, Sebastian; Cheng, Bastian; Knaack, Christian; Thomalla, Götz; Menon, Bijoy K.; Forkert, Nils D.

2017-03-01

Stroke is a leading cause of death and disability in the western hemisphere. Acute ischemic strokes can be broadly classified based on the underlying cause into atherosclerotic strokes, cardioembolic strokes, small vessels disease, and stroke with other causes. The ability to determine the exact origin of an acute ischemic stroke is highly relevant for optimal treatment decision and preventing recurrent events. However, the differentiation of atherosclerotic and cardioembolic phenotypes can be especially challenging due to similar appearance and symptoms. The aim of this study was to develop and evaluate the feasibility of an image-based machine learning approach for discriminating between arteriosclerotic and cardioembolic acute ischemic strokes using 56 apparent diffusion coefficient (ADC) datasets from acute stroke patients. For this purpose, acute infarct lesions were semi-atomically segmented and 30,981 geometric and texture image features were extracted for each stroke volume. To improve the performance and accuracy, categorical Pearson's χ2 test was used to select the most informative features while removing redundant attributes. As a result, only 289 features were finally included for training of a deep multilayer feed-forward neural network without bootstrapping. The proposed method was evaluated using a leave-one-out cross validation scheme. The proposed classification method achieved an average area under receiver operator characteristic curve value of 0.93 and a classification accuracy of 94.64%. These first results suggest that the proposed image-based classification framework can support neurologists in clinical routine differentiating between atherosclerotic and cardioembolic phenotypes.

Violent Interaction Detection in Video Based on Deep Learning

NASA Astrophysics Data System (ADS)

Zhou, Peipei; Ding, Qinghai; Luo, Haibo; Hou, Xinglin

2017-06-01

Violent interaction detection is of vital importance in some video surveillance scenarios like railway stations, prisons or psychiatric centres. Existing vision-based methods are mainly based on hand-crafted features such as statistic features between motion regions, leading to a poor adaptability to another dataset. En lightened by the development of convolutional networks on common activity recognition, we construct a FightNet to represent the complicated visual violence interaction. In this paper, a new input modality, image acceleration field is proposed to better extract the motion attributes. Firstly, each video is framed as RGB images. Secondly, optical flow field is computed using the consecutive frames and acceleration field is obtained according to the optical flow field. Thirdly, the FightNet is trained with three kinds of input modalities, i.e., RGB images for spatial networks, optical flow images and acceleration images for temporal networks. By fusing results from different inputs, we conclude whether a video tells a violent event or not. To provide researchers a common ground for comparison, we have collected a violent interaction dataset (VID), containing 2314 videos with 1077 fight ones and 1237 no-fight ones. By comparison with other algorithms, experimental results demonstrate that the proposed model for violent interaction detection shows higher accuracy and better robustness.

Extraction of Built-Up Areas Using Convolutional Neural Networks and Transfer Learning from SENTINEL-2 Satellite Images

NASA Astrophysics Data System (ADS)

Bramhe, V. S.; Ghosh, S. K.; Garg, P. K.

2018-04-01

With rapid globalization, the extent of built-up areas is continuously increasing. Extraction of features for classifying built-up areas that are more robust and abstract is a leading research topic from past many years. Although, various studies have been carried out where spatial information along with spectral features has been utilized to enhance the accuracy of classification. Still, these feature extraction techniques require a large number of user-specific parameters and generally application specific. On the other hand, recently introduced Deep Learning (DL) techniques requires less number of parameters to represent more abstract aspects of the data without any manual effort. Since, it is difficult to acquire high-resolution datasets for applications that require large scale monitoring of areas. Therefore, in this study Sentinel-2 image has been used for built-up areas extraction. In this work, pre-trained Convolutional Neural Networks (ConvNets) i.e. Inception v3 and VGGNet are employed for transfer learning. Since these networks are trained on generic images of ImageNet dataset which are having very different characteristics from satellite images. Therefore, weights of networks are fine-tuned using data derived from Sentinel-2 images. To compare the accuracies with existing shallow networks, two state of art classifiers i.e. Gaussian Support Vector Machine (SVM) and Back-Propagation Neural Network (BP-NN) are also implemented. Both SVM and BP-NN gives 84.31 % and 82.86 % overall accuracies respectively. Inception-v3 and VGGNet gives 89.43 % of overall accuracy using fine-tuned VGGNet and 92.10 % when using Inception-v3. The results indicate high accuracy of proposed fine-tuned ConvNets on a 4-channel Sentinel-2 dataset for built-up area extraction.

Going deeper in the automated identification of Herbarium specimens.

PubMed

Carranza-Rojas, Jose; Goeau, Herve; Bonnet, Pierre; Mata-Montero, Erick; Joly, Alexis

2017-08-11

Hundreds of herbarium collections have accumulated a valuable heritage and knowledge of plants over several centuries. Recent initiatives started ambitious preservation plans to digitize this information and make it available to botanists and the general public through web portals. However, thousands of sheets are still unidentified at the species level while numerous sheets should be reviewed and updated following more recent taxonomic knowledge. These annotations and revisions require an unrealistic amount of work for botanists to carry out in a reasonable time. Computer vision and machine learning approaches applied to herbarium sheets are promising but are still not well studied compared to automated species identification from leaf scans or pictures of plants in the field. In this work, we propose to study and evaluate the accuracy with which herbarium images can be potentially exploited for species identification with deep learning technology. In addition, we propose to study if the combination of herbarium sheets with photos of plants in the field is relevant in terms of accuracy, and finally, we explore if herbarium images from one region that has one specific flora can be used to do transfer learning to another region with other species; for example, on a region under-represented in terms of collected data. This is, to our knowledge, the first study that uses deep learning to analyze a big dataset with thousands of species from herbaria. Results show the potential of Deep Learning on herbarium species identification, particularly by training and testing across different datasets from different herbaria. This could potentially lead to the creation of a semi, or even fully automated system to help taxonomists and experts with their annotation, classification, and revision works.

Emergence of Convolutional Neural Network in Future Medicine: Why and How. A Review on Brain Tumor Segmentation

NASA Astrophysics Data System (ADS)

Alizadeh Savareh, Behrouz; Emami, Hassan; Hajiabadi, Mohamadreza; Ghafoori, Mahyar; Majid Azimi, Seyed

2018-03-01

Manual analysis of brain tumors magnetic resonance images is usually accompanied by some problem. Several techniques have been proposed for the brain tumor segmentation. This study will be focused on searching popular databases for related studies, theoretical and practical aspects of Convolutional Neural Network surveyed in brain tumor segmentation. Based on our findings, details about related studies including the datasets used, evaluation parameters, preferred architectures and complementary steps analyzed. Deep learning as a revolutionary idea in image processing, achieved brilliant results in brain tumor segmentation too. This can be continuing until the next revolutionary idea emerging.

Location Sensitive Deep Convolutional Neural Networks for Segmentation of White Matter Hyperintensities.

PubMed

Ghafoorian, Mohsen; Karssemeijer, Nico; Heskes, Tom; van Uden, Inge W M; Sanchez, Clara I; Litjens, Geert; de Leeuw, Frank-Erik; van Ginneken, Bram; Marchiori, Elena; Platel, Bram

2017-07-11

The anatomical location of imaging features is of crucial importance for accurate diagnosis in many medical tasks. Convolutional neural networks (CNN) have had huge successes in computer vision, but they lack the natural ability to incorporate the anatomical location in their decision making process, hindering success in some medical image analysis tasks. In this paper, to integrate the anatomical location information into the network, we propose several deep CNN architectures that consider multi-scale patches or take explicit location features while training. We apply and compare the proposed architectures for segmentation of white matter hyperintensities in brain MR images on a large dataset. As a result, we observe that the CNNs that incorporate location information substantially outperform a conventional segmentation method with handcrafted features as well as CNNs that do not integrate location information. On a test set of 50 scans, the best configuration of our networks obtained a Dice score of 0.792, compared to 0.805 for an independent human observer. Performance levels of the machine and the independent human observer were not statistically significantly different (p-value = 0.06).

Deep Adaptive Log-Demons: Diffeomorphic Image Registration with Very Large Deformations

PubMed Central

Jia, Kebin

2015-01-01

This paper proposes a new framework for capturing large and complex deformation in image registration. Traditionally, this challenging problem relies firstly on a preregistration, usually an affine matrix containing rotation, scale, and translation and afterwards on a nonrigid transformation. According to preregistration, the directly calculated affine matrix, which is obtained by limited pixel information, may misregistrate when large biases exist, thus misleading following registration subversively. To address this problem, for two-dimensional (2D) images, the two-layer deep adaptive registration framework proposed in this paper firstly accurately classifies the rotation parameter through multilayer convolutional neural networks (CNNs) and then identifies scale and translation parameters separately. For three-dimensional (3D) images, affine matrix is located through feature correspondences by a triplanar 2D CNNs. Then deformation removal is done iteratively through preregistration and demons registration. By comparison with the state-of-the-art registration framework, our method gains more accurate registration results on both synthetic and real datasets. Besides, principal component analysis (PCA) is combined with correlation like Pearson and Spearman to form new similarity standards in 2D and 3D registration. Experiment results also show faster convergence speed. PMID:26120356

Deep Adaptive Log-Demons: Diffeomorphic Image Registration with Very Large Deformations.

PubMed

Zhao, Liya; Jia, Kebin

2015-01-01

This paper proposes a new framework for capturing large and complex deformation in image registration. Traditionally, this challenging problem relies firstly on a preregistration, usually an affine matrix containing rotation, scale, and translation and afterwards on a nonrigid transformation. According to preregistration, the directly calculated affine matrix, which is obtained by limited pixel information, may misregistrate when large biases exist, thus misleading following registration subversively. To address this problem, for two-dimensional (2D) images, the two-layer deep adaptive registration framework proposed in this paper firstly accurately classifies the rotation parameter through multilayer convolutional neural networks (CNNs) and then identifies scale and translation parameters separately. For three-dimensional (3D) images, affine matrix is located through feature correspondences by a triplanar 2D CNNs. Then deformation removal is done iteratively through preregistration and demons registration. By comparison with the state-of-the-art registration framework, our method gains more accurate registration results on both synthetic and real datasets. Besides, principal component analysis (PCA) is combined with correlation like Pearson and Spearman to form new similarity standards in 2D and 3D registration. Experiment results also show faster convergence speed.

Fully Convolutional Neural Networks Improve Abdominal Organ Segmentation.

PubMed

Bobo, Meg F; Bao, Shunxing; Huo, Yuankai; Yao, Yuang; Virostko, Jack; Plassard, Andrew J; Lyu, Ilwoo; Assad, Albert; Abramson, Richard G; Hilmes, Melissa A; Landman, Bennett A

2018-03-01

Abdominal image segmentation is a challenging, yet important clinical problem. Variations in body size, position, and relative organ positions greatly complicate the segmentation process. Historically, multi-atlas methods have achieved leading results across imaging modalities and anatomical targets. However, deep learning is rapidly overtaking classical approaches for image segmentation. Recently, Zhou et al. showed that fully convolutional networks produce excellent results in abdominal organ segmentation of computed tomography (CT) scans. Yet, deep learning approaches have not been applied to whole abdomen magnetic resonance imaging (MRI) segmentation. Herein, we evaluate the applicability of an existing fully convolutional neural network (FCNN) designed for CT imaging to segment abdominal organs on T2 weighted (T2w) MRI's with two examples. In the primary example, we compare a classical multi-atlas approach with FCNN on forty-five T2w MRI's acquired from splenomegaly patients with five organs labeled (liver, spleen, left kidney, right kidney, and stomach). Thirty-six images were used for training while nine were used for testing. The FCNN resulted in a Dice similarity coefficient (DSC) of 0.930 in spleens, 0.730 in left kidneys, 0.780 in right kidneys, 0.913 in livers, and 0.556 in stomachs. The performance measures for livers, spleens, right kidneys, and stomachs were significantly better than multi-atlas (p < 0.05, Wilcoxon rank-sum test). In a secondary example, we compare the multi-atlas approach with FCNN on 138 distinct T2w MRI's with manually labeled pancreases (one label). On the pancreas dataset, the FCNN resulted in a median DSC of 0.691 in pancreases versus 0.287 for multi-atlas. The results are highly promising given relatively limited training data and without specific training of the FCNN model and illustrate the potential of deep learning approaches to transcend imaging modalities.

Fully convolutional neural networks improve abdominal organ segmentation

NASA Astrophysics Data System (ADS)

Bobo, Meg F.; Bao, Shunxing; Huo, Yuankai; Yao, Yuang; Virostko, Jack; Plassard, Andrew J.; Lyu, Ilwoo; Assad, Albert; Abramson, Richard G.; Hilmes, Melissa A.; Landman, Bennett A.

2018-03-01

Abdominal image segmentation is a challenging, yet important clinical problem. Variations in body size, position, and relative organ positions greatly complicate the segmentation process. Historically, multi-atlas methods have achieved leading results across imaging modalities and anatomical targets. However, deep learning is rapidly overtaking classical approaches for image segmentation. Recently, Zhou et al. showed that fully convolutional networks produce excellent results in abdominal organ segmentation of computed tomography (CT) scans. Yet, deep learning approaches have not been applied to whole abdomen magnetic resonance imaging (MRI) segmentation. Herein, we evaluate the applicability of an existing fully convolutional neural network (FCNN) designed for CT imaging to segment abdominal organs on T2 weighted (T2w) MRI's with two examples. In the primary example, we compare a classical multi-atlas approach with FCNN on forty-five T2w MRI's acquired from splenomegaly patients with five organs labeled (liver, spleen, left kidney, right kidney, and stomach). Thirty-six images were used for training while nine were used for testing. The FCNN resulted in a Dice similarity coefficient (DSC) of 0.930 in spleens, 0.730 in left kidneys, 0.780 in right kidneys, 0.913 in livers, and 0.556 in stomachs. The performance measures for livers, spleens, right kidneys, and stomachs were significantly better than multi-atlas (p < 0.05, Wilcoxon rank-sum test). In a secondary example, we compare the multi-atlas approach with FCNN on 138 distinct T2w MRI's with manually labeled pancreases (one label). On the pancreas dataset, the FCNN resulted in a median DSC of 0.691 in pancreases versus 0.287 for multi-atlas. The results are highly promising given relatively limited training data and without specific training of the FCNN model and illustrate the potential of deep learning approaches to transcend imaging modalities. 1

Benchmarking Deep Learning Models on Large Healthcare Datasets.

PubMed

Purushotham, Sanjay; Meng, Chuizheng; Che, Zhengping; Liu, Yan

2018-06-04

Deep learning models (aka Deep Neural Networks) have revolutionized many fields including computer vision, natural language processing, speech recognition, and is being increasingly used in clinical healthcare applications. However, few works exist which have benchmarked the performance of the deep learning models with respect to the state-of-the-art machine learning models and prognostic scoring systems on publicly available healthcare datasets. In this paper, we present the benchmarking results for several clinical prediction tasks such as mortality prediction, length of stay prediction, and ICD-9 code group prediction using Deep Learning models, ensemble of machine learning models (Super Learner algorithm), SAPS II and SOFA scores. We used the Medical Information Mart for Intensive Care III (MIMIC-III) (v1.4) publicly available dataset, which includes all patients admitted to an ICU at the Beth Israel Deaconess Medical Center from 2001 to 2012, for the benchmarking tasks. Our results show that deep learning models consistently outperform all the other approaches especially when the 'raw' clinical time series data is used as input features to the models. Copyright © 2018 Elsevier Inc. All rights reserved.

Automated sub-cortical brain structure segmentation combining spatial and deep convolutional features.

PubMed

Kushibar, Kaisar; Valverde, Sergi; González-Villà, Sandra; Bernal, Jose; Cabezas, Mariano; Oliver, Arnau; Lladó, Xavier

2018-06-15

Sub-cortical brain structure segmentation in Magnetic Resonance Images (MRI) has attracted the interest of the research community for a long time as morphological changes in these structures are related to different neurodegenerative disorders. However, manual segmentation of these structures can be tedious and prone to variability, highlighting the need for robust automated segmentation methods. In this paper, we present a novel convolutional neural network based approach for accurate segmentation of the sub-cortical brain structures that combines both convolutional and prior spatial features for improving the segmentation accuracy. In order to increase the accuracy of the automated segmentation, we propose to train the network using a restricted sample selection to force the network to learn the most difficult parts of the structures. We evaluate the accuracy of the proposed method on the public MICCAI 2012 challenge and IBSR 18 datasets, comparing it with different traditional and deep learning state-of-the-art methods. On the MICCAI 2012 dataset, our method shows an excellent performance comparable to the best participant strategy on the challenge, while performing significantly better than state-of-the-art techniques such as FreeSurfer and FIRST. On the IBSR 18 dataset, our method also exhibits a significant increase in the performance with respect to not only FreeSurfer and FIRST, but also comparable or better results than other recent deep learning approaches. Moreover, our experiments show that both the addition of the spatial priors and the restricted sampling strategy have a significant effect on the accuracy of the proposed method. In order to encourage the reproducibility and the use of the proposed method, a public version of our approach is available to download for the neuroimaging community. Copyright © 2018 The Authors. Published by Elsevier B.V. All rights reserved.

DeepSig: deep learning improves signal peptide detection in proteins.

PubMed

Savojardo, Castrense; Martelli, Pier Luigi; Fariselli, Piero; Casadio, Rita

2018-05-15

The identification of signal peptides in protein sequences is an important step toward protein localization and function characterization. Here, we present DeepSig, an improved approach for signal peptide detection and cleavage-site prediction based on deep learning methods. Comparative benchmarks performed on an updated independent dataset of proteins show that DeepSig is the current best performing method, scoring better than other available state-of-the-art approaches on both signal peptide detection and precise cleavage-site identification. DeepSig is available as both standalone program and web server at https://deepsig.biocomp.unibo.it. All datasets used in this study can be obtained from the same website. pierluigi.martelli@unibo.it. Supplementary data are available at Bioinformatics online.

Deep embedding convolutional neural network for synthesizing CT image from T1-Weighted MR image.

PubMed

Xiang, Lei; Wang, Qian; Nie, Dong; Zhang, Lichi; Jin, Xiyao; Qiao, Yu; Shen, Dinggang

2018-07-01

Recently, more and more attention is drawn to the field of medical image synthesis across modalities. Among them, the synthesis of computed tomography (CT) image from T1-weighted magnetic resonance (MR) image is of great importance, although the mapping between them is highly complex due to large gaps of appearances of the two modalities. In this work, we aim to tackle this MR-to-CT synthesis task by a novel deep embedding convolutional neural network (DECNN). Specifically, we generate the feature maps from MR images, and then transform these feature maps forward through convolutional layers in the network. We can further compute a tentative CT synthesis from the midway of the flow of feature maps, and then embed this tentative CT synthesis result back to the feature maps. This embedding operation results in better feature maps, which are further transformed forward in DECNN. After repeating this embedding procedure for several times in the network, we can eventually synthesize a final CT image in the end of the DECNN. We have validated our proposed method on both brain and prostate imaging datasets, by also comparing with the state-of-the-art methods. Experimental results suggest that our DECNN (with repeated embedding operations) demonstrates its superior performances, in terms of both the perceptive quality of the synthesized CT image and the run-time cost for synthesizing a CT image. Copyright © 2018. Published by Elsevier B.V.

Can surgical simulation be used to train detection and classification of neural networks?

PubMed

Zisimopoulos, Odysseas; Flouty, Evangello; Stacey, Mark; Muscroft, Sam; Giataganas, Petros; Nehme, Jean; Chow, Andre; Stoyanov, Danail

2017-10-01

Computer-assisted interventions (CAI) aim to increase the effectiveness, precision and repeatability of procedures to improve surgical outcomes. The presence and motion of surgical tools is a key information input for CAI surgical phase recognition algorithms. Vision-based tool detection and recognition approaches are an attractive solution and can be designed to take advantage of the powerful deep learning paradigm that is rapidly advancing image recognition and classification. The challenge for such algorithms is the availability and quality of labelled data used for training. In this Letter, surgical simulation is used to train tool detection and segmentation based on deep convolutional neural networks and generative adversarial networks. The authors experiment with two network architectures for image segmentation in tool classes commonly encountered during cataract surgery. A commercially-available simulator is used to create a simulated cataract dataset for training models prior to performing transfer learning on real surgical data. To the best of authors' knowledge, this is the first attempt to train deep learning models for surgical instrument detection on simulated data while demonstrating promising results to generalise on real data. Results indicate that simulated data does have some potential for training advanced classification methods for CAI systems.

Cell dynamic morphology classification using deep convolutional neural networks.

PubMed

Li, Heng; Pang, Fengqian; Shi, Yonggang; Liu, Zhiwen

2018-05-15

Cell morphology is often used as a proxy measurement of cell status to understand cell physiology. Hence, interpretation of cell dynamic morphology is a meaningful task in biomedical research. Inspired by the recent success of deep learning, we here explore the application of convolutional neural networks (CNNs) to cell dynamic morphology classification. An innovative strategy for the implementation of CNNs is introduced in this study. Mouse lymphocytes were collected to observe the dynamic morphology, and two datasets were thus set up to investigate the performances of CNNs. Considering the installation of deep learning, the classification problem was simplified from video data to image data, and was then solved by CNNs in a self-taught manner with the generated image data. CNNs were separately performed in three installation scenarios and compared with existing methods. Experimental results demonstrated the potential of CNNs in cell dynamic morphology classification, and validated the effectiveness of the proposed strategy. CNNs were successfully applied to the classification problem, and outperformed the existing methods in the classification accuracy. For the installation of CNNs, transfer learning was proved to be a promising scheme. © 2018 International Society for Advancement of Cytometry. © 2018 International Society for Advancement of Cytometry.

TH-CD-206-05: Machine-Learning Based Segmentation of Organs at Risks for Head and Neck Radiotherapy Planning

DOE Office of Scientific and Technical Information (OSTI.GOV)

Ibragimov, B; Pernus, F; Strojan, P

Purpose: Accurate and efficient delineation of tumor target and organs-at-risks is essential for the success of radiotherapy. In reality, despite of decades of intense research efforts, auto-segmentation has not yet become clinical practice. In this study, we present, for the first time, a deep learning-based classification algorithm for autonomous segmentation in head and neck (HaN) treatment planning. Methods: Fifteen HN datasets of CT, MR and PET images with manual annotation of organs-at-risk (OARs) including spinal cord, brainstem, optic nerves, chiasm, eyes, mandible, tongue, parotid glands were collected and saved in a library of plans. We also have ten super-resolution MRmore » images of the tongue area, where the genioglossus and inferior longitudinalis tongue muscles are defined as organs of interest. We applied the concepts of random forest- and deep learning-based object classification for automated image annotation with the aim of using machine learning to facilitate head and neck radiotherapy planning process. In this new paradigm of segmentation, random forests were used for landmark-assisted segmentation of super-resolution MR images. Alternatively to auto-segmentation with random forest-based landmark detection, deep convolutional neural networks were developed for voxel-wise segmentation of OARs in single and multi-modal images. The network consisted of three pairs of convolution and pooing layer, one RuLU layer and a softmax layer. Results: We present a comprehensive study on using machine learning concepts for auto-segmentation of OARs and tongue muscles for the HaN radiotherapy planning. An accuracy of 81.8% in terms of Dice coefficient was achieved for segmentation of genioglossus and inferior longitudinalis tongue muscles. Preliminary results of OARs regimentation also indicate that deep-learning afforded an unprecedented opportunities to improve the accuracy and robustness of radiotherapy planning. Conclusion: A novel machine learning framework has been developed for image annotation and structure segmentation. Our results indicate the great potential of deep learning in radiotherapy treatment planning.« less

A deep convolutional neural network-based automatic delineation strategy for multiple brain metastases stereotactic radiosurgery.

PubMed

Liu, Yan; Stojadinovic, Strahinja; Hrycushko, Brian; Wardak, Zabi; Lau, Steven; Lu, Weiguo; Yan, Yulong; Jiang, Steve B; Zhen, Xin; Timmerman, Robert; Nedzi, Lucien; Gu, Xuejun

2017-01-01

Accurate and automatic brain metastases target delineation is a key step for efficient and effective stereotactic radiosurgery (SRS) treatment planning. In this work, we developed a deep learning convolutional neural network (CNN) algorithm for segmenting brain metastases on contrast-enhanced T1-weighted magnetic resonance imaging (MRI) datasets. We integrated the CNN-based algorithm into an automatic brain metastases segmentation workflow and validated on both Multimodal Brain Tumor Image Segmentation challenge (BRATS) data and clinical patients' data. Validation on BRATS data yielded average DICE coefficients (DCs) of 0.75±0.07 in the tumor core and 0.81±0.04 in the enhancing tumor, which outperformed most techniques in the 2015 BRATS challenge. Segmentation results of patient cases showed an average of DCs 0.67±0.03 and achieved an area under the receiver operating characteristic curve of 0.98±0.01. The developed automatic segmentation strategy surpasses current benchmark levels and offers a promising tool for SRS treatment planning for multiple brain metastases.

In-situ Planetary Subsurface Imaging System

NASA Astrophysics Data System (ADS)

Song, W.; Weber, R. C.; Dimech, J. L.; Kedar, S.; Neal, C. R.; Siegler, M.

2017-12-01

Geophysical and seismic instruments are considered the most effective tools for studying the detailed global structures of planetary interiors. A planet's interior bears the geochemical markers of its evolutionary history, as well as its present state of activity, which has direct implications to habitability. On Earth, subsurface imaging often involves massive data collection from hundreds to thousands of geophysical sensors (seismic, acoustic, etc) followed by transfer by hard links or wirelessly to a central location for post processing and computing, which will not be possible in planetary environments due to imposed mission constraints on mass, power, and bandwidth. Emerging opportunities for geophysical exploration of the solar system from Venus to the icy Ocean Worlds of Jupiter and Saturn dictate that subsurface imaging of the deep interior will require substantial data reduction and processing in-situ. The Real-time In-situ Subsurface Imaging (RISI) technology is a mesh network that senses and processes geophysical signals. Instead of data collection then post processing, the mesh network performs the distributed data processing and computing in-situ, and generates an evolving 3D subsurface image in real-time that can be transmitted under bandwidth and resource constraints. Seismic imaging algorithms (including traveltime tomography, ambient noise imaging, and microseismic imaging) have been successfully developed and validated using both synthetic and real-world terrestrial seismic data sets. The prototype hardware system has been implemented and can be extended as a general field instrumentation platform tailored specifically for a wide variety of planetary uses, including crustal mapping, ice and ocean structure, and geothermal systems. The team is applying the RISI technology to real off-world seismic datasets. For example, the Lunar Seismic Profiling Experiment (LSPE) deployed during the Apollo 17 Moon mission consisted of four geophone instruments spaced up to 100 meters apart, which in essence forms a small aperture seismic network. A pattern recognition technique based on Hidden Markov Models was able to characterize this dataset, and we are exploring how the RISI technology can be adapted for this dataset.

Detection of colorectal masses in CT colonography: application of deep residual networks for differentiating masses from normal colon anatomy

NASA Astrophysics Data System (ADS)

Näppi, Janne J.; Hironaka, Toru; Yoshida, Hiroyuki

2018-02-01

Even though the clinical consequences of a missed colorectal cancer far outweigh those of a missed polyp, there has been little work on computer-aided detection (CADe) for colorectal masses in CT colonography (CTC). One of the problems is that it is not clear how to manually design mathematical image-based features that could be used to differentiate effectively between masses and certain types of normal colon anatomy such as ileocecal valves (ICVs). Deep learning has demonstrated ability to automatically determine effective discriminating features in many image-based problems. Recently, residual networks (ResNets) were developed to address the practical problems of constructing deep network architectures for optimizing the performance of deep learning. In this pilot study, we compared the classification performance of a conventional 2D-convolutional ResNet (2D-ResNet) with that of a volumetric 3D-convolutional ResNet (3D-ResNet) in differentiating masses from normal colon anatomy in CTC. For the development and evaluation of the ResNets, 695 volumetric images of biopsy-proven colorectal masses, ICVs, haustral folds, and rectal tubes were sampled from 196 clinical CTC cases and divided randomly into independent training, validation, and test datasets. The training set was expanded by use of volumetric data augmentation. Our preliminary results on the 140 test samples indicate that it is feasible to train a deep volumetric 3D-ResNet for performing effective image-based discriminations in CTC. The 3D-ResNet slightly outperformed the 2D-ResNet in the discrimination of masses and normal colon anatomy, but the statistical difference between their very high classification accuracies was not significant. The highest classification accuracy was obtained by combining the mass-likelihood estimates of the 2D- and 3D-ResNets, which enabled correct classification of all of the masses.

Using Deep Learning for Tropical Cyclone Intensity Estimation

NASA Astrophysics Data System (ADS)

Miller, J.; Maskey, M.; Berendes, T.

2017-12-01

Satellite-based techniques are the primary approach to estimating tropical cyclone (TC) intensity. Tropical cyclone warning centers worldwide still apply variants of the Dvorak technique for such estimations that include visual inspection of the satellite images. The National Hurricane Center (NHC) estimates about 10-20% uncertainty in its post analyses when only satellite-based estimates are available. The success of the Dvorak technique proves that spatial patterns in infrared (IR) imagery strongly relate to TC intensity. With the ever-increasing quality and quantity of satellite observations of TCs, deep learning techniques designed to excel at pattern recognition have become more relevant in this area of study. In our current study, we aim to provide a fully objective approach to TC intensity estimation by utilizing deep learning in the form of a convolutional neural network trained to predict TC intensity (maximum sustained wind speed) using IR satellite imagery. Large amounts of training data are needed to train a convolutional neural network, so we use GOES IR images from historical tropical storms from the Atlantic and Pacific basins spanning years 2000 to 2015. Images are labeled using a special subset of the HURDAT2 dataset restricted to time periods with airborne reconnaissance data available in order to improve the quality of the HURDAT2 data. Results and the advantages of this technique are to be discussed.

A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition.

PubMed

Fuentes, Alvaro; Yoon, Sook; Kim, Sang Cheol; Park, Dong Sun

2017-09-04

Plant Diseases and Pests are a major challenge in the agriculture sector. An accurate and a faster detection of diseases and pests in plants could help to develop an early treatment technique while substantially reducing economic losses. Recent developments in Deep Neural Networks have allowed researchers to drastically improve the accuracy of object detection and recognition systems. In this paper, we present a deep-learning-based approach to detect diseases and pests in tomato plants using images captured in-place by camera devices with various resolutions. Our goal is to find the more suitable deep-learning architecture for our task. Therefore, we consider three main families of detectors: Faster Region-based Convolutional Neural Network (Faster R-CNN), Region-based Fully Convolutional Network (R-FCN), and Single Shot Multibox Detector (SSD), which for the purpose of this work are called "deep learning meta-architectures". We combine each of these meta-architectures with "deep feature extractors" such as VGG net and Residual Network (ResNet). We demonstrate the performance of deep meta-architectures and feature extractors, and additionally propose a method for local and global class annotation and data augmentation to increase the accuracy and reduce the number of false positives during training. We train and test our systems end-to-end on our large Tomato Diseases and Pests Dataset, which contains challenging images with diseases and pests, including several inter- and extra-class variations, such as infection status and location in the plant. Experimental results show that our proposed system can effectively recognize nine different types of diseases and pests, with the ability to deal with complex scenarios from a plant's surrounding area.

A Robust Deep-Learning-Based Detector for Real-Time Tomato Plant Diseases and Pests Recognition

PubMed Central

Yoon, Sook; Kim, Sang Cheol; Park, Dong Sun

2017-01-01

Plant Diseases and Pests are a major challenge in the agriculture sector. An accurate and a faster detection of diseases and pests in plants could help to develop an early treatment technique while substantially reducing economic losses. Recent developments in Deep Neural Networks have allowed researchers to drastically improve the accuracy of object detection and recognition systems. In this paper, we present a deep-learning-based approach to detect diseases and pests in tomato plants using images captured in-place by camera devices with various resolutions. Our goal is to find the more suitable deep-learning architecture for our task. Therefore, we consider three main families of detectors: Faster Region-based Convolutional Neural Network (Faster R-CNN), Region-based Fully Convolutional Network (R-FCN), and Single Shot Multibox Detector (SSD), which for the purpose of this work are called “deep learning meta-architectures”. We combine each of these meta-architectures with “deep feature extractors” such as VGG net and Residual Network (ResNet). We demonstrate the performance of deep meta-architectures and feature extractors, and additionally propose a method for local and global class annotation and data augmentation to increase the accuracy and reduce the number of false positives during training. We train and test our systems end-to-end on our large Tomato Diseases and Pests Dataset, which contains challenging images with diseases and pests, including several inter- and extra-class variations, such as infection status and location in the plant. Experimental results show that our proposed system can effectively recognize nine different types of diseases and pests, with the ability to deal with complex scenarios from a plant’s surrounding area. PMID:28869539

Convolutional Neural Network for Histopathological Analysis of Osteosarcoma.

PubMed

Mishra, Rashika; Daescu, Ovidiu; Leavey, Patrick; Rakheja, Dinesh; Sengupta, Anita

2018-03-01

Pathologists often deal with high complexity and sometimes disagreement over osteosarcoma tumor classification due to cellular heterogeneity in the dataset. Segmentation and classification of histology tissue in H&E stained tumor image datasets is a challenging task because of intra-class variations, inter-class similarity, crowded context, and noisy data. In recent years, deep learning approaches have led to encouraging results in breast cancer and prostate cancer analysis. In this article, we propose convolutional neural network (CNN) as a tool to improve efficiency and accuracy of osteosarcoma tumor classification into tumor classes (viable tumor, necrosis) versus nontumor. The proposed CNN architecture contains eight learned layers: three sets of stacked two convolutional layers interspersed with max pooling layers for feature extraction and two fully connected layers with data augmentation strategies to boost performance. The use of a neural network results in higher accuracy of average 92% for the classification. We compare the proposed architecture with three existing and proven CNN architectures for image classification: AlexNet, LeNet, and VGGNet. We also provide a pipeline to calculate percentage necrosis in a given whole slide image. We conclude that the use of neural networks can assure both high accuracy and efficiency in osteosarcoma classification.

Deformable MR Prostate Segmentation via Deep Feature Learning and Sparse Patch Matching

PubMed Central

Guo, Yanrong; Gao, Yaozong

2016-01-01

Automatic and reliable segmentation of the prostate is an important but difficult task for various clinical applications such as prostate cancer radiotherapy. The main challenges for accurate MR prostate localization lie in two aspects: (1) inhomogeneous and inconsistent appearance around prostate boundary, and (2) the large shape variation across different patients. To tackle these two problems, we propose a new deformable MR prostate segmentation method by unifying deep feature learning with the sparse patch matching. First, instead of directly using handcrafted features, we propose to learn the latent feature representation from prostate MR images by the stacked sparse auto-encoder (SSAE). Since the deep learning algorithm learns the feature hierarchy from the data, the learned features are often more concise and effective than the handcrafted features in describing the underlying data. To improve the discriminability of learned features, we further refine the feature representation in a supervised fashion. Second, based on the learned features, a sparse patch matching method is proposed to infer a prostate likelihood map by transferring the prostate labels from multiple atlases to the new prostate MR image. Finally, a deformable segmentation is used to integrate a sparse shape model with the prostate likelihood map for achieving the final segmentation. The proposed method has been extensively evaluated on the dataset that contains 66 T2-wighted prostate MR images. Experimental results show that the deep-learned features are more effective than the handcrafted features in guiding MR prostate segmentation. Moreover, our method shows superior performance than other state-of-the-art segmentation methods. PMID:26685226

Applying a machine learning model using a locally preserving projection based feature regeneration algorithm to predict breast cancer risk

NASA Astrophysics Data System (ADS)

Heidari, Morteza; Zargari Khuzani, Abolfazl; Danala, Gopichandh; Mirniaharikandehei, Seyedehnafiseh; Qian, Wei; Zheng, Bin

2018-03-01

Both conventional and deep machine learning has been used to develop decision-support tools applied in medical imaging informatics. In order to take advantages of both conventional and deep learning approach, this study aims to investigate feasibility of applying a locally preserving projection (LPP) based feature regeneration algorithm to build a new machine learning classifier model to predict short-term breast cancer risk. First, a computer-aided image processing scheme was used to segment and quantify breast fibro-glandular tissue volume. Next, initially computed 44 image features related to the bilateral mammographic tissue density asymmetry were extracted. Then, an LLP-based feature combination method was applied to regenerate a new operational feature vector using a maximal variance approach. Last, a k-nearest neighborhood (KNN) algorithm based machine learning classifier using the LPP-generated new feature vectors was developed to predict breast cancer risk. A testing dataset involving negative mammograms acquired from 500 women was used. Among them, 250 were positive and 250 remained negative in the next subsequent mammography screening. Applying to this dataset, LLP-generated feature vector reduced the number of features from 44 to 4. Using a leave-onecase-out validation method, area under ROC curve produced by the KNN classifier significantly increased from 0.62 to 0.68 (p < 0.05) and odds ratio was 4.60 with a 95% confidence interval of [3.16, 6.70]. Study demonstrated that this new LPP-based feature regeneration approach enabled to produce an optimal feature vector and yield improved performance in assisting to predict risk of women having breast cancer detected in the next subsequent mammography screening.

A Consistent EPIC Visible Channel Calibration Using VIIRS and MODIS as a Reference.

NASA Astrophysics Data System (ADS)

Haney, C.; Doelling, D. R.; Minnis, P.; Bhatt, R.; Scarino, B. R.; Gopalan, A.

2017-12-01

The Earth Polychromatic Imaging Camera (EPIC) aboard the Deep Space Climate Observatory (DSCOVR) satellite constantly images the sunlit disk of Earth from the Lagrange-1 (L1) point in 10 spectral channels spanning the UV, VIS, and NIR spectrums. Recently, the DSCOVR EPIC team has publicly released version 2 dataset, which has implemented improved navigation, stray-light correction, and flat-fielding of the CCD array. The EPIC 2-year data record must be well-calibrated for consistent cloud, aerosol, trace gas, land use and other retrievals. Because EPIC lacks onboard calibrators, the observations made by EPIC channels must be calibrated vicariously using the coincident measurements from radiometrically stable instruments that have onboard calibration systems. MODIS and VIIRS are best-suited instruments for this task as they contain similar spectral bands that are well-calibrated onboard using solar diffusers and lunar tracking. We have previously calibrated the EPIC version 1 dataset by using EPIC and VIIRS angularly matched radiance pairs over both all-sky ocean and deep convective clouds (DCC). We noted that the EPIC image required navigations adjustments, and that the EPIC stray-light correction provided an offset term closer to zero based on the linear regression of the EPIC and VIIRS ray-matched radiance pairs. We will evaluate the EPIC version 2 navigation and stray-light improvements using the same techniques. In addition, we will monitor the EPIC channel calibration over the two years for any temporal degradation or anomalous behavior. These two calibration methods will be further validated using desert and DCC invariant Earth targets. The radiometric characterization of the selected invariant targets is performed using multiple years of MODIS and VIIRS measurements. Results of these studies will be shown at the conference.

A Consistent EPIC Visible Channel Calibration using VIIRS and MODIS as a Reference

NASA Technical Reports Server (NTRS)

Haney, C. O.; Doelling, D. R.; Minnis, P.; Bhatt, R.; Scarino, B. R.; Gopalan, A.

2017-01-01

The Earth Polychromatic Imaging Camera (EPIC) aboard the Deep Space Climate Observatory (DSCOVR) satellite constantly images the sunlit disk of Earth from the Lagrange-1 (L1) point in 10 spectral channels spanning the UV, VIS, and NIR spectrums. Recently, the DSCOVR EPIC team has publicly released version 2 dataset, which has implemented improved navigation, stray-light correction, and flat-fielding of the CCD array. The EPIC 2-year data record must be well-calibrated for consistent cloud, aerosol, trace gas, land use and other retrievals. Because EPIC lacks onboard calibrators, the observations made by EPIC channels must be calibrated vicariously using the coincident measurements from radiometrically stable instruments that have onboard calibration systems. MODIS and VIIRS are best-suited instruments for this task as they contain similar spectral bands that are well-calibrated onboard using solar diffusers and lunar tracking. We have previously calibrated the EPIC version 1 dataset by using EPIC and VIIRS angularly matched radiance pairs over both all-sky ocean and deep convective clouds (DCC). We noted that the EPIC image required navigations adjustments, and that the EPIC stray-light correction provided an offset term closer to zero based on the linear regression of the EPIC and VIIRS ray-matched radiance pairs. We will evaluate the EPIC version 2 navigation and stray-light improvements using the same techniques. In addition, we will monitor the EPIC channel calibration over the two years for any temporal degradation or anomalous behavior. These two calibration methods will be further validated using desert and DCC invariant Earth targets. The radiometric characterization of the selected invariant targets is performed using multiple years of MODIS and VIIRS measurements. Results of these studies will be shown at the conference.

Two-phase deep convolutional neural network for reducing class skewness in histopathological images based breast cancer detection.

PubMed

Wahab, Noorul; Khan, Asifullah; Lee, Yeon Soo

2017-06-01

Different types of breast cancer are affecting lives of women across the world. Common types include Ductal carcinoma in situ (DCIS), Invasive ductal carcinoma (IDC), Tubular carcinoma, Medullary carcinoma, and Invasive lobular carcinoma (ILC). While detecting cancer, one important factor is mitotic count - showing how rapidly the cells are dividing. But the class imbalance problem, due to the small number of mitotic nuclei in comparison to the overwhelming number of non-mitotic nuclei, affects the performance of classification models. This work presents a two-phase model to mitigate the class biasness issue while classifying mitotic and non-mitotic nuclei in breast cancer histopathology images through a deep convolutional neural network (CNN). First, nuclei are segmented out using blue ratio and global binary thresholding. In Phase-1 a CNN is then trained on the segmented out 80×80 pixel patches based on a standard dataset. Hard non-mitotic examples are identified and augmented; mitotic examples are oversampled by rotation and flipping; whereas non-mitotic examples are undersampled by blue ratio histogram based k-means clustering. Based on this information from Phase-1, the dataset is modified for Phase-2 in order to reduce the effects of class imbalance. The proposed CNN architecture and data balancing technique yielded an F-measure of 0.79, and outperformed all the methods relying on specific handcrafted features, as well as those using a combination of handcrafted and CNN-generated features. Copyright © 2017 Elsevier Ltd. All rights reserved.

A deep-learning based automatic pulmonary nodule detection system

NASA Astrophysics Data System (ADS)

Zhao, Yiyuan; Zhao, Liang; Yan, Zhennan; Wolf, Matthias; Zhan, Yiqiang

2018-02-01

Lung cancer is the deadliest cancer worldwide. Early detection of lung cancer is a promising way to lower the risk of dying. Accurate pulmonary nodule detection in computed tomography (CT) images is crucial for early diagnosis of lung cancer. The development of computer-aided detection (CAD) system of pulmonary nodules contributes to making the CT analysis more accurate and with more efficiency. Recent studies from other groups have been focusing on lung cancer diagnosis CAD system by detecting medium to large nodules. However, to fully investigate the relevance between nodule features and cancer diagnosis, a CAD that is capable of detecting nodules with all sizes is needed. In this paper, we present a deep-learning based automatic all size pulmonary nodule detection system by cascading two artificial neural networks. We firstly use a U-net like 3D network to generate nodule candidates from CT images. Then, we use another 3D neural network to refine the locations of the nodule candidates generated from the previous subsystem. With the second sub-system, we bring the nodule candidates closer to the center of the ground truth nodule locations. We evaluate our system on a public CT dataset provided by the Lung Nodule Analysis (LUNA) 2016 grand challenge. The performance on the testing dataset shows that our system achieves 90% sensitivity with an average of 4 false positives per scan. This indicates that our system can be an aid for automatic nodule detection, which is beneficial for lung cancer diagnosis.

Convolutional neural network-based classification system design with compressed wireless sensor network images.

PubMed

Ahn, Jungmo; Park, JaeYeon; Park, Donghwan; Paek, Jeongyeup; Ko, JeongGil

2018-01-01

With the introduction of various advanced deep learning algorithms, initiatives for image classification systems have transitioned over from traditional machine learning algorithms (e.g., SVM) to Convolutional Neural Networks (CNNs) using deep learning software tools. A prerequisite in applying CNN to real world applications is a system that collects meaningful and useful data. For such purposes, Wireless Image Sensor Networks (WISNs), that are capable of monitoring natural environment phenomena using tiny and low-power cameras on resource-limited embedded devices, can be considered as an effective means of data collection. However, with limited battery resources, sending high-resolution raw images to the backend server is a burdensome task that has direct impact on network lifetime. To address this problem, we propose an energy-efficient pre- and post- processing mechanism using image resizing and color quantization that can significantly reduce the amount of data transferred while maintaining the classification accuracy in the CNN at the backend server. We show that, if well designed, an image in its highly compressed form can be well-classified with a CNN model trained in advance using adequately compressed data. Our evaluation using a real image dataset shows that an embedded device can reduce the amount of transmitted data by ∼71% while maintaining a classification accuracy of ∼98%. Under the same conditions, this process naturally reduces energy consumption by ∼71% compared to a WISN that sends the original uncompressed images.

Completing the Legacy of Hubble's Wide/Deep Fields: An Aligned Complete Dataset of 1220 Orbits on the GOODS-N/CANDELS-N Region

NASA Astrophysics Data System (ADS)

Illingworth, Garth

2017-08-01

The GOODS-N/CANDELS-N region is second only to the GOODS-S/ECDF-S region in the extent of its HST and Spitzer coverage, making it a remarkable science resource. Yet of 1220 orbits of ACS and WFC3/IR imaging from 27 programs on the GOODS-N region, fully 42% of the total, about 520 orbits of imaging data from 22 programs, remains unavailable in MAST as a high-level science data product (HLSP). The GOODS-N region dataset is a key Legacy field ( 3 Msec from HST, 6 Msec from Spitzer, and 2 Msec from Chandra). We propose to deliver, with catalogs, HST ACS and WFC3/IR HLSPs to MAST for all 1220 orbits of GOODS-N data. We will also deliver HLSPs for the EGS, UDS and the COSMOS CANDELS regions, including new data not included to date. These four HLSPs, 2300 orbits of HST data ( 75% of a HST Cycle ), will add substantially to (1) our understanding of the build-up of galaxies to z 6 in the first Gyr during reionization, (2) the development of galaxies over the subsequent Gyr to the peak of the star formation rate in the universe at z 2-3, and (3) the transition at z<2 of early star-forming galaxies to the full splendor of the Hubble sequence. We can do this major AR Legacy program, having submitted a HLSP of ALL 2442 orbits of HST data on the GOODS- S region (>950 orbits new). The total volume of data in the GOODS-S Hubble Legacy Field (HLF-GOODS-S) is 5.8 Msec in 7211 exposures ( 70% of a HST cycle). The HLF-GOODS-S includes 4 new deep areas akin to the HUDF/XDF. The four proposed NEW Hubble Legacy Field datasets will complement the Frontier Field datasets and our recent HLF-GOODS-S and HUDF/XDF HLSP submissions. They will be cornerstones of Hubble's Legacy as the JWST era dawns.

The 3D Reference Earth Model (REM-3D): Update and Outlook

NASA Astrophysics Data System (ADS)

Lekic, V.; Moulik, P.; Romanowicz, B. A.; Dziewonski, A. M.

2016-12-01

Elastic properties of the Earth's interior (e.g. density, rigidity, compressibility, anisotropy) vary spatially due to changes in temperature, pressure, composition, and flow. In the 20th century, seismologists have constructed reference models of how these quantities vary with depth, notably the PREM model of Dziewonski and Anderson (1981). These 1D reference earth models have proven indispensable in earthquake location, imaging of interior structure, understanding material properties under extreme conditions, and as a reference in other fields, such as particle physics and astronomy. Over the past three decades, more sophisticated efforts by seismologists have yielded several generations of models of how properties vary not only with depth, but also laterally. Yet, though these three-dimensional (3D) models exhibit compelling similarities at large scales, differences in the methodology, representation of structure, and dataset upon which they are based, have prevented the creation of 3D community reference models. We propose to overcome these challenges by compiling, reconciling, and distributing a long period (>15 s) reference seismic dataset, from which we will construct a 3D seismic reference model (REM-3D) for the Earth's mantle, which will come in two flavors: a long wavelength smoothly parameterized model and a set of regional profiles. Here, we summarize progress made in the construction of the reference long period dataset, and present preliminary versions of the REM-3D in order to illustrate the two flavors of REM-3D and their relative advantages and disadvantages. As a community reference model and with fully quantified uncertainties and tradeoffs, REM-3D will facilitate Earth imaging studies, earthquake characterization, inferences on temperature and composition in the deep interior, and be of improved utility to emerging scientific endeavors, such as neutrino geoscience. In this presentation, we outline the outlook for setting up advisory community working groups and the community workshop that would assess progress, evaluate model and dataset performance, identify avenues for improvement, and recommend strategies for maximizing model adoption in and utility for the deep Earth community.

Traffic sign recognition based on deep convolutional neural network

NASA Astrophysics Data System (ADS)

Yin, Shi-hao; Deng, Ji-cai; Zhang, Da-wei; Du, Jing-yuan

2017-11-01

Traffic sign recognition (TSR) is an important component of automated driving systems. It is a rather challenging task to design a high-performance classifier for the TSR system. In this paper, we propose a new method for TSR system based on deep convolutional neural network. In order to enhance the expression of the network, a novel structure (dubbed block-layer below) which combines network-in-network and residual connection is designed. Our network has 10 layers with parameters (block-layer seen as a single layer): the first seven are alternate convolutional layers and block-layers, and the remaining three are fully-connected layers. We train our TSR network on the German traffic sign recognition benchmark (GTSRB) dataset. To reduce overfitting, we perform data augmentation on the training images and employ a regularization method named "dropout". The activation function we employ in our network adopts scaled exponential linear units (SELUs), which can induce self-normalizing properties. To speed up the training, we use an efficient GPU to accelerate the convolutional operation. On the test dataset of GTSRB, we achieve the accuracy rate of 99.67%, exceeding the state-of-the-art results.

Marginal Space Deep Learning: Efficient Architecture for Volumetric Image Parsing.

PubMed

Ghesu, Florin C; Krubasik, Edward; Georgescu, Bogdan; Singh, Vivek; Yefeng Zheng; Hornegger, Joachim; Comaniciu, Dorin

2016-05-01

Robust and fast solutions for anatomical object detection and segmentation support the entire clinical workflow from diagnosis, patient stratification, therapy planning, intervention and follow-up. Current state-of-the-art techniques for parsing volumetric medical image data are typically based on machine learning methods that exploit large annotated image databases. Two main challenges need to be addressed, these are the efficiency in scanning high-dimensional parametric spaces and the need for representative image features which require significant efforts of manual engineering. We propose a pipeline for object detection and segmentation in the context of volumetric image parsing, solving a two-step learning problem: anatomical pose estimation and boundary delineation. For this task we introduce Marginal Space Deep Learning (MSDL), a novel framework exploiting both the strengths of efficient object parametrization in hierarchical marginal spaces and the automated feature design of Deep Learning (DL) network architectures. In the 3D context, the application of deep learning systems is limited by the very high complexity of the parametrization. More specifically 9 parameters are necessary to describe a restricted affine transformation in 3D, resulting in a prohibitive amount of billions of scanning hypotheses. The mechanism of marginal space learning provides excellent run-time performance by learning classifiers in clustered, high-probability regions in spaces of gradually increasing dimensionality. To further increase computational efficiency and robustness, in our system we learn sparse adaptive data sampling patterns that automatically capture the structure of the input. Given the object localization, we propose a DL-based active shape model to estimate the non-rigid object boundary. Experimental results are presented on the aortic valve in ultrasound using an extensive dataset of 2891 volumes from 869 patients, showing significant improvements of up to 45.2% over the state-of-the-art. To our knowledge, this is the first successful demonstration of the DL potential to detection and segmentation in full 3D data with parametrized representations.

A survey on object detection in optical remote sensing images

NASA Astrophysics Data System (ADS)

Cheng, Gong; Han, Junwei

2016-07-01

Object detection in optical remote sensing images, being a fundamental but challenging problem in the field of aerial and satellite image analysis, plays an important role for a wide range of applications and is receiving significant attention in recent years. While enormous methods exist, a deep review of the literature concerning generic object detection is still lacking. This paper aims to provide a review of the recent progress in this field. Different from several previously published surveys that focus on a specific object class such as building and road, we concentrate on more generic object categories including, but are not limited to, road, building, tree, vehicle, ship, airport, urban-area. Covering about 270 publications we survey (1) template matching-based object detection methods, (2) knowledge-based object detection methods, (3) object-based image analysis (OBIA)-based object detection methods, (4) machine learning-based object detection methods, and (5) five publicly available datasets and three standard evaluation metrics. We also discuss the challenges of current studies and propose two promising research directions, namely deep learning-based feature representation and weakly supervised learning-based geospatial object detection. It is our hope that this survey will be beneficial for the researchers to have better understanding of this research field.

Geoseq: a tool for dissecting deep-sequencing datasets.

PubMed

Gurtowski, James; Cancio, Anthony; Shah, Hardik; Levovitz, Chaya; George, Ajish; Homann, Robert; Sachidanandam, Ravi

2010-10-12

Datasets generated on deep-sequencing platforms have been deposited in various public repositories such as the Gene Expression Omnibus (GEO), Sequence Read Archive (SRA) hosted by the NCBI, or the DNA Data Bank of Japan (ddbj). Despite being rich data sources, they have not been used much due to the difficulty in locating and analyzing datasets of interest. Geoseq http://geoseq.mssm.edu provides a new method of analyzing short reads from deep sequencing experiments. Instead of mapping the reads to reference genomes or sequences, Geoseq maps a reference sequence against the sequencing data. It is web-based, and holds pre-computed data from public libraries. The analysis reduces the input sequence to tiles and measures the coverage of each tile in a sequence library through the use of suffix arrays. The user can upload custom target sequences or use gene/miRNA names for the search and get back results as plots and spreadsheet files. Geoseq organizes the public sequencing data using a controlled vocabulary, allowing identification of relevant libraries by organism, tissue and type of experiment. Analysis of small sets of sequences against deep-sequencing datasets, as well as identification of public datasets of interest, is simplified by Geoseq. We applied Geoseq to, a) identify differential isoform expression in mRNA-seq datasets, b) identify miRNAs (microRNAs) in libraries, and identify mature and star sequences in miRNAS and c) to identify potentially mis-annotated miRNAs. The ease of using Geoseq for these analyses suggests its utility and uniqueness as an analysis tool.

A method for medulloblastoma tumor differentiation based on convolutional neural networks and transfer learning

NASA Astrophysics Data System (ADS)

Cruz-Roa, Angel; Arévalo, John; Judkins, Alexander; Madabhushi, Anant; González, Fabio

2015-12-01

Convolutional neural networks (CNN) have been very successful at addressing different computer vision tasks thanks to their ability to learn image representations directly from large amounts of labeled data. Features learned from a dataset can be used to represent images from a different dataset via an approach called transfer learning. In this paper we apply transfer learning to the challenging task of medulloblastoma tumor differentiation. We compare two different CNN models which were previously trained in two different domains (natural and histopathology images). The first CNN is a state-of-the-art approach in computer vision, a large and deep CNN with 16-layers, Visual Geometry Group (VGG) CNN. The second (IBCa-CNN) is a 2-layer CNN trained for invasive breast cancer tumor classification. Both CNNs are used as visual feature extractors of histopathology image regions of anaplastic and non-anaplastic medulloblastoma tumor from digitized whole-slide images. The features from the two models are used, separately, to train a softmax classifier to discriminate between anaplastic and non-anaplastic medulloblastoma image regions. Experimental results show that the transfer learning approach produce competitive results in comparison with the state of the art approaches for IBCa detection. Results also show that features extracted from the IBCa-CNN have better performance in comparison with features extracted from the VGG-CNN. The former obtains 89.8% while the latter obtains 76.6% in terms of average accuracy.

Artificial Intelligence and Machine Learning in Radiology: Opportunities, Challenges, Pitfalls, and Criteria for Success.

PubMed

Thrall, James H; Li, Xiang; Li, Quanzheng; Cruz, Cinthia; Do, Synho; Dreyer, Keith; Brink, James

2018-03-01

Worldwide interest in artificial intelligence (AI) applications, including imaging, is high and growing rapidly, fueled by availability of large datasets ("big data"), substantial advances in computing power, and new deep-learning algorithms. Apart from developing new AI methods per se, there are many opportunities and challenges for the imaging community, including the development of a common nomenclature, better ways to share image data, and standards for validating AI program use across different imaging platforms and patient populations. AI surveillance programs may help radiologists prioritize work lists by identifying suspicious or positive cases for early review. AI programs can be used to extract "radiomic" information from images not discernible by visual inspection, potentially increasing the diagnostic and prognostic value derived from image datasets. Predictions have been made that suggest AI will put radiologists out of business. This issue has been overstated, and it is much more likely that radiologists will beneficially incorporate AI methods into their practices. Current limitations in availability of technical expertise and even computing power will be resolved over time and can also be addressed by remote access solutions. Success for AI in imaging will be measured by value created: increased diagnostic certainty, faster turnaround, better outcomes for patients, and better quality of work life for radiologists. AI offers a new and promising set of methods for analyzing image data. Radiologists will explore these new pathways and are likely to play a leading role in medical applications of AI. Copyright © 2017 American College of Radiology. Published by Elsevier Inc. All rights reserved.

Land-use Scene Classification in High-Resolution Remote Sensing Images by Multiscale Deeply Described Correlatons

NASA Astrophysics Data System (ADS)

Qi, K.; Qingfeng, G.

2017-12-01

With the popular use of High-Resolution Satellite (HRS) images, more and more research efforts have been placed on land-use scene classification. However, it makes the task difficult with HRS images for the complex background and multiple land-cover classes or objects. This article presents a multiscale deeply described correlaton model for land-use scene classification. Specifically, the convolutional neural network is introduced to learn and characterize the local features at different scales. Then, learnt multiscale deep features are explored to generate visual words. The spatial arrangement of visual words is achieved through the introduction of adaptive vector quantized correlograms at different scales. Experiments on two publicly available land-use scene datasets demonstrate that the proposed model is compact and yet discriminative for efficient representation of land-use scene images, and achieves competitive classification results with the state-of-art methods.

Fast and robust segmentation of the striatum using deep convolutional neural networks.

PubMed

Choi, Hongyoon; Jin, Kyong Hwan

2016-12-01

Automated segmentation of brain structures is an important task in structural and functional image analysis. We developed a fast and accurate method for the striatum segmentation using deep convolutional neural networks (CNN). T1 magnetic resonance (MR) images were used for our CNN-based segmentation, which require neither image feature extraction nor nonlinear transformation. We employed two serial CNN, Global and Local CNN: The Global CNN determined approximate locations of the striatum. It performed a regression of input MR images fitted to smoothed segmentation maps of the striatum. From the output volume of Global CNN, cropped MR volumes which included the striatum were extracted. The cropped MR volumes and the output volumes of Global CNN were used for inputs of Local CNN. Local CNN predicted the accurate label of all voxels. Segmentation results were compared with a widely used segmentation method, FreeSurfer. Our method showed higher Dice Similarity Coefficient (DSC) (0.893±0.017 vs. 0.786±0.015) and precision score (0.905±0.018 vs. 0.690±0.022) than FreeSurfer-based striatum segmentation (p=0.06). Our approach was also tested using another independent dataset, which showed high DSC (0.826±0.038) comparable with that of FreeSurfer. Comparison with existing method Segmentation performance of our proposed method was comparable with that of FreeSurfer. The running time of our approach was approximately three seconds. We suggested a fast and accurate deep CNN-based segmentation for small brain structures which can be widely applied to brain image analysis. Copyright © 2016 Elsevier B.V. All rights reserved.

Wave equation datuming applied to marine OBS data and to land high resolution seismic profiling

NASA Astrophysics Data System (ADS)

Barison, Erika; Brancatelli, Giuseppe; Nicolich, Rinaldo; Accaino, Flavio; Giustiniani, Michela; Tinivella, Umberta

2011-03-01

One key step in seismic data processing flows is the computation of static corrections, which relocate shots and receivers at the same datum plane and remove near surface weathering effects. We applied a standard static correction and a wave equation datuming and compared the obtained results in two case studies: 1) a sparse ocean bottom seismometers dataset for deep crustal prospecting; 2) a high resolution land reflection dataset for hydrogeological investigation. In both cases, a detailed velocity field, obtained by tomographic inversion of the first breaks, was adopted to relocate shots and receivers to the datum plane. The results emphasize the importance of wave equation datuming to properly handle complex near surface conditions. In the first dataset, the deployed ocean bottom seismometers were relocated to the sea level (shot positions) and a standard processing sequence was subsequently applied to the output. In the second dataset, the application of wave equation datuming allowed us to remove the coherent noise, such as ground roll, and to improve the image quality with respect to the application of static correction. The comparison of the two approaches evidences that the main reflecting markers are better resolved when the wave equation datuming procedure is adopted.

Pixel-Level Deep Segmentation: Artificial Intelligence Quantifies Muscle on Computed Tomography for Body Morphometric Analysis.

PubMed

Lee, Hyunkwang; Troschel, Fabian M; Tajmir, Shahein; Fuchs, Georg; Mario, Julia; Fintelmann, Florian J; Do, Synho

2017-08-01

Pretreatment risk stratification is key for personalized medicine. While many physicians rely on an "eyeball test" to assess whether patients will tolerate major surgery or chemotherapy, "eyeballing" is inherently subjective and difficult to quantify. The concept of morphometric age derived from cross-sectional imaging has been found to correlate well with outcomes such as length of stay, morbidity, and mortality. However, the determination of the morphometric age is time intensive and requires highly trained experts. In this study, we propose a fully automated deep learning system for the segmentation of skeletal muscle cross-sectional area (CSA) on an axial computed tomography image taken at the third lumbar vertebra. We utilized a fully automated deep segmentation model derived from an extended implementation of a fully convolutional network with weight initialization of an ImageNet pre-trained model, followed by post processing to eliminate intramuscular fat for a more accurate analysis. This experiment was conducted by varying window level (WL), window width (WW), and bit resolutions in order to better understand the effects of the parameters on the model performance. Our best model, fine-tuned on 250 training images and ground truth labels, achieves 0.93 ± 0.02 Dice similarity coefficient (DSC) and 3.68 ± 2.29% difference between predicted and ground truth muscle CSA on 150 held-out test cases. Ultimately, the fully automated segmentation system can be embedded into the clinical environment to accelerate the quantification of muscle and expanded to volume analysis of 3D datasets.

Improved Automated Detection of Diabetic Retinopathy on a Publicly Available Dataset Through Integration of Deep Learning.

PubMed

Abràmoff, Michael David; Lou, Yiyue; Erginay, Ali; Clarida, Warren; Amelon, Ryan; Folk, James C; Niemeijer, Meindert

2016-10-01

To compare performance of a deep-learning enhanced algorithm for automated detection of diabetic retinopathy (DR), to the previously published performance of that algorithm, the Iowa Detection Program (IDP)-without deep learning components-on the same publicly available set of fundus images and previously reported consensus reference standard set, by three US Board certified retinal specialists. We used the previously reported consensus reference standard of referable DR (rDR), defined as International Clinical Classification of Diabetic Retinopathy moderate, severe nonproliferative (NPDR), proliferative DR, and/or macular edema (ME). Neither Messidor-2 images, nor the three retinal specialists setting the Messidor-2 reference standard were used for training IDx-DR version X2.1. Sensitivity, specificity, negative predictive value, area under the curve (AUC), and their confidence intervals (CIs) were calculated. Sensitivity was 96.8% (95% CI: 93.3%-98.8%), specificity was 87.0% (95% CI: 84.2%-89.4%), with 6/874 false negatives, resulting in a negative predictive value of 99.0% (95% CI: 97.8%-99.6%). No cases of severe NPDR, PDR, or ME were missed. The AUC was 0.980 (95% CI: 0.968-0.992). Sensitivity was not statistically different from published IDP sensitivity, which had a CI of 94.4% to 99.3%, but specificity was significantly better than the published IDP specificity CI of 55.7% to 63.0%. A deep-learning enhanced algorithm for the automated detection of DR, achieves significantly better performance than a previously reported, otherwise essentially identical, algorithm that does not employ deep learning. Deep learning enhanced algorithms have the potential to improve the efficiency of DR screening, and thereby to prevent visual loss and blindness from this devastating disease.

A deep convolutional neural network for recognizing foods

NASA Astrophysics Data System (ADS)

Jahani Heravi, Elnaz; Habibi Aghdam, Hamed; Puig, Domenec

2015-12-01

Controlling the food intake is an efficient way that each person can undertake to tackle the obesity problem in countries worldwide. This is achievable by developing a smartphone application that is able to recognize foods and compute their calories. State-of-art methods are chiefly based on hand-crafted feature extraction methods such as HOG and Gabor. Recent advances in large-scale object recognition datasets such as ImageNet have revealed that deep Convolutional Neural Networks (CNN) possess more representation power than the hand-crafted features. The main challenge with CNNs is to find the appropriate architecture for each problem. In this paper, we propose a deep CNN which consists of 769; 988 parameters. Our experiments show that the proposed CNN outperforms the state-of-art methods and improves the best result of traditional methods 17%. Moreover, using an ensemble of two CNNs that have been trained two different times, we are able to improve the classification performance 21:5%.

Image Classification Using Biomimetic Pattern Recognition with Convolutional Neural Networks Features

PubMed Central

Huo, Guanying

2017-01-01

As a typical deep-learning model, Convolutional Neural Networks (CNNs) can be exploited to automatically extract features from images using the hierarchical structure inspired by mammalian visual system. For image classification tasks, traditional CNN models employ the softmax function for classification. However, owing to the limited capacity of the softmax function, there are some shortcomings of traditional CNN models in image classification. To deal with this problem, a new method combining Biomimetic Pattern Recognition (BPR) with CNNs is proposed for image classification. BPR performs class recognition by a union of geometrical cover sets in a high-dimensional feature space and therefore can overcome some disadvantages of traditional pattern recognition. The proposed method is evaluated on three famous image classification benchmarks, that is, MNIST, AR, and CIFAR-10. The classification accuracies of the proposed method for the three datasets are 99.01%, 98.40%, and 87.11%, respectively, which are much higher in comparison with the other four methods in most cases. PMID:28316614

Low-complexity object detection with deep convolutional neural network for embedded systems

NASA Astrophysics Data System (ADS)

Tripathi, Subarna; Kang, Byeongkeun; Dane, Gokce; Nguyen, Truong

2017-09-01

We investigate low-complexity convolutional neural networks (CNNs) for object detection for embedded vision applications. It is well-known that consolidation of an embedded system for CNN-based object detection is more challenging due to computation and memory requirement comparing with problems like image classification. To achieve these requirements, we design and develop an end-to-end TensorFlow (TF)-based fully-convolutional deep neural network for generic object detection task inspired by one of the fastest framework, YOLO.1 The proposed network predicts the localization of every object by regressing the coordinates of the corresponding bounding box as in YOLO. Hence, the network is able to detect any objects without any limitations in the size of the objects. However, unlike YOLO, all the layers in the proposed network is fully-convolutional. Thus, it is able to take input images of any size. We pick face detection as an use case. We evaluate the proposed model for face detection on FDDB dataset and Widerface dataset. As another use case of generic object detection, we evaluate its performance on PASCAL VOC dataset. The experimental results demonstrate that the proposed network can predict object instances of different sizes and poses in a single frame. Moreover, the results show that the proposed method achieves comparative accuracy comparing with the state-of-the-art CNN-based object detection methods while reducing the model size by 3× and memory-BW by 3 - 4× comparing with one of the best real-time CNN-based object detectors, YOLO. Our 8-bit fixed-point TF-model provides additional 4× memory reduction while keeping the accuracy nearly as good as the floating-point model. Moreover, the fixed- point model is capable of achieving 20× faster inference speed comparing with the floating-point model. Thus, the proposed method is promising for embedded implementations.

An improved advertising CTR prediction approach based on the fuzzy deep neural network

PubMed Central

Gao, Shu; Li, Mingjiang

2018-01-01

Combining a deep neural network with fuzzy theory, this paper proposes an advertising click-through rate (CTR) prediction approach based on a fuzzy deep neural network (FDNN). In this approach, fuzzy Gaussian-Bernoulli restricted Boltzmann machine (FGBRBM) is first applied to input raw data from advertising datasets. Next, fuzzy restricted Boltzmann machine (FRBM) is used to construct the fuzzy deep belief network (FDBN) with the unsupervised method layer by layer. Finally, fuzzy logistic regression (FLR) is utilized for modeling the CTR. The experimental results show that the proposed FDNN model outperforms several baseline models in terms of both data representation capability and robustness in advertising click log datasets with noise. PMID:29727443

An improved advertising CTR prediction approach based on the fuzzy deep neural network.

PubMed

Jiang, Zilong; Gao, Shu; Li, Mingjiang

2018-01-01

Combining a deep neural network with fuzzy theory, this paper proposes an advertising click-through rate (CTR) prediction approach based on a fuzzy deep neural network (FDNN). In this approach, fuzzy Gaussian-Bernoulli restricted Boltzmann machine (FGBRBM) is first applied to input raw data from advertising datasets. Next, fuzzy restricted Boltzmann machine (FRBM) is used to construct the fuzzy deep belief network (FDBN) with the unsupervised method layer by layer. Finally, fuzzy logistic regression (FLR) is utilized for modeling the CTR. The experimental results show that the proposed FDNN model outperforms several baseline models in terms of both data representation capability and robustness in advertising click log datasets with noise.

Significance of perceptually relevant image decolorization for scene classification

NASA Astrophysics Data System (ADS)

Viswanathan, Sowmya; Divakaran, Govind; Soman, Kutti Padanyl

2017-11-01

Color images contain luminance and chrominance components representing the intensity and color information, respectively. The objective of this paper is to show the significance of incorporating chrominance information to the task of scene classification. An improved color-to-grayscale image conversion algorithm that effectively incorporates chrominance information is proposed using the color-to-gray structure similarity index and singular value decomposition to improve the perceptual quality of the converted grayscale images. The experimental results based on an image quality assessment for image decolorization and its success rate (using the Cadik and COLOR250 datasets) show that the proposed image decolorization technique performs better than eight existing benchmark algorithms for image decolorization. In the second part of the paper, the effectiveness of incorporating the chrominance component for scene classification tasks is demonstrated using a deep belief network-based image classification system developed using dense scale-invariant feature transforms. The amount of chrominance information incorporated into the proposed image decolorization technique is confirmed with the improvement to the overall scene classification accuracy. Moreover, the overall scene classification performance improved by combining the models obtained using the proposed method and conventional decolorization methods.

Applying NASA Imaging Radar Datasets to Investigate the Geomorphology of the Amazon's Planalto

NASA Astrophysics Data System (ADS)

McDonald, K. C.; Campbell, K.; Islam, R.; Alexander, P. M.; Cracraft, J.

2016-12-01

The Amazon basin is a biodiversity rich biome and plays a significant role into shaping Earth's climate, ocean and atmospheric gases. Understanding the history of the formation of this basin is essential to our understanding of the region's biodiversity and its response to climate change. During March 2013, the NASA/JPL L-band polarimetric airborne imaging radar, UAVSAR, conducted airborne studies over regions of South America including portions of the western Amazon basin. We utilize UAVSAR imagery acquired during that time over the Planalto, in the Madre de Dios region of southeastern Peru in an assessment of the underlying geomorphology, its relationship to the current distribution of vegetation, and its relationship to geologic processes through deep time. We employ UAVSAR data collections to assess the utility of these high quality imaging radar data for use in identifying geomorphologic features and vegetation communities within the context of improving the understanding of evolutionary processes, and their utility in aiding interpretation of datasets from Earth-orbiting satellites to support a basin-wide characterization across the Amazon. We derive maps of landcover and river branching structure from UAVSAR imagery. We compare these maps to those derived using imaging radar datasets from the Japanese Space Agency's ALOS PALSAR and Digital Elevation Models (DEMs) from NASA's Shuttle Radar Topography Mission (SRTM). Results provide an understanding of the underlying geomorphology of the Amazon planalto as well as its relationship to geologic processes and will support interpretation of the evolutionary history of the Amazon Basin. Portions of this work have been carried out within the framework of the ALOS Kyoto & Carbon Initiative. PALSAR data were provided by JAXA/EORC and the Alaska Satellite Facility.This work is carried out with support from the NASA Biodiversity Program and the NSF DIMENSIONS of Biodiversity Program.

Conversion of Continuous-Valued Deep Networks to Efficient Event-Driven Networks for Image Classification.

PubMed

Rueckauer, Bodo; Lungu, Iulia-Alexandra; Hu, Yuhuang; Pfeiffer, Michael; Liu, Shih-Chii

2017-01-01

Spiking neural networks (SNNs) can potentially offer an efficient way of doing inference because the neurons in the networks are sparsely activated and computations are event-driven. Previous work showed that simple continuous-valued deep Convolutional Neural Networks (CNNs) can be converted into accurate spiking equivalents. These networks did not include certain common operations such as max-pooling, softmax, batch-normalization and Inception-modules. This paper presents spiking equivalents of these operations therefore allowing conversion of nearly arbitrary CNN architectures. We show conversion of popular CNN architectures, including VGG-16 and Inception-v3, into SNNs that produce the best results reported to date on MNIST, CIFAR-10 and the challenging ImageNet dataset. SNNs can trade off classification error rate against the number of available operations whereas deep continuous-valued neural networks require a fixed number of operations to achieve their classification error rate. From the examples of LeNet for MNIST and BinaryNet for CIFAR-10, we show that with an increase in error rate of a few percentage points, the SNNs can achieve more than 2x reductions in operations compared to the original CNNs. This highlights the potential of SNNs in particular when deployed on power-efficient neuromorphic spiking neuron chips, for use in embedded applications.

Conversion of Continuous-Valued Deep Networks to Efficient Event-Driven Networks for Image Classification

PubMed Central

Rueckauer, Bodo; Lungu, Iulia-Alexandra; Hu, Yuhuang; Pfeiffer, Michael; Liu, Shih-Chii

2017-01-01

Spiking neural networks (SNNs) can potentially offer an efficient way of doing inference because the neurons in the networks are sparsely activated and computations are event-driven. Previous work showed that simple continuous-valued deep Convolutional Neural Networks (CNNs) can be converted into accurate spiking equivalents. These networks did not include certain common operations such as max-pooling, softmax, batch-normalization and Inception-modules. This paper presents spiking equivalents of these operations therefore allowing conversion of nearly arbitrary CNN architectures. We show conversion of popular CNN architectures, including VGG-16 and Inception-v3, into SNNs that produce the best results reported to date on MNIST, CIFAR-10 and the challenging ImageNet dataset. SNNs can trade off classification error rate against the number of available operations whereas deep continuous-valued neural networks require a fixed number of operations to achieve their classification error rate. From the examples of LeNet for MNIST and BinaryNet for CIFAR-10, we show that with an increase in error rate of a few percentage points, the SNNs can achieve more than 2x reductions in operations compared to the original CNNs. This highlights the potential of SNNs in particular when deployed on power-efficient neuromorphic spiking neuron chips, for use in embedded applications. PMID:29375284

Scaling Deep Learning on GPU and Knights Landing clusters

DOE PAGES

You, Yang; Buluc, Aydin; Demmel, James

2017-09-26

The speed of deep neural networks training has become a big bottleneck of deep learning research and development. For example, training GoogleNet by ImageNet dataset on one Nvidia K20 GPU needs 21 days. To speed up the training process, the current deep learning systems heavily rely on the hardware accelerators. However, these accelerators have limited on-chip memory compared with CPUs. To handle large datasets, they need to fetch data from either CPU memory or remote processors. We use both self-hosted Intel Knights Landing (KNL) clusters and multi-GPU clusters as our target platforms. From an algorithm aspect, current distributed machine learningmore » systems are mainly designed for cloud systems. These methods are asynchronous because of the slow network and high fault-tolerance requirement on cloud systems. We focus on Elastic Averaging SGD (EASGD) to design algorithms for HPC clusters. Original EASGD used round-robin method for communication and updating. The communication is ordered by the machine rank ID, which is inefficient on HPC clusters. First, we redesign four efficient algorithms for HPC systems to improve EASGD's poor scaling on clusters. Async EASGD, Async MEASGD, and Hogwild EASGD are faster \\textcolor{black}{than} their existing counterparts (Async SGD, Async MSGD, and Hogwild SGD, resp.) in all the comparisons. Finally, we design Sync EASGD, which ties for the best performance among all the methods while being deterministic. In addition to the algorithmic improvements, we use some system-algorithm codesign techniques to scale up the algorithms. By reducing the percentage of communication from 87% to 14%, our Sync EASGD achieves 5.3x speedup over original EASGD on the same platform. We get 91.5% weak scaling efficiency on 4253 KNL cores, which is higher than the state-of-the-art implementation.« less

Scaling Deep Learning on GPU and Knights Landing clusters

DOE Office of Scientific and Technical Information (OSTI.GOV)

You, Yang; Buluc, Aydin; Demmel, James

The speed of deep neural networks training has become a big bottleneck of deep learning research and development. For example, training GoogleNet by ImageNet dataset on one Nvidia K20 GPU needs 21 days. To speed up the training process, the current deep learning systems heavily rely on the hardware accelerators. However, these accelerators have limited on-chip memory compared with CPUs. To handle large datasets, they need to fetch data from either CPU memory or remote processors. We use both self-hosted Intel Knights Landing (KNL) clusters and multi-GPU clusters as our target platforms. From an algorithm aspect, current distributed machine learningmore » systems are mainly designed for cloud systems. These methods are asynchronous because of the slow network and high fault-tolerance requirement on cloud systems. We focus on Elastic Averaging SGD (EASGD) to design algorithms for HPC clusters. Original EASGD used round-robin method for communication and updating. The communication is ordered by the machine rank ID, which is inefficient on HPC clusters. First, we redesign four efficient algorithms for HPC systems to improve EASGD's poor scaling on clusters. Async EASGD, Async MEASGD, and Hogwild EASGD are faster \\textcolor{black}{than} their existing counterparts (Async SGD, Async MSGD, and Hogwild SGD, resp.) in all the comparisons. Finally, we design Sync EASGD, which ties for the best performance among all the methods while being deterministic. In addition to the algorithmic improvements, we use some system-algorithm codesign techniques to scale up the algorithms. By reducing the percentage of communication from 87% to 14%, our Sync EASGD achieves 5.3x speedup over original EASGD on the same platform. We get 91.5% weak scaling efficiency on 4253 KNL cores, which is higher than the state-of-the-art implementation.« less

Friend or foe: exploiting sensor failures for transparent object localization and classification

NASA Astrophysics Data System (ADS)

Seib, Viktor; Barthen, Andreas; Marohn, Philipp; Paulus, Dietrich

2017-02-01

In this work we address the problem of detecting and recognizing transparent objects using depth images from an RGB-D camera. Using this type of sensor usually prohibits the localization of transparent objects since the structured light pattern of these cameras is not reflected by transparent surfaces. Instead, transparent surfaces often appear as undefined values in the resulting images. However, these erroneous sensor readings form characteristic patterns that we exploit in the presented approach. The sensor data is fed into a deep convolutional neural network that is trained to classify and localize drinking glasses. We evaluate our approach with four different types of transparent objects. To our best knowledge, no datasets offering depth images of transparent objects exist so far. With this work we aim at closing this gap by providing our data to the public.

Discriminatively learning for representing local image features with quadruplet model

NASA Astrophysics Data System (ADS)

Zhang, Da-long; Zhao, Lei; Xu, Duan-qing; Lu, Dong-ming

2017-11-01

Traditional hand-crafted features for representing local image patches are evolving into current data-driven and learning-based image feature, but learning a robust and discriminative descriptor which is capable of controlling various patch-level computer vision tasks is still an open problem. In this work, we propose a novel deep convolutional neural network (CNN) to learn local feature descriptors. We utilize the quadruplets with positive and negative training samples, together with a constraint to restrict the intra-class variance, to learn good discriminative CNN representations. Compared with previous works, our model reduces the overlap in feature space between corresponding and non-corresponding patch pairs, and mitigates margin varying problem caused by commonly used triplet loss. We demonstrate that our method achieves better embedding result than some latest works, like PN-Net and TN-TG, on benchmark dataset.

Pairwise domain adaptation module for CNN-based 2-D/3-D registration.

PubMed

Zheng, Jiannan; Miao, Shun; Jane Wang, Z; Liao, Rui

2018-04-01

Accurate two-dimensional to three-dimensional (2-D/3-D) registration of preoperative 3-D data and intraoperative 2-D x-ray images is a key enabler for image-guided therapy. Recent advances in 2-D/3-D registration formulate the problem as a learning-based approach and exploit the modeling power of convolutional neural networks (CNN) to significantly improve the accuracy and efficiency of 2-D/3-D registration. However, for surgery-related applications, collecting a large clinical dataset with accurate annotations for training can be very challenging or impractical. Therefore, deep learning-based 2-D/3-D registration methods are often trained with synthetically generated data, and a performance gap is often observed when testing the trained model on clinical data. We propose a pairwise domain adaptation (PDA) module to adapt the model trained on source domain (i.e., synthetic data) to target domain (i.e., clinical data) by learning domain invariant features with only a few paired real and synthetic data. The PDA module is designed to be flexible for different deep learning-based 2-D/3-D registration frameworks, and it can be plugged into any pretrained CNN model such as a simple Batch-Norm layer. The proposed PDA module has been quantitatively evaluated on two clinical applications using different frameworks of deep networks, demonstrating its significant advantages of generalizability and flexibility for 2-D/3-D medical image registration when a small number of paired real-synthetic data can be obtained.

Applying a deep learning based CAD scheme to segment and quantify visceral and subcutaneous fat areas from CT images

NASA Astrophysics Data System (ADS)

Wang, Yunzhi; Qiu, Yuchen; Thai, Theresa; Moore, Kathleen; Liu, Hong; Zheng, Bin

2017-03-01

Abdominal obesity is strongly associated with a number of diseases and accurately assessment of subtypes of adipose tissue volume plays a significant role in predicting disease risk, diagnosis and prognosis. The objective of this study is to develop and evaluate a new computer-aided detection (CAD) scheme based on deep learning models to automatically segment subcutaneous fat areas (SFA) and visceral (VFA) fat areas depicting on CT images. A dataset involving CT images from 40 patients were retrospectively collected and equally divided into two independent groups (i.e. training and testing group). The new CAD scheme consisted of two sequential convolutional neural networks (CNNs) namely, Selection-CNN and Segmentation-CNN. Selection-CNN was trained using 2,240 CT slices to automatically select CT slices belonging to abdomen areas and SegmentationCNN was trained using 84,000 fat-pixel patches to classify fat-pixels as belonging to SFA or VFA. Then, data from the testing group was used to evaluate the performance of the optimized CAD scheme. Comparing to manually labelled results, the classification accuracy of CT slices selection generated by Selection-CNN yielded 95.8%, while the accuracy of fat pixel segmentation using Segmentation-CNN yielded 96.8%. Therefore, this study demonstrated the feasibility of using deep learning based CAD scheme to recognize human abdominal section from CT scans and segment SFA and VFA from CT slices with high agreement compared with subjective segmentation results.

Results of the 2016 International Skin Imaging Collaboration International Symposium on Biomedical Imaging challenge: Comparison of the accuracy of computer algorithms to dermatologists for the diagnosis of melanoma from dermoscopic images.

PubMed

Marchetti, Michael A; Codella, Noel C F; Dusza, Stephen W; Gutman, David A; Helba, Brian; Kalloo, Aadi; Mishra, Nabin; Carrera, Cristina; Celebi, M Emre; DeFazio, Jennifer L; Jaimes, Natalia; Marghoob, Ashfaq A; Quigley, Elizabeth; Scope, Alon; Yélamos, Oriol; Halpern, Allan C

2018-02-01

Computer vision may aid in melanoma detection. We sought to compare melanoma diagnostic accuracy of computer algorithms to dermatologists using dermoscopic images. We conducted a cross-sectional study using 100 randomly selected dermoscopic images (50 melanomas, 44 nevi, and 6 lentigines) from an international computer vision melanoma challenge dataset (n = 379), along with individual algorithm results from 25 teams. We used 5 methods (nonlearned and machine learning) to combine individual automated predictions into "fusion" algorithms. In a companion study, 8 dermatologists classified the lesions in the 100 images as either benign or malignant. The average sensitivity and specificity of dermatologists in classification was 82% and 59%. At 82% sensitivity, dermatologist specificity was similar to the top challenge algorithm (59% vs. 62%, P = .68) but lower than the best-performing fusion algorithm (59% vs. 76%, P = .02). Receiver operating characteristic area of the top fusion algorithm was greater than the mean receiver operating characteristic area of dermatologists (0.86 vs. 0.71, P = .001). The dataset lacked the full spectrum of skin lesions encountered in clinical practice, particularly banal lesions. Readers and algorithms were not provided clinical data (eg, age or lesion history/symptoms). Results obtained using our study design cannot be extrapolated to clinical practice. Deep learning computer vision systems classified melanoma dermoscopy images with accuracy that exceeded some but not all dermatologists. Copyright © 2017 American Academy of Dermatology, Inc. Published by Elsevier Inc. All rights reserved.

Bag of Visual Words Model with Deep Spatial Features for Geographical Scene Classification

PubMed Central

Wu, Lin

2017-01-01

With the popular use of geotagging images, more and more research efforts have been placed on geographical scene classification. In geographical scene classification, valid spatial feature selection can significantly boost the final performance. Bag of visual words (BoVW) can do well in selecting feature in geographical scene classification; nevertheless, it works effectively only if the provided feature extractor is well-matched. In this paper, we use convolutional neural networks (CNNs) for optimizing proposed feature extractor, so that it can learn more suitable visual vocabularies from the geotagging images. Our approach achieves better performance than BoVW as a tool for geographical scene classification, respectively, in three datasets which contain a variety of scene categories. PMID:28706534

Deep visual-semantic for crowded video understanding

NASA Astrophysics Data System (ADS)

Deng, Chunhua; Zhang, Junwen

2018-03-01

Visual-semantic features play a vital role for crowded video understanding. Convolutional Neural Networks (CNNs) have experienced a significant breakthrough in learning representations from images. However, the learning of visualsemantic features, and how it can be effectively extracted for video analysis, still remains a challenging task. In this study, we propose a novel visual-semantic method to capture both appearance and dynamic representations. In particular, we propose a spatial context method, based on the fractional Fisher vector (FV) encoding on CNN features, which can be regarded as our main contribution. In addition, to capture temporal context information, we also applied fractional encoding method on dynamic images. Experimental results on the WWW crowed video dataset demonstrate that the proposed method outperform the state of the art.

Automated Critical Test Findings Identification and Online Notification System Using Artificial Intelligence in Imaging.

PubMed

Prevedello, Luciano M; Erdal, Barbaros S; Ryu, John L; Little, Kevin J; Demirer, Mutlu; Qian, Songyue; White, Richard D

2017-12-01

Purpose To evaluate the performance of an artificial intelligence (AI) tool using a deep learning algorithm for detecting hemorrhage, mass effect, or hydrocephalus (HMH) at non-contrast material-enhanced head computed tomographic (CT) examinations and to determine algorithm performance for detection of suspected acute infarct (SAI). Materials and Methods This HIPAA-compliant retrospective study was completed after institutional review board approval. A training and validation dataset of noncontrast-enhanced head CT examinations that comprised 100 examinations of HMH, 22 of SAI, and 124 of noncritical findings was obtained resulting in 2583 representative images. Examinations were processed by using a convolutional neural network (deep learning) using two different window and level configurations (brain window and stroke window). AI algorithm performance was tested on a separate dataset containing 50 examinations with HMH findings, 15 with SAI findings, and 35 with noncritical findings. Results Final algorithm performance for HMH showed 90% (45 of 50) sensitivity (95% confidence interval [CI]: 78%, 97%) and 85% (68 of 80) specificity (95% CI: 76%, 92%), with area under the receiver operating characteristic curve (AUC) of 0.91 with the brain window. For SAI, the best performance was achieved with the stroke window showing 62% (13 of 21) sensitivity (95% CI: 38%, 82%) and 96% (27 of 28) specificity (95% CI: 82%, 100%), with AUC of 0.81. Conclusion AI using deep learning demonstrates promise for detecting critical findings at noncontrast-enhanced head CT. A dedicated algorithm was required to detect SAI. Detection of SAI showed lower sensitivity in comparison to detection of HMH, but showed reasonable performance. Findings support further investigation of the algorithm in a controlled and prospective clinical setting to determine whether it can independently screen noncontrast-enhanced head CT examinations and notify the interpreting radiologist of critical findings. © RSNA, 2017 Online supplemental material is available for this article.

Automatic bladder segmentation from CT images using deep CNN and 3D fully connected CRF-RNN.

PubMed

Xu, Xuanang; Zhou, Fugen; Liu, Bo

2018-03-19

Automatic approach for bladder segmentation from computed tomography (CT) images is highly desirable in clinical practice. It is a challenging task since the bladder usually suffers large variations of appearance and low soft-tissue contrast in CT images. In this study, we present a deep learning-based approach which involves a convolutional neural network (CNN) and a 3D fully connected conditional random fields recurrent neural network (CRF-RNN) to perform accurate bladder segmentation. We also propose a novel preprocessing method, called dual-channel preprocessing, to further advance the segmentation performance of our approach. The presented approach works as following: first, we apply our proposed preprocessing method on the input CT image and obtain a dual-channel image which consists of the CT image and an enhanced bladder density map. Second, we exploit a CNN to predict a coarse voxel-wise bladder score map on this dual-channel image. Finally, a 3D fully connected CRF-RNN refines the coarse bladder score map and produce final fine-localized segmentation result. We compare our approach to the state-of-the-art V-net on a clinical dataset. Results show that our approach achieves superior segmentation accuracy, outperforming the V-net by a significant margin. The Dice Similarity Coefficient of our approach (92.24%) is 8.12% higher than that of the V-net. Moreover, the bladder probability maps performed by our approach present sharper boundaries and more accurate localizations compared with that of the V-net. Our approach achieves higher segmentation accuracy than the state-of-the-art method on clinical data. Both the dual-channel processing and the 3D fully connected CRF-RNN contribute to this improvement. The united deep network composed of the CNN and 3D CRF-RNN also outperforms a system where the CRF model acts as a post-processing method disconnected from the CNN.

Pubface: Celebrity face identification based on deep learning

NASA Astrophysics Data System (ADS)

Ouanan, H.; Ouanan, M.; Aksasse, B.

2018-05-01

In this paper, we describe a new real time application called PubFace, which allows to recognize celebrities in public spaces by employs a new pose invariant face recognition deep neural network algorithm with an extremely low error rate. To build this application, we make the following contributions: firstly, we build a novel dataset with over five million faces labelled. Secondly, we fine tuning the deep convolutional neural network (CNN) VGG-16 architecture on our new dataset that we have built. Finally, we deploy this model on the Raspberry Pi 3 model B using the OpenCv dnn module (OpenCV 3.3).

A deep convolutional neural network-based automatic delineation strategy for multiple brain metastases stereotactic radiosurgery

PubMed Central

Stojadinovic, Strahinja; Hrycushko, Brian; Wardak, Zabi; Lau, Steven; Lu, Weiguo; Yan, Yulong; Jiang, Steve B.; Zhen, Xin; Timmerman, Robert; Nedzi, Lucien

2017-01-01

Accurate and automatic brain metastases target delineation is a key step for efficient and effective stereotactic radiosurgery (SRS) treatment planning. In this work, we developed a deep learning convolutional neural network (CNN) algorithm for segmenting brain metastases on contrast-enhanced T1-weighted magnetic resonance imaging (MRI) datasets. We integrated the CNN-based algorithm into an automatic brain metastases segmentation workflow and validated on both Multimodal Brain Tumor Image Segmentation challenge (BRATS) data and clinical patients' data. Validation on BRATS data yielded average DICE coefficients (DCs) of 0.75±0.07 in the tumor core and 0.81±0.04 in the enhancing tumor, which outperformed most techniques in the 2015 BRATS challenge. Segmentation results of patient cases showed an average of DCs 0.67±0.03 and achieved an area under the receiver operating characteristic curve of 0.98±0.01. The developed automatic segmentation strategy surpasses current benchmark levels and offers a promising tool for SRS treatment planning for multiple brain metastases. PMID:28985229

An Unsupervised Deep Hyperspectral Anomaly Detector

PubMed Central

Ma, Ning; Peng, Yu; Wang, Shaojun

2018-01-01

Hyperspectral image (HSI) based detection has attracted considerable attention recently in agriculture, environmental protection and military applications as different wavelengths of light can be advantageously used to discriminate different types of objects. Unfortunately, estimating the background distribution and the detection of interesting local objects is not straightforward, and anomaly detectors may give false alarms. In this paper, a Deep Belief Network (DBN) based anomaly detector is proposed. The high-level features and reconstruction errors are learned through the network in a manner which is not affected by previous background distribution assumption. To reduce contamination by local anomalies, adaptive weights are constructed from reconstruction errors and statistical information. By using the code image which is generated during the inference of DBN and modified by adaptively updated weights, a local Euclidean distance between under test pixels and their neighboring pixels is used to determine the anomaly targets. Experimental results on synthetic and recorded HSI datasets show the performance of proposed method outperforms the classic global Reed-Xiaoli detector (RXD), local RX detector (LRXD) and the-state-of-the-art Collaborative Representation detector (CRD). PMID:29495410

Deep Spatial-Temporal Joint Feature Representation for Video Object Detection.

PubMed

Zhao, Baojun; Zhao, Boya; Tang, Linbo; Han, Yuqi; Wang, Wenzheng

2018-03-04

With the development of deep neural networks, many object detection frameworks have shown great success in the fields of smart surveillance, self-driving cars, and facial recognition. However, the data sources are usually videos, and the object detection frameworks are mostly established on still images and only use the spatial information, which means that the feature consistency cannot be ensured because the training procedure loses temporal information. To address these problems, we propose a single, fully-convolutional neural network-based object detection framework that involves temporal information by using Siamese networks. In the training procedure, first, the prediction network combines the multiscale feature map to handle objects of various sizes. Second, we introduce a correlation loss by using the Siamese network, which provides neighboring frame features. This correlation loss represents object co-occurrences across time to aid the consistent feature generation. Since the correlation loss should use the information of the track ID and detection label, our video object detection network has been evaluated on the large-scale ImageNet VID dataset where it achieves a 69.5% mean average precision (mAP).

Multiple Active Contours Guided by Differential Evolution for Medical Image Segmentation

PubMed Central

Cruz-Aceves, I.; Avina-Cervantes, J. G.; Lopez-Hernandez, J. M.; Rostro-Gonzalez, H.; Garcia-Capulin, C. H.; Torres-Cisneros, M.; Guzman-Cabrera, R.

2013-01-01

This paper presents a new image segmentation method based on multiple active contours guided by differential evolution, called MACDE. The segmentation method uses differential evolution over a polar coordinate system to increase the exploration and exploitation capabilities regarding the classical active contour model. To evaluate the performance of the proposed method, a set of synthetic images with complex objects, Gaussian noise, and deep concavities is introduced. Subsequently, MACDE is applied on datasets of sequential computed tomography and magnetic resonance images which contain the human heart and the human left ventricle, respectively. Finally, to obtain a quantitative and qualitative evaluation of the medical image segmentations compared to regions outlined by experts, a set of distance and similarity metrics has been adopted. According to the experimental results, MACDE outperforms the classical active contour model and the interactive Tseng method in terms of efficiency and robustness for obtaining the optimal control points and attains a high accuracy segmentation. PMID:23983809

Crowdsourcing the creation of image segmentation algorithms for connectomics.

PubMed

Arganda-Carreras, Ignacio; Turaga, Srinivas C; Berger, Daniel R; Cireşan, Dan; Giusti, Alessandro; Gambardella, Luca M; Schmidhuber, Jürgen; Laptev, Dmitry; Dwivedi, Sarvesh; Buhmann, Joachim M; Liu, Ting; Seyedhosseini, Mojtaba; Tasdizen, Tolga; Kamentsky, Lee; Burget, Radim; Uher, Vaclav; Tan, Xiao; Sun, Changming; Pham, Tuan D; Bas, Erhan; Uzunbas, Mustafa G; Cardona, Albert; Schindelin, Johannes; Seung, H Sebastian

2015-01-01

To stimulate progress in automating the reconstruction of neural circuits, we organized the first international challenge on 2D segmentation of electron microscopic (EM) images of the brain. Participants submitted boundary maps predicted for a test set of images, and were scored based on their agreement with a consensus of human expert annotations. The winning team had no prior experience with EM images, and employed a convolutional network. This "deep learning" approach has since become accepted as a standard for segmentation of EM images. The challenge has continued to accept submissions, and the best so far has resulted from cooperation between two teams. The challenge has probably saturated, as algorithms cannot progress beyond limits set by ambiguities inherent in 2D scoring and the size of the test dataset. Retrospective evaluation of the challenge scoring system reveals that it was not sufficiently robust to variations in the widths of neurite borders. We propose a solution to this problem, which should be useful for a future 3D segmentation challenge.

Digital mammographic tumor classification using transfer learning from deep convolutional neural networks.

PubMed

Huynh, Benjamin Q; Li, Hui; Giger, Maryellen L

2016-07-01

Convolutional neural networks (CNNs) show potential for computer-aided diagnosis (CADx) by learning features directly from the image data instead of using analytically extracted features. However, CNNs are difficult to train from scratch for medical images due to small sample sizes and variations in tumor presentations. Instead, transfer learning can be used to extract tumor information from medical images via CNNs originally pretrained for nonmedical tasks, alleviating the need for large datasets. Our database includes 219 breast lesions (607 full-field digital mammographic images). We compared support vector machine classifiers based on the CNN-extracted image features and our prior computer-extracted tumor features in the task of distinguishing between benign and malignant breast lesions. Five-fold cross validation (by lesion) was conducted with the area under the receiver operating characteristic (ROC) curve as the performance metric. Results show that classifiers based on CNN-extracted features (with transfer learning) perform comparably to those using analytically extracted features [area under the ROC curve [Formula: see text

Brain tumor segmentation using holistically nested neural networks in MRI images.

PubMed

Zhuge, Ying; Krauze, Andra V; Ning, Holly; Cheng, Jason Y; Arora, Barbara C; Camphausen, Kevin; Miller, Robert W

2017-10-01

Gliomas are rapidly progressive, neurologically devastating, largely fatal brain tumors. Magnetic resonance imaging (MRI) is a widely used technique employed in the diagnosis and management of gliomas in clinical practice. MRI is also the standard imaging modality used to delineate the brain tumor target as part of treatment planning for the administration of radiation therapy. Despite more than 20 yr of research and development, computational brain tumor segmentation in MRI images remains a challenging task. We are presenting a novel method of automatic image segmentation based on holistically nested neural networks that could be employed for brain tumor segmentation of MRI images. Two preprocessing techniques were applied to MRI images. The N4ITK method was employed for correction of bias field distortion. A novel landmark-based intensity normalization method was developed so that tissue types have a similar intensity scale in images of different subjects for the same MRI protocol. The holistically nested neural networks (HNN), which extend from the convolutional neural networks (CNN) with a deep supervision through an additional weighted-fusion output layer, was trained to learn the multiscale and multilevel hierarchical appearance representation of the brain tumor in MRI images and was subsequently applied to produce a prediction map of the brain tumor on test images. Finally, the brain tumor was obtained through an optimum thresholding on the prediction map. The proposed method was evaluated on both the Multimodal Brain Tumor Image Segmentation (BRATS) Benchmark 2013 training datasets, and clinical data from our institute. A dice similarity coefficient (DSC) and sensitivity of 0.78 and 0.81 were achieved on 20 BRATS 2013 training datasets with high-grade gliomas (HGG), based on a two-fold cross-validation. The HNN model built on the BRATS 2013 training data was applied to ten clinical datasets with HGG from a locally developed database. DSC and sensitivity of 0.83 and 0.85 were achieved. A quantitative comparison indicated that the proposed method outperforms the popular fully convolutional network (FCN) method. In terms of efficiency, the proposed method took around 10 h for training with 50,000 iterations, and approximately 30 s for testing of a typical MRI image in the BRATS 2013 dataset with a size of 160 × 216 × 176, using a DELL PRECISION workstation T7400, with an NVIDIA Tesla K20c GPU. An effective brain tumor segmentation method for MRI images based on a HNN has been developed. The high level of accuracy and efficiency make this method practical in brain tumor segmentation. It may play a crucial role in both brain tumor diagnostic analysis and in the treatment planning of radiation therapy. Published 2017. This article is a U.S. Government work and is in the public domain in the USA.

Transferring x-ray based automated threat detection between scanners with different energies and resolution

NASA Astrophysics Data System (ADS)

Caldwell, M.; Ransley, M.; Rogers, T. W.; Griffin, L. D.

2017-10-01

A significant obstacle to developing high performance Deep Learning algorithms for Automated Threat Detection (ATD) in security X-ray imagery, is the difficulty of obtaining large training datasets. In our previous work, we circumvented this problem for ATD in cargo containers, using Threat Image Projection and data augmentation. In this work, we investigate whether data scarcity for other modalities, such as parcels and baggage, can be ameliorated by transforming data from one domain so that it approximates the appearance of another. We present an ontology of ATD datasets to assess where transfer learning may be applied. We define frameworks for transfer at the training and testing stages, and compare the results for both methods against ATD where a common data source is used for training and testing. Our results show very poor transfer, which we attribute to the difficulty of accurately matching the blur and contrast characteristics of different scanners.

Deep learning for the detection of barchan dunes in satellite images

NASA Astrophysics Data System (ADS)

Azzaoui, A. M.; Adnani, M.; Elbelrhiti, H.; Chaouki, B. E. K.; Masmoudi, L.

2017-12-01

Barchan dunes are known to be the fastest moving sand dunes in deserts as they form under unidirectional winds and limited sand supply over a firm coherent basement (Elbelrhiti and Hargitai,2015). They were studied in the context of natural hazard monitoring as they could be a threat to human activities and infrastructures. Also, they were studied as a natural phenomenon occurring in other planetary landforms such as Mars or Venus (Bourke et al., 2010). Our region of interest was located in a desert region in the south of Morocco, in a barchan dunes corridor next to the town of Tarfaya. This region which is part of the Sahara desert contained thousands of barchans; which limits the number of dunes that could be studied during field missions. Therefore, we chose to monitor barchan dunes with satellite imagery, which can be seen as a complementary approach to field missions. We collected data from the Sentinel platform (https://scihub.copernicus.eu/dhus/); we used a machine learning method as a basis for the detection of barchan dunes positions in the satellite image. We trained a deep learning model on a mid-sized dataset that contained blocks representing images of barchan dunes, and images of other desert features, that we collected by cropping and annotating the source image. During testing, we browsed the satellite image with a gliding window that evaluated each block, and then produced a probability map. Finally, a threshold on the latter map exposed the location of barchan dunes. We used a subsample of data to train the model and we gradually incremented the size of the training set to get finer results and avoid over fitting. The positions of barchan dunes were successfully detected and deep learning was an effective method for this application. Sentinel-2 images were chosen for their availability and good temporal resolution, which will allow the tracking of barchan dunes in future work. While Sentinel images had sufficient spatial resolution for the detection of mid-size to large size barchans, we noted that it was relatively difficult to detect smaller barchan dunes. Overall, deep learning allowed us to achieve a high accuracy in the detection of barchan dunes. The tracking of hundreds of barchans using this detection method would provide an insight into the understanding of the dynamics of this natural phenomenon.

[Spatial domain display for interference image dataset].

PubMed

Wang, Cai-Ling; Li, Yu-Shan; Liu, Xue-Bin; Hu, Bing-Liang; Jing, Juan-Juan; Wen, Jia

2011-11-01

The requirements of imaging interferometer visualization is imminent for the user of image interpretation and information extraction. However, the conventional researches on visualization only focus on the spectral image dataset in spectral domain. Hence, the quick show of interference spectral image dataset display is one of the nodes in interference image processing. The conventional visualization of interference dataset chooses classical spectral image dataset display method after Fourier transformation. In the present paper, the problem of quick view of interferometer imager in image domain is addressed and the algorithm is proposed which simplifies the matter. The Fourier transformation is an obstacle since its computation time is very large and the complexion would be even deteriorated with the size of dataset increasing. The algorithm proposed, named interference weighted envelopes, makes the dataset divorced from transformation. The authors choose three interference weighted envelopes respectively based on the Fourier transformation, features of interference data and human visual system. After comparing the proposed with the conventional methods, the results show the huge difference in display time.

An introduction to deep learning on biological sequence data: examples and solutions.

PubMed

Jurtz, Vanessa Isabell; Johansen, Alexander Rosenberg; Nielsen, Morten; Almagro Armenteros, Jose Juan; Nielsen, Henrik; Sønderby, Casper Kaae; Winther, Ole; Sønderby, Søren Kaae

2017-11-15

Deep neural network architectures such as convolutional and long short-term memory networks have become increasingly popular as machine learning tools during the recent years. The availability of greater computational resources, more data, new algorithms for training deep models and easy to use libraries for implementation and training of neural networks are the drivers of this development. The use of deep learning has been especially successful in image recognition; and the development of tools, applications and code examples are in most cases centered within this field rather than within biology. Here, we aim to further the development of deep learning methods within biology by providing application examples and ready to apply and adapt code templates. Given such examples, we illustrate how architectures consisting of convolutional and long short-term memory neural networks can relatively easily be designed and trained to state-of-the-art performance on three biological sequence problems: prediction of subcellular localization, protein secondary structure and the binding of peptides to MHC Class II molecules. All implementations and datasets are available online to the scientific community at https://github.com/vanessajurtz/lasagne4bio. skaaesonderby@gmail.com. Supplementary data are available at Bioinformatics online. © The Author (2017). Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

Quantum-assisted Helmholtz machines: A quantum–classical deep learning framework for industrial datasets in near-term devices

NASA Astrophysics Data System (ADS)

Benedetti, Marcello; Realpe-Gómez, John; Perdomo-Ortiz, Alejandro

2018-07-01

Machine learning has been presented as one of the key applications for near-term quantum technologies, given its high commercial value and wide range of applicability. In this work, we introduce the quantum-assisted Helmholtz machine:a hybrid quantum–classical framework with the potential of tackling high-dimensional real-world machine learning datasets on continuous variables. Instead of using quantum computers only to assist deep learning, as previous approaches have suggested, we use deep learning to extract a low-dimensional binary representation of data, suitable for processing on relatively small quantum computers. Then, the quantum hardware and deep learning architecture work together to train an unsupervised generative model. We demonstrate this concept using 1644 quantum bits of a D-Wave 2000Q quantum device to model a sub-sampled version of the MNIST handwritten digit dataset with 16 × 16 continuous valued pixels. Although we illustrate this concept on a quantum annealer, adaptations to other quantum platforms, such as ion-trap technologies or superconducting gate-model architectures, could be explored within this flexible framework.

Autofocusing in digital holography using deep learning

NASA Astrophysics Data System (ADS)

Ren, Zhenbo; Xu, Zhimin; Lam, Edmund Y.

2018-02-01

In digital holography, it is critical to know the distance in order to reconstruct the multi-sectional object. This autofocusing is traditionally solved by reconstructing a stack of in-focus and out-of-focus images and using some focus metric, such as entropy or variance, to calculate the sharpness of each reconstructed image. Then the distance corresponding to the sharpest image is determined as the focal position. This method is effective but computationally demanding and time-consuming. To get an accurate estimation, one has to reconstruct many images. Sometimes after a coarse search, a refinement is needed. To overcome this problem in autofocusing, we propose to use deep learning, i.e., a convolutional neural network (CNN), to solve this problem. Autofocusing is viewed as a classification problem, in which the true distance is transferred as a label. To estimate the distance is equated to labeling a hologram correctly. To train such an algorithm, totally 1000 holograms are captured under the same environment, i.e., exposure time, incident angle, object, except the distance. There are 5 labels corresponding to 5 distances. These data are randomly split into three datasets to train, validate and test a CNN network. Experimental results show that the trained network is capable of predicting the distance without reconstructing or knowing any physical parameters about the setup. The prediction time using this method is far less than traditional autofocusing methods.

Deep learning applications for predicting pharmacological properties of drugs and drug repurposing using transcriptomic data

PubMed Central

Aliper, Alexander; Plis, Sergey; Artemov, Artem; Ulloa, Alvaro; Mamoshina, Polina; Zhavoronkov, Alex

2016-01-01

Deep learning is rapidly advancing many areas of science and technology with multiple success stories in image, text, voice and video recognition, robotics and autonomous driving. In this paper we demonstrate how deep neural networks (DNN) trained on large transcriptional response data sets can classify various drugs to therapeutic categories solely based on their transcriptional profiles. We used the perturbation samples of 678 drugs across A549, MCF‐7 and PC‐3 cell lines from the LINCS project and linked those to 12 therapeutic use categories derived from MeSH. To train the DNN, we utilized both gene level transcriptomic data and transcriptomic data processed using a pathway activation scoring algorithm, for a pooled dataset of samples perturbed with different concentrations of the drug for 6 and 24 hours. In both gene and pathway level classification, DNN convincingly outperformed support vector machine (SVM) model on every multiclass classification problem, however, models based on a pathway level classification perform better. For the first time we demonstrate a deep learning neural net trained on transcriptomic data to recognize pharmacological properties of multiple drugs across different biological systems and conditions. We also propose using deep neural net confusion matrices for drug repositioning. This work is a proof of principle for applying deep learning to drug discovery and development. PMID:27200455

Speaker emotion recognition: from classical classifiers to deep neural networks

NASA Astrophysics Data System (ADS)

Mezghani, Eya; Charfeddine, Maha; Nicolas, Henri; Ben Amar, Chokri

2018-04-01

Speaker emotion recognition is considered among the most challenging tasks in recent years. In fact, automatic systems for security, medicine or education can be improved when considering the speech affective state. In this paper, a twofold approach for speech emotion classification is proposed. At the first side, a relevant set of features is adopted, and then at the second one, numerous supervised training techniques, involving classic methods as well as deep learning, are experimented. Experimental results indicate that deep architecture can improve classification performance on two affective databases, the Berlin Dataset of Emotional Speech and the SAVEE Dataset Surrey Audio-Visual Expressed Emotion.

Global detection approach for clustered microcalcifications in mammograms using a deep learning network.

PubMed

Wang, Juan; Nishikawa, Robert M; Yang, Yongyi

2017-04-01

In computerized detection of clustered microcalcifications (MCs) from mammograms, the traditional approach is to apply a pattern detector to locate the presence of individual MCs, which are subsequently grouped into clusters. Such an approach is often susceptible to the occurrence of false positives (FPs) caused by local image patterns that resemble MCs. We investigate the feasibility of a direct detection approach to determining whether an image region contains clustered MCs or not. Toward this goal, we develop a deep convolutional neural network (CNN) as the classifier model to which the input consists of a large image window ([Formula: see text] in size). The multiple layers in the CNN classifier are trained to automatically extract image features relevant to MCs at different spatial scales. In the experiments, we demonstrated this approach on a dataset consisting of both screen-film mammograms and full-field digital mammograms. We evaluated the detection performance both on classifying image regions of clustered MCs using a receiver operating characteristic (ROC) analysis and on detecting clustered MCs from full mammograms by a free-response receiver operating characteristic analysis. For comparison, we also considered a recently developed MC detector with FP suppression. In classifying image regions of clustered MCs, the CNN classifier achieved 0.971 in the area under the ROC curve, compared to 0.944 for the MC detector. In detecting clustered MCs from full mammograms, at 90% sensitivity, the CNN classifier obtained an FP rate of 0.69 clusters/image, compared to 1.17 clusters/image by the MC detector. These results indicate that using global image features can be more effective in discriminating clustered MCs from FPs caused by various sources, such as linear structures, thereby providing a more accurate detection of clustered MCs on mammograms.

Hierarchical Context Modeling for Video Event Recognition.

PubMed

Wang, Xiaoyang; Ji, Qiang

2016-10-11

Current video event recognition research remains largely target-centered. For real-world surveillance videos, targetcentered event recognition faces great challenges due to large intra-class target variation, limited image resolution, and poor detection and tracking results. To mitigate these challenges, we introduced a context-augmented video event recognition approach. Specifically, we explicitly capture different types of contexts from three levels including image level, semantic level, and prior level. At the image level, we introduce two types of contextual features including the appearance context features and interaction context features to capture the appearance of context objects and their interactions with the target objects. At the semantic level, we propose a deep model based on deep Boltzmann machine to learn event object representations and their interactions. At the prior level, we utilize two types of prior-level contexts including scene priming and dynamic cueing. Finally, we introduce a hierarchical context model that systematically integrates the contextual information at different levels. Through the hierarchical context model, contexts at different levels jointly contribute to the event recognition. We evaluate the hierarchical context model for event recognition on benchmark surveillance video datasets. Results show that incorporating contexts in each level can improve event recognition performance, and jointly integrating three levels of contexts through our hierarchical model achieves the best performance.

Joint deep shape and appearance learning: application to optic pathway glioma segmentation

NASA Astrophysics Data System (ADS)

Mansoor, Awais; Li, Ien; Packer, Roger J.; Avery, Robert A.; Linguraru, Marius George

2017-03-01

Automated tissue characterization is one of the major applications of computer-aided diagnosis systems. Deep learning techniques have recently demonstrated impressive performance for the image patch-based tissue characterization. However, existing patch-based tissue classification techniques struggle to exploit the useful shape information. Local and global shape knowledge such as the regional boundary changes, diameter, and volumetrics can be useful in classifying the tissues especially in scenarios where the appearance signature does not provide significant classification information. In this work, we present a deep neural network-based method for the automated segmentation of the tumors referred to as optic pathway gliomas (OPG) located within the anterior visual pathway (AVP; optic nerve, chiasm or tracts) using joint shape and appearance learning. Voxel intensity values of commonly used MRI sequences are generally not indicative of OPG. To be considered an OPG, current clinical practice dictates that some portion of AVP must demonstrate shape enlargement. The method proposed in this work integrates multiple sequence magnetic resonance image (T1, T2, and FLAIR) along with local boundary changes to train a deep neural network. For training and evaluation purposes, we used a dataset of multiple sequence MRI obtained from 20 subjects (10 controls, 10 NF1+OPG). To our best knowledge, this is the first deep representation learning-based approach designed to merge shape and multi-channel appearance data for the glioma detection. In our experiments, mean misclassification errors of 2:39% and 0:48% were observed respectively for glioma and control patches extracted from the AVP. Moreover, an overall dice similarity coefficient of 0:87+/-0:13 (0:93+/-0:06 for healthy tissue, 0:78+/-0:18 for glioma tissue) demonstrates the potential of the proposed method in the accurate localization and early detection of OPG.

Automatic lung nodule graph cuts segmentation with deep learning false positive reduction

NASA Astrophysics Data System (ADS)

Sun, Wenqing; Huang, Xia; Tseng, Tzu-Liang Bill; Qian, Wei

2017-03-01

To automatic detect lung nodules from CT images, we designed a two stage computer aided detection (CAD) system. The first stage is graph cuts segmentation to identify and segment the nodule candidates, and the second stage is convolutional neural network for false positive reduction. The dataset contains 595 CT cases randomly selected from Lung Image Database Consortium and Image Database Resource Initiative (LIDC/IDRI) and the 305 pulmonary nodules achieved diagnosis consensus by all four experienced radiologists were our detection targets. Consider each slice as an individual sample, 2844 nodules were included in our database. The graph cuts segmentation was conducted in a two-dimension manner, 2733 lung nodule ROIs are successfully identified and segmented. With a false positive reduction by a seven-layer convolutional neural network, 2535 nodules remain detected while the false positive dropped to 31.6%. The average F-measure of segmented lung nodule tissue is 0.8501.

Influence of Domain Shift Factors on Deep Segmentation of the Drivable Path of AN Autonomous Vehicle

NASA Astrophysics Data System (ADS)

Bormans, R. P. A.; Lindenbergh, R. C.; Karimi Nejadasl, F.

2018-05-01

One of the biggest challenges for an autonomous vehicle (and hence the WEpod) is to see the world as humans would see it. This understanding is the base for a successful and reliable future of autonomous vehicles. Real-world data and semantic segmentation generally are used to achieve full understanding of its surroundings. However, deploying a pretrained segmentation network to a new, previously unseen domain will not attain similar performance as it would on the domain where it is trained on due to the differences between the domains. Although research is done concerning the mitigation of this domain shift, the factors that cause these differences are not yet fully explored. We filled this gap with the investigation of several factors. A base network was created by a two-step finetuning procedure on a convolutional neural network (SegNet) which is pretrained on CityScapes (a dataset for semantic segmentation). The first tuning step is based on RobotCar (road scenery dataset recorded in Oxford, UK) while afterwards this network is fine-tuned for a second time but now on the KITTI (road scenery dataset recorded in Germany) dataset. With this base, experiments are used to obtain the importance of factors such as horizon line, colour and training order for a successful domain adaptation. In this case the domain adaptation is from the KITTI and RobotCar domain to the WEpod domain. For evaluation, groundtruth labels are created in a weakly-supervised setting. Negative influence was obtained for training on greyscale images instead of RGB images. This resulted in drops of IoU values up to 23.9 % for WEpod test images. The training order is a main contributor for domain adaptation with an increase in IoU of 4.7 %. This shows that the target domain (WEpod) is more closely related to RobotCar than to KITTI.

Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes.

PubMed

Ting, Daniel Shu Wei; Cheung, Carol Yim-Lui; Lim, Gilbert; Tan, Gavin Siew Wei; Quang, Nguyen D; Gan, Alfred; Hamzah, Haslina; Garcia-Franco, Renata; San Yeo, Ian Yew; Lee, Shu Yen; Wong, Edmund Yick Mun; Sabanayagam, Charumathi; Baskaran, Mani; Ibrahim, Farah; Tan, Ngiap Chuan; Finkelstein, Eric A; Lamoureux, Ecosse L; Wong, Ian Y; Bressler, Neil M; Sivaprasad, Sobha; Varma, Rohit; Jonas, Jost B; He, Ming Guang; Cheng, Ching-Yu; Cheung, Gemmy Chui Ming; Aung, Tin; Hsu, Wynne; Lee, Mong Li; Wong, Tien Yin

2017-12-12

A deep learning system (DLS) is a machine learning technology with potential for screening diabetic retinopathy and related eye diseases. To evaluate the performance of a DLS in detecting referable diabetic retinopathy, vision-threatening diabetic retinopathy, possible glaucoma, and age-related macular degeneration (AMD) in community and clinic-based multiethnic populations with diabetes. Diagnostic performance of a DLS for diabetic retinopathy and related eye diseases was evaluated using 494 661 retinal images. A DLS was trained for detecting diabetic retinopathy (using 76 370 images), possible glaucoma (125 189 images), and AMD (72 610 images), and performance of DLS was evaluated for detecting diabetic retinopathy (using 112 648 images), possible glaucoma (71 896 images), and AMD (35 948 images). Training of the DLS was completed in May 2016, and validation of the DLS was completed in May 2017 for detection of referable diabetic retinopathy (moderate nonproliferative diabetic retinopathy or worse) and vision-threatening diabetic retinopathy (severe nonproliferative diabetic retinopathy or worse) using a primary validation data set in the Singapore National Diabetic Retinopathy Screening Program and 10 multiethnic cohorts with diabetes. Use of a deep learning system. Area under the receiver operating characteristic curve (AUC) and sensitivity and specificity of the DLS with professional graders (retinal specialists, general ophthalmologists, trained graders, or optometrists) as the reference standard. In the primary validation dataset (n = 14 880 patients; 71 896 images; mean [SD] age, 60.2 [2.2] years; 54.6% men), the prevalence of referable diabetic retinopathy was 3.0%; vision-threatening diabetic retinopathy, 0.6%; possible glaucoma, 0.1%; and AMD, 2.5%. The AUC of the DLS for referable diabetic retinopathy was 0.936 (95% CI, 0.925-0.943), sensitivity was 90.5% (95% CI, 87.3%-93.0%), and specificity was 91.6% (95% CI, 91.0%-92.2%). For vision-threatening diabetic retinopathy, AUC was 0.958 (95% CI, 0.956-0.961), sensitivity was 100% (95% CI, 94.1%-100.0%), and specificity was 91.1% (95% CI, 90.7%-91.4%). For possible glaucoma, AUC was 0.942 (95% CI, 0.929-0.954), sensitivity was 96.4% (95% CI, 81.7%-99.9%), and specificity was 87.2% (95% CI, 86.8%-87.5%). For AMD, AUC was 0.931 (95% CI, 0.928-0.935), sensitivity was 93.2% (95% CI, 91.1%-99.8%), and specificity was 88.7% (95% CI, 88.3%-89.0%). For referable diabetic retinopathy in the 10 additional datasets, AUC range was 0.889 to 0.983 (n = 40 752 images). In this evaluation of retinal images from multiethnic cohorts of patients with diabetes, the DLS had high sensitivity and specificity for identifying diabetic retinopathy and related eye diseases. Further research is necessary to evaluate the applicability of the DLS in health care settings and the utility of the DLS to improve vision outcomes.

Development and Validation of a Deep Learning System for Diabetic Retinopathy and Related Eye Diseases Using Retinal Images From Multiethnic Populations With Diabetes

PubMed Central

Ting, Daniel Shu Wei; Cheung, Carol Yim-Lui; Lim, Gilbert; Tan, Gavin Siew Wei; Quang, Nguyen D.; Gan, Alfred; Hamzah, Haslina; Garcia-Franco, Renata; San Yeo, Ian Yew; Lee, Shu Yen; Wong, Edmund Yick Mun; Sabanayagam, Charumathi; Baskaran, Mani; Ibrahim, Farah; Tan, Ngiap Chuan; Finkelstein, Eric A.; Lamoureux, Ecosse L.; Wong, Ian Y.; Bressler, Neil M.; Sivaprasad, Sobha; Varma, Rohit; Jonas, Jost B.; He, Ming Guang; Cheng, Ching-Yu; Cheung, Gemmy Chui Ming; Aung, Tin; Hsu, Wynne; Lee, Mong Li

2017-01-01

Importance A deep learning system (DLS) is a machine learning technology with potential for screening diabetic retinopathy and related eye diseases. Objective To evaluate the performance of a DLS in detecting referable diabetic retinopathy, vision-threatening diabetic retinopathy, possible glaucoma, and age-related macular degeneration (AMD) in community and clinic-based multiethnic populations with diabetes. Design, Setting, and Participants Diagnostic performance of a DLS for diabetic retinopathy and related eye diseases was evaluated using 494 661 retinal images. A DLS was trained for detecting diabetic retinopathy (using 76 370 images), possible glaucoma (125 189 images), and AMD (72 610 images), and performance of DLS was evaluated for detecting diabetic retinopathy (using 112 648 images), possible glaucoma (71 896 images), and AMD (35 948 images). Training of the DLS was completed in May 2016, and validation of the DLS was completed in May 2017 for detection of referable diabetic retinopathy (moderate nonproliferative diabetic retinopathy or worse) and vision-threatening diabetic retinopathy (severe nonproliferative diabetic retinopathy or worse) using a primary validation data set in the Singapore National Diabetic Retinopathy Screening Program and 10 multiethnic cohorts with diabetes. Exposures Use of a deep learning system. Main Outcomes and Measures Area under the receiver operating characteristic curve (AUC) and sensitivity and specificity of the DLS with professional graders (retinal specialists, general ophthalmologists, trained graders, or optometrists) as the reference standard. Results In the primary validation dataset (n = 14 880 patients; 71 896 images; mean [SD] age, 60.2 [2.2] years; 54.6% men), the prevalence of referable diabetic retinopathy was 3.0%; vision-threatening diabetic retinopathy, 0.6%; possible glaucoma, 0.1%; and AMD, 2.5%. The AUC of the DLS for referable diabetic retinopathy was 0.936 (95% CI, 0.925-0.943), sensitivity was 90.5% (95% CI, 87.3%-93.0%), and specificity was 91.6% (95% CI, 91.0%-92.2%). For vision-threatening diabetic retinopathy, AUC was 0.958 (95% CI, 0.956-0.961), sensitivity was 100% (95% CI, 94.1%-100.0%), and specificity was 91.1% (95% CI, 90.7%-91.4%). For possible glaucoma, AUC was 0.942 (95% CI, 0.929-0.954), sensitivity was 96.4% (95% CI, 81.7%-99.9%), and specificity was 87.2% (95% CI, 86.8%-87.5%). For AMD, AUC was 0.931 (95% CI, 0.928-0.935), sensitivity was 93.2% (95% CI, 91.1%-99.8%), and specificity was 88.7% (95% CI, 88.3%-89.0%). For referable diabetic retinopathy in the 10 additional datasets, AUC range was 0.889 to 0.983 (n = 40 752 images). Conclusions and Relevance In this evaluation of retinal images from multiethnic cohorts of patients with diabetes, the DLS had high sensitivity and specificity for identifying diabetic retinopathy and related eye diseases. Further research is necessary to evaluate the applicability of the DLS in health care settings and the utility of the DLS to improve vision outcomes. PMID:29234807

Exploring the Earth Using Deep Learning Techniques

NASA Astrophysics Data System (ADS)

Larraondo, P. R.; Evans, B. J. K.; Antony, J.

2016-12-01

Research using deep neural networks have significantly matured in recent times, and there is now a surge in interest to apply such methods to Earth systems science and the geosciences. When combined with Big Data, we believe there are opportunities for significantly transforming a number of areas relevant to researchers and policy makers. In particular, by using a combination of data from a range of satellite Earth observations as well as computer simulations from climate models and reanalysis, we can gain new insights into the information that is locked within the data. Global geospatial datasets describe a wide range of physical and chemical parameters, which are mostly available using regular grids covering large spatial and temporal extents. This makes them perfect candidates to apply deep learning methods. So far, these techniques have been successfully applied to image analysis through the use of convolutional neural networks. However, this is only one field of interest, and there is potential for many more use cases to be explored. The deep learning algorithms require fast access to large amounts of data in the form of tensors and make intensive use of CPU in order to train its models. The Australian National Computational Infrastructure (NCI) has recently augmented its Raijin 1.2 PFlop supercomputer with hardware accelerators. Together with NCI's 3000 core high performance OpenStack cloud, these computational systems have direct access to NCI's 10+ PBytes of datasets and associated Big Data software technologies (see http://geonetwork.nci.org.au/ and http://nci.org.au/systems-services/national-facility/nerdip/). To effectively use these computing infrastructures requires that both the data and software are organised in a way that readily supports the deep learning software ecosystem. Deep learning software, such as the open source TensorFlow library, has allowed us to demonstrate the possibility of generating geospatial models by combining information from our different data sources. This opens the door to an exciting new way of generating products and extracting features that have previously been labour intensive. In this paper, we will explore some of these geospatial use cases and share some of the lessons learned from this experience.

Using deep learning to segment breast and fibroglandular tissue in MRI volumes.

PubMed

Dalmış, Mehmet Ufuk; Litjens, Geert; Holland, Katharina; Setio, Arnaud; Mann, Ritse; Karssemeijer, Nico; Gubern-Mérida, Albert

2017-02-01

Automated segmentation of breast and fibroglandular tissue (FGT) is required for various computer-aided applications of breast MRI. Traditional image analysis and computer vision techniques, such atlas, template matching, or, edge and surface detection, have been applied to solve this task. However, applicability of these methods is usually limited by the characteristics of the images used in the study datasets, while breast MRI varies with respect to the different MRI protocols used, in addition to the variability in breast shapes. All this variability, in addition to various MRI artifacts, makes it a challenging task to develop a robust breast and FGT segmentation method using traditional approaches. Therefore, in this study, we investigated the use of a deep-learning approach known as "U-net." We used a dataset of 66 breast MRI's randomly selected from our scientific archive, which includes five different MRI acquisition protocols and breasts from four breast density categories in a balanced distribution. To prepare reference segmentations, we manually segmented breast and FGT for all images using an in-house developed workstation. We experimented with the application of U-net in two different ways for breast and FGT segmentation. In the first method, following the same pipeline used in traditional approaches, we trained two consecutive (2C) U-nets: first for segmenting the breast in the whole MRI volume and the second for segmenting FGT inside the segmented breast. In the second method, we used a single 3-class (3C) U-net, which performs both tasks simultaneously by segmenting the volume into three regions: nonbreast, fat inside the breast, and FGT inside the breast. For comparison, we applied two existing and published methods to our dataset: an atlas-based method and a sheetness-based method. We used Dice Similarity Coefficient (DSC) to measure the performances of the automated methods, with respect to the manual segmentations. Additionally, we computed Pearson's correlation between the breast density values computed based on manual and automated segmentations. The average DSC values for breast segmentation were 0.933, 0.944, 0.863, and 0.848 obtained from 3C U-net, 2C U-nets, atlas-based method, and sheetness-based method, respectively. The average DSC values for FGT segmentation obtained from 3C U-net, 2C U-nets, and atlas-based methods were 0.850, 0.811, and 0.671, respectively. The correlation between breast density values based on 3C U-net and manual segmentations was 0.974. This value was significantly higher than 0.957 as obtained from 2C U-nets (P < 0.0001, Steiger's Z-test with Bonferoni correction) and 0.938 as obtained from atlas-based method (P = 0.0016). In conclusion, we applied a deep-learning method, U-net, for segmenting breast and FGT in MRI in a dataset that includes a variety of MRI protocols and breast densities. Our results showed that U-net-based methods significantly outperformed the existing algorithms and resulted in significantly more accurate breast density computation. © 2016 American Association of Physicists in Medicine.

Comparison of breast DCE-MRI contrast time points for predicting response to neoadjuvant chemotherapy using deep convolutional neural network features with transfer learning

NASA Astrophysics Data System (ADS)

Huynh, Benjamin Q.; Antropova, Natasha; Giger, Maryellen L.

2017-03-01

DCE-MRI datasets have a temporal aspect to them, resulting in multiple regions of interest (ROIs) per subject, based on contrast time points. It is unclear how the different contrast time points vary in terms of usefulness for computer-aided diagnosis tasks in conjunction with deep learning methods. We thus sought to compare the different DCE-MRI contrast time points with regard to how well their extracted features predict response to neoadjuvant chemotherapy within a deep convolutional neural network. Our dataset consisted of 561 ROIs from 64 subjects. Each subject was categorized as a non-responder or responder, determined by recurrence-free survival. First, features were extracted from each ROI using a convolutional neural network (CNN) pre-trained on non-medical images. Linear discriminant analysis classifiers were then trained on varying subsets of these features, based on their contrast time points of origin. Leave-one-out cross validation (by subject) was used to assess performance in the task of estimating probability of response to therapy, with area under the ROC curve (AUC) as the metric. The classifier trained on features from strictly the pre-contrast time point performed the best, with an AUC of 0.85 (SD = 0.033). The remaining classifiers resulted in AUCs ranging from 0.71 (SD = 0.028) to 0.82 (SD = 0.027). Overall, we found the pre-contrast time point to be the most effective at predicting response to therapy and that including additional contrast time points moderately reduces variance.

A predictive penetrative fracture mapping method from regional potential field and geologic datasets, southwest Colorado Plateau, U.S.A

USGS Publications Warehouse

Gettings, M.E.; Bultman, M.W.

2005-01-01

Some aquifers of the southwest Colorado Plateau, U.S.A., are deeply buried and overlain by several impermeable units, and thus recharge to the aquifer is probably mainly by seepage down penetrative fracture systems. This purpose of this study was to develop a method to map the location of candidate deep penetrative fractures over a 120,000 km2 area using gravity and aeromagnetic anomaly data together with surficial fracture data. The resulting database constitutes a spatially registered estimate of recharge location. Candidate deep fractures were obtained by spatial correlation of horizontal gradient and analytic signal maxima of gravity and magnetic anomalies vertically with major surficial lineaments obtained from geologic, topographic, side-looking airborne radar, and satellite imagery. The maps define a sub-set of possible penetrative fractures because of limitations of data coverage and the analysis technique. The data and techniques employed do not yield any indication as to whether fractures are open or closed. Correlations were carried out using image processing software in such a way that every pixel on the resulting grids was coded to uniquely identify which datasets correlated. The technique correctly identified known deep fracture systems and many new ones. Maps of the correlations also define in detail the tectonic fabrics of the Southwestern Colorado Plateau. Copyright ?? The Society of Geomagnetism and Earth, Planetary and Space Sciences (SGEPSS); The Seismological Society of Japan; The Volcanological Society of Japan; The Geodetic Society of Japan; The Japanese Society for Planetary Sciences; TERRAPUB.

Seismic Imaging Reveals Deep-Penetrating Fault Planes in the Wharton Basin Oceanic Mantle

NASA Astrophysics Data System (ADS)

Carton, H. D.; Singh, S. C.; Dyment, J.; Hananto, N. D.; Chauhan, A.

2011-12-01

We present images from a deep multi-channel seismic reflection survey acquired in 2006 over the oceanic lithosphere of the Wharton Basin offshore northern Sumatra, NW of Simeulue island. The main ~230-km long seismic profile is roughly parallel to the trench at ~32-66 km distance from the subduction front and crosses (at oblique angles to both flow line and isochron directions) an entire segment of 55-57 my-old fast-spread crust formed at the extinct Wharton spreading center, as well as two bounding ~N5°E trending fracture zones near its extremities; complementary data is provided by the oceanic portions of two margin-crossing profiles on either side shot during the same survey. This high-quality, 12-km streamer dataset acquired for deep reflection imaging (10000 cu in tuned airgun array and 15-m source and streamer depths) reveals the presence of mostly SE-dipping (20 to 40 degrees dip) events cutting across and extending below the oceanic Moho, down to a maximum depth below seafloor of ~37 km, at ~5 km spacing along the trench-parallel profile. Similar dipping mantle events are imaged on the oceanic portion of another long-offset profile acquired in 2009 offshore central Sumatra south of Pagai island, which will also be presented. Such events are unlikely to be imaging artefacts of the 2D acquisition, such as out-of-plane energy originating from sharp, buried basement reliefs trending obliquely to the profile. Due to their geometry, they do not seem to be associated with plate bending at the trench outer-rise, which has a relatively modest expression at the seafloor and within the incoming sedimentary section north of the Simeulue elbow. We propose that these deep-penetrating dipping reflectors are fossil fault planes formed due to compressive stresses at the beginning of the continent-continent collision between India and Eurasia, the early stages of which were responsible for the cessation of seafloor spreading at the Wharton ridge at ca 40 Ma.

Development of levees on deep-sea channels: Insights from high-resolution AUV exploration of the Lucia Chica system, offshore central California

NASA Astrophysics Data System (ADS)

Maier, K. L.; Fildani, A.; Romans, B.; Paull, C. K.; McHargue, T.; Graham, S. A.; Caress, D. W.

2010-12-01

The Lucia Chica, a tributary channel system of the Lucia Canyon, offshore central California, was imaged using the Monterey Bay Aquarium Research Institute’s (MBARI) Autonomous Underwater Vehicle (AUV) in order to investigate seafloor and subsurface morphologies associated with low-relief submarine channels. In larger, previously investigated seafloor channel-levee systems, initial deposits are either eroded, compacted, or below the resolution of available imaging. In this dataset from the Lucia Chica, the unprecedented high-resolution multibeam bathymetry (1 m lateral resolution) and chirp sub-bottom profiles (11 cm vertical resolution) reveal a highly irregular seafloor with scours, depressions, and discontinuous low-relief conduits over an area of ~70 km2. Sediment packages associated with channels, levees, and deposits related to less confined flows are correlated between chirp profiles and with the multibeam bathymetric image to determine the stratigraphic evolution of the Lucia Chica and the sequence of channel-levee development. In the Lucia Chica, channels appear to have initiated as trains of scours that eventually coalesced into continuous channel thalwegs carved by erosional turbidity currents. Channel incision and stepped lateral migration led to the development of terraces, complex levee stratigraphy, and distinct morphologies associated with inner and outer bends of sinuous channels. The inner bend levee stratigraphy indicates that the channel position migrated in discrete shifts, as opposed to continuous channel migration associated with lateral accretion. Discrete levee packages, formed from flow-stripped turbidity currents, later infilled abandoned portions of the channel and overbank areas. While processes of initial channel and levee development are well established in fluvial settings, detailed examples are lacking for deep-sea systems. These results highlight the differences in initiation between submarine channel systems, their fluvial counterparts, and larger submarine channel-levee systems imaged only with lower-resolution technologies. High-resolution imaging and detailed mapping made possible by cutting-edge oceanographic technology provide an unprecedented examination of deep-water channel-levee morphology and improve understanding of deep-water channel migration and levee development.

A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans

NASA Astrophysics Data System (ADS)

Chen, Quan; Xu, Xiang; Hu, Shiliang; Li, Xiao; Zou, Qing; Li, Yunpeng

2017-03-01

Deep learning has shown a great potential in computer aided diagnosis. However, in many applications, large dataset is not available. This makes the training of a sophisticated deep learning neural network (DNN) difficult. In this study, we demonstrated that with transfer learning, we can quickly retrain start-of-the-art DNN models with limited data provided by the prostateX challenge. The training data consists of 330 lesions, only 78 were clinical significant. Efforts were made to balance the data during training. We used ImageNet pre-trained inceptionV3 and Vgg-16 model and obtained AUC of 0.81 and 0.83 respectively on the prostateX test data, good for a 4th place finish. We noticed that models trained for different prostate zone has different sensitivity. Applying scaling factors before merging the result improves the AUC for the final result.

Deep hierarchical attention network for video description

NASA Astrophysics Data System (ADS)

Li, Shuohao; Tang, Min; Zhang, Jun

2018-03-01

Pairing video to natural language description remains a challenge in computer vision and machine translation. Inspired by image description, which uses an encoder-decoder model for reducing visual scene into a single sentence, we propose a deep hierarchical attention network for video description. The proposed model uses convolutional neural network (CNN) and bidirectional LSTM network as encoders while a hierarchical attention network is used as the decoder. Compared to encoder-decoder models used in video description, the bidirectional LSTM network can capture the temporal structure among video frames. Moreover, the hierarchical attention network has an advantage over single-layer attention network on global context modeling. To make a fair comparison with other methods, we evaluate the proposed architecture with different types of CNN structures and decoders. Experimental results on the standard datasets show that our model has a more superior performance than the state-of-the-art techniques.

Drawing a baseline in aesthetic quality assessment

NASA Astrophysics Data System (ADS)

Rubio, Fernando; Flores, M. Julia; Puerta, Jose M.

2018-04-01

Aesthetic classification of images is an inherently subjective task. There does not exist a validated collection of images/photographs labeled as having good or bad quality from experts. Nowadays, the closest approximation to that is to use databases of photos where a group of users rate each image. Hence, there is not a unique good/bad label but a rating distribution given by users voting. Due to this peculiarity, it is not possible to state the problem of binary aesthetic supervised classification in such a direct mode as other Computer Vision tasks. Recent literature follows an approach where researchers utilize the average rates from the users for each image, and they establish an arbitrary threshold to determine their class or label. In this way, images above the threshold are considered of good quality, while images below the threshold are seen as bad quality. This paper analyzes current literature, and it reviews those attributes able to represent an image, differentiating into three families: specific, general and deep features. Among those which have been proved more competitive, we have selected a representative subset, being our main goal to establish a clear experimental framework. Finally, once features were selected, we have used them for the full AVA dataset. We have to remark that to perform validation we report not only accuracy values, which is not that informative in this case, but also, metrics able to evaluate classification power within imbalanced datasets. We have conducted a series of experiments so that distinct well-known classifiers are learned from data. Like that, this paper provides what we could consider valuable and valid baseline results for the given problem.

A hybrid MLP-CNN classifier for very fine resolution remotely sensed image classification

NASA Astrophysics Data System (ADS)

Zhang, Ce; Pan, Xin; Li, Huapeng; Gardiner, Andy; Sargent, Isabel; Hare, Jonathon; Atkinson, Peter M.

2018-06-01

The contextual-based convolutional neural network (CNN) with deep architecture and pixel-based multilayer perceptron (MLP) with shallow structure are well-recognized neural network algorithms, representing the state-of-the-art deep learning method and the classical non-parametric machine learning approach, respectively. The two algorithms, which have very different behaviours, were integrated in a concise and effective way using a rule-based decision fusion approach for the classification of very fine spatial resolution (VFSR) remotely sensed imagery. The decision fusion rules, designed primarily based on the classification confidence of the CNN, reflect the generally complementary patterns of the individual classifiers. In consequence, the proposed ensemble classifier MLP-CNN harvests the complementary results acquired from the CNN based on deep spatial feature representation and from the MLP based on spectral discrimination. Meanwhile, limitations of the CNN due to the adoption of convolutional filters such as the uncertainty in object boundary partition and loss of useful fine spatial resolution detail were compensated. The effectiveness of the ensemble MLP-CNN classifier was tested in both urban and rural areas using aerial photography together with an additional satellite sensor dataset. The MLP-CNN classifier achieved promising performance, consistently outperforming the pixel-based MLP, spectral and textural-based MLP, and the contextual-based CNN in terms of classification accuracy. This research paves the way to effectively address the complicated problem of VFSR image classification.

A patch-based convolutional neural network for remote sensing image classification.

PubMed

Sharma, Atharva; Liu, Xiuwen; Yang, Xiaojun; Shi, Di

2017-11-01

Availability of accurate land cover information over large areas is essential to the global environment sustainability; digital classification using medium-resolution remote sensing data would provide an effective method to generate the required land cover information. However, low accuracy of existing per-pixel based classification methods for medium-resolution data is a fundamental limiting factor. While convolutional neural networks (CNNs) with deep layers have achieved unprecedented improvements in object recognition applications that rely on fine image structures, they cannot be applied directly to medium-resolution data due to lack of such fine structures. In this paper, considering the spatial relation of a pixel to its neighborhood, we propose a new deep patch-based CNN system tailored for medium-resolution remote sensing data. The system is designed by incorporating distinctive characteristics of medium-resolution data; in particular, the system computes patch-based samples from multidimensional top of atmosphere reflectance data. With a test site from the Florida Everglades area (with a size of 771 square kilometers), the proposed new system has outperformed pixel-based neural network, pixel-based CNN and patch-based neural network by 24.36%, 24.23% and 11.52%, respectively, in overall classification accuracy. By combining the proposed deep CNN and the huge collection of medium-resolution remote sensing data, we believe that much more accurate land cover datasets can be produced over large areas. Copyright © 2017 Elsevier Ltd. All rights reserved.

Vehicle Color Recognition with Vehicle-Color Saliency Detection and Dual-Orientational Dimensionality Reduction of CNN Deep Features

NASA Astrophysics Data System (ADS)

Zhang, Qiang; Li, Jiafeng; Zhuo, Li; Zhang, Hui; Li, Xiaoguang

2017-12-01

Color is one of the most stable attributes of vehicles and often used as a valuable cue in some important applications. Various complex environmental factors, such as illumination, weather, noise and etc., result in the visual characteristics of the vehicle color being obvious diversity. Vehicle color recognition in complex environments has been a challenging task. The state-of-the-arts methods roughly take the whole image for color recognition, but many parts of the images such as car windows; wheels and background contain no color information, which will have negative impact on the recognition accuracy. In this paper, a novel vehicle color recognition method using local vehicle-color saliency detection and dual-orientational dimensionality reduction of convolutional neural network (CNN) deep features has been proposed. The novelty of the proposed method includes two parts: (1) a local vehicle-color saliency detection method has been proposed to determine the vehicle color region of the vehicle image and exclude the influence of non-color regions on the recognition accuracy; (2) dual-orientational dimensionality reduction strategy has been designed to greatly reduce the dimensionality of deep features that are learnt from CNN, which will greatly mitigate the storage and computational burden of the subsequent processing, while improving the recognition accuracy. Furthermore, linear support vector machine is adopted as the classifier to train the dimensionality reduced features to obtain the recognition model. The experimental results on public dataset demonstrate that the proposed method can achieve superior recognition performance over the state-of-the-arts methods.

Neural Networks for the Classification of Building Use from Street-View Imagery

NASA Astrophysics Data System (ADS)

Laupheimer, D.; Tutzauer, P.; Haala, N.; Spicker, M.

2018-05-01

Within this paper we propose an end-to-end approach for classifying terrestrial images of building facades into five different utility classes (commercial, hybrid, residential, specialUse, underConstruction) by using Convolutional Neural Networks (CNNs). For our examples we use images provided by Google Street View. These images are automatically linked to a coarse city model, including the outlines of the buildings as well as their respective use classes. By these means an extensive dataset is available for training and evaluation of our Deep Learning pipeline. The paper describes the implemented end-to-end approach for classifying street-level images of building facades and discusses our experiments with various CNNs. In addition to the classification results, so-called Class Activation Maps (CAMs) are evaluated. These maps give further insights into decisive facade parts that are learned as features during the training process. Furthermore, they can be used for the generation of abstract presentations which facilitate the comprehension of semantic image content. The abstract representations are a result of the stippling method, an importance-based image rendering.

Medical Image Data and Datasets in the Era of Machine Learning-Whitepaper from the 2016 C-MIMI Meeting Dataset Session.

PubMed

Kohli, Marc D; Summers, Ronald M; Geis, J Raymond

2017-08-01

At the first annual Conference on Machine Intelligence in Medical Imaging (C-MIMI), held in September 2016, a conference session on medical image data and datasets for machine learning identified multiple issues. The common theme from attendees was that everyone participating in medical image evaluation with machine learning is data starved. There is an urgent need to find better ways to collect, annotate, and reuse medical imaging data. Unique domain issues with medical image datasets require further study, development, and dissemination of best practices and standards, and a coordinated effort among medical imaging domain experts, medical imaging informaticists, government and industry data scientists, and interested commercial, academic, and government entities. High-level attributes of reusable medical image datasets suitable to train, test, validate, verify, and regulate ML products should be better described. NIH and other government agencies should promote and, where applicable, enforce, access to medical image datasets. We should improve communication among medical imaging domain experts, medical imaging informaticists, academic clinical and basic science researchers, government and industry data scientists, and interested commercial entities.

Fast DCNN based on FWT, intelligent dropout and layer skipping for image retrieval.

PubMed

ElAdel, Asma; Zaied, Mourad; Amar, Chokri Ben

2017-11-01

Deep Convolutional Neural Network (DCNN) can be marked as a powerful tool for object and image classification and retrieval. However, the training stage of such networks is highly consuming in terms of storage space and time. Also, the optimization is still a challenging subject. In this paper, we propose a fast DCNN based on Fast Wavelet Transform (FWT), intelligent dropout and layer skipping. The proposed approach led to improve the image retrieval accuracy as well as the searching time. This was possible thanks to three key advantages: First, the rapid way to compute the features using FWT. Second, the proposed intelligent dropout method is based on whether or not a unit is efficiently and not randomly selected. Third, it is possible to classify the image using efficient units of earlier layer(s) and skipping all the subsequent hidden layers directly to the output layer. Our experiments were performed on CIFAR-10 and MNIST datasets and the obtained results are very promising. Copyright © 2017 Elsevier Ltd. All rights reserved.

Representational Distance Learning for Deep Neural Networks

PubMed Central

McClure, Patrick; Kriegeskorte, Nikolaus

2016-01-01

Deep neural networks (DNNs) provide useful models of visual representational transformations. We present a method that enables a DNN (student) to learn from the internal representational spaces of a reference model (teacher), which could be another DNN or, in the future, a biological brain. Representational spaces of the student and the teacher are characterized by representational distance matrices (RDMs). We propose representational distance learning (RDL), a stochastic gradient descent method that drives the RDMs of the student to approximate the RDMs of the teacher. We demonstrate that RDL is competitive with other transfer learning techniques for two publicly available benchmark computer vision datasets (MNIST and CIFAR-100), while allowing for architectural differences between student and teacher. By pulling the student's RDMs toward those of the teacher, RDL significantly improved visual classification performance when compared to baseline networks that did not use transfer learning. In the future, RDL may enable combined supervised training of deep neural networks using task constraints (e.g., images and category labels) and constraints from brain-activity measurements, so as to build models that replicate the internal representational spaces of biological brains. PMID:28082889

Representational Distance Learning for Deep Neural Networks.

PubMed

McClure, Patrick; Kriegeskorte, Nikolaus

2016-01-01

Deep neural networks (DNNs) provide useful models of visual representational transformations. We present a method that enables a DNN (student) to learn from the internal representational spaces of a reference model (teacher), which could be another DNN or, in the future, a biological brain. Representational spaces of the student and the teacher are characterized by representational distance matrices (RDMs). We propose representational distance learning (RDL), a stochastic gradient descent method that drives the RDMs of the student to approximate the RDMs of the teacher. We demonstrate that RDL is competitive with other transfer learning techniques for two publicly available benchmark computer vision datasets (MNIST and CIFAR-100), while allowing for architectural differences between student and teacher. By pulling the student's RDMs toward those of the teacher, RDL significantly improved visual classification performance when compared to baseline networks that did not use transfer learning. In the future, RDL may enable combined supervised training of deep neural networks using task constraints (e.g., images and category labels) and constraints from brain-activity measurements, so as to build models that replicate the internal representational spaces of biological brains.

Deep learning for biomarker regression: application to osteoporosis and emphysema on chest CT scans

NASA Astrophysics Data System (ADS)

González, Germán.; Washko, George R.; San José Estépar, Raúl

2018-03-01

Introduction: Biomarker computation using deep-learning often relies on a two-step process, where the deep learning algorithm segments the region of interest and then the biomarker is measured. We propose an alternative paradigm, where the biomarker is estimated directly using a regression network. We showcase this image-tobiomarker paradigm using two biomarkers: the estimation of bone mineral density (BMD) and the estimation of lung percentage of emphysema from CT scans. Materials and methods: We use a large database of 9,925 CT scans to train, validate and test the network for which reference standard BMD and percentage emphysema have been already computed. First, the 3D dataset is reduced to a set of canonical 2D slices where the organ of interest is visible (either spine for BMD or lungs for emphysema). This data reduction is performed using an automatic object detector. Second, The regression neural network is composed of three convolutional layers, followed by a fully connected and an output layer. The network is optimized using a momentum optimizer with an exponential decay rate, using the root mean squared error as cost function. Results: The Pearson correlation coefficients obtained against the reference standards are r = 0.940 (p < 0.00001) and r = 0.976 (p < 0.00001) for BMD and percentage emphysema respectively. Conclusions: The deep-learning regression architecture can learn biomarkers from images directly, without indicating the structures of interest. This approach simplifies the development of biomarker extraction algorithms. The proposed data reduction based on object detectors conveys enough information to compute the biomarkers of interest.

A comparative study of deep learning models for medical image classification

NASA Astrophysics Data System (ADS)

Dutta, Suvajit; Manideep, B. C. S.; Rai, Shalva; Vijayarajan, V.

2017-11-01

Deep Learning(DL) techniques are conquering over the prevailing traditional approaches of neural network, when it comes to the huge amount of dataset, applications requiring complex functions demanding increase accuracy with lower time complexities. Neurosciences has already exploited DL techniques, thus portrayed itself as an inspirational source for researchers exploring the domain of Machine learning. DL enthusiasts cover the areas of vision, speech recognition, motion planning and NLP as well, moving back and forth among fields. This concerns with building models that can successfully solve variety of tasks requiring intelligence and distributed representation. The accessibility to faster CPUs, introduction of GPUs-performing complex vector and matrix computations, supported agile connectivity to network. Enhanced software infrastructures for distributed computing worked in strengthening the thought that made researchers suffice DL methodologies. The paper emphases on the following DL procedures to traditional approaches which are performed manually for classifying medical images. The medical images are used for the study Diabetic Retinopathy(DR) and computed tomography (CT) emphysema data. Both DR and CT data diagnosis is difficult task for normal image classification methods. The initial work was carried out with basic image processing along with K-means clustering for identification of image severity levels. After determining image severity levels ANN has been applied on the data to get the basic classification result, then it is compared with the result of DNNs (Deep Neural Networks), which performed efficiently because of its multiple hidden layer features basically which increases accuracy factors, but the problem of vanishing gradient in DNNs made to consider Convolution Neural Networks (CNNs) as well for better results. The CNNs are found to be providing better outcomes when compared to other learning models aimed at classification of images. CNNs are favoured as they provide better visual processing models successfully classifying the noisy data as well. The work centres on the detection on Diabetic Retinopathy-loss in vision and recognition of computed tomography (CT) emphysema data measuring the severity levels for both cases. The paper discovers how various Machine Learning algorithms can be implemented ensuing a supervised approach, so as to get accurate results with less complexity possible.

Use of a Land Streamer System to Image the Potomac Formation in Northern Delaware

NASA Astrophysics Data System (ADS)

Velez, C. C.; McLaughlin, P. P.; McGeary, S.; Sargent, S. L.

2008-12-01

A land streamer system, an alternative to conventional seismic acquisition equipment for collecting large amounts of seismic reflection data in urbanized and semi-urbanized areas, was used to collect a network of high-resolution seismic reflection data in northern Delaware. The principal objective of this work is to image the distribution and geometry of sand bodies in the Cretaceous (Aptian to Cenomanian) non-marine deposits of the Potomac Formation. The Potomac Formation includes the most important confined aquifers in the Coastal Plain of northern Delaware. Previous studies indicate these deposits onlap Paleozoic basement at depths from 115 m to 400 m in the study area and are truncated by an unconformity. Previous descriptions of sedimentary facies from nearby cores and geophysical logs indicate that the Potomac Formation is a predominantly fine-grained alluvial unit with laterally discontinuous fluvial sand bodies, resulting in a "labyrinth style heterogeneity" for aquifer facies. The 20-km seismic dataset collected for this study indicates that land-streamer seismic methods can be used in this area to image the subsurface geology as shallow as 18 m and as deep as the basement at 315 m. The theoretical quarter wavelength of the seismic dataset suggests a resolution of 2 to 4 m, which is sufficient to resolve aquifer sands in the Potomac Formation ranging from 10 to 20 m thick. Final processed seismic sections will be integrated with geophysical logs and core data to provide a robust 2-D dataset that will allow assessment of current concepts for facies and correlations in the Potomac Formation, thus benefiting understanding of critical ground-water resources.

Application of Super-Resolution Convolutional Neural Network for Enhancing Image Resolution in Chest CT.

PubMed

Umehara, Kensuke; Ota, Junko; Ishida, Takayuki

2017-10-18

In this study, the super-resolution convolutional neural network (SRCNN) scheme, which is the emerging deep-learning-based super-resolution method for enhancing image resolution in chest CT images, was applied and evaluated using the post-processing approach. For evaluation, 89 chest CT cases were sampled from The Cancer Imaging Archive. The 89 CT cases were divided randomly into 45 training cases and 44 external test cases. The SRCNN was trained using the training dataset. With the trained SRCNN, a high-resolution image was reconstructed from a low-resolution image, which was down-sampled from an original test image. For quantitative evaluation, two image quality metrics were measured and compared to those of the conventional linear interpolation methods. The image restoration quality of the SRCNN scheme was significantly higher than that of the linear interpolation methods (p < 0.001 or p < 0.05). The high-resolution image reconstructed by the SRCNN scheme was highly restored and comparable to the original reference image, in particular, for a ×2 magnification. These results indicate that the SRCNN scheme significantly outperforms the linear interpolation methods for enhancing image resolution in chest CT images. The results also suggest that SRCNN may become a potential solution for generating high-resolution CT images from standard CT images.

Comparison of lung tumor motion measured using a model-based 4DCT technique and a commercial protocol.

PubMed

O'Connell, Dylan; Shaverdian, Narek; Kishan, Amar U; Thomas, David H; Dou, Tai H; Lewis, John H; Lamb, James M; Cao, Minsong; Tenn, Stephen; Percy, Lee P; Low, Daniel A

To compare lung tumor motion measured with a model-based technique to commercial 4-dimensional computed tomography (4DCT) scans and describe a workflow for using model-based 4DCT as a clinical simulation protocol. Twenty patients were imaged using a model-based technique and commercial 4DCT. Tumor motion was measured on each commercial 4DCT dataset and was calculated on model-based datasets for 3 breathing amplitude percentile intervals: 5th to 85th, 5th to 95th, and 0th to 100th. Internal target volumes (ITVs) were defined on the 4DCT and 5th to 85th interval datasets and compared using Dice similarity. Images were evaluated for noise and rated by 2 radiation oncologists for artifacts. Mean differences in tumor motion magnitude between commercial and model-based images were 0.47 ± 3.0, 1.63 ± 3.17, and 5.16 ± 4.90 mm for the 5th to 85th, 5th to 95th, and 0th to 100th amplitude intervals, respectively. Dice coefficients between ITVs defined on commercial and 5th to 85th model-based images had a mean value of 0.77 ± 0.09. Single standard deviation image noise was 11.6 ± 9.6 HU in the liver and 6.8 ± 4.7 HU in the aorta for the model-based images compared with 57.7 ± 30 and 33.7 ± 15.4 for commercial 4DCT. Mean model error within the ITV regions was 1.71 ± 0.81 mm. Model-based images exhibited reduced presence of artifacts at the tumor compared with commercial images. Tumor motion measured with the model-based technique using the 5th to 85th percentile breathing amplitude interval corresponded more closely to commercial 4DCT than the 5th to 95th or 0th to 100th intervals, which showed greater motion on average. The model-based technique tended to display increased tumor motion when breathing amplitude intervals wider than 5th to 85th were used because of the influence of unusually deep inhalations. These results suggest that care must be taken in selecting the appropriate interval during image generation when using model-based 4DCT methods. Copyright © 2017 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.

Transforming the Geocomputational Battlespace Framework with HDF5

DTIC Science & Technology

2010-08-01

layout level, dataset arrays can be stored in chunks or tiles , enabling fast subsetting of large datasets, including compressed datasets. HDF software...Image Base (CIB) image of the AOI: an orthophoto made from rectified grayscale aerial images b. An IKONOS satellite image made up of 3 spectral

Large-scale image region documentation for fully automated image biomarker algorithm development and evaluation.

PubMed

Reeves, Anthony P; Xie, Yiting; Liu, Shuang

2017-04-01

With the advent of fully automated image analysis and modern machine learning methods, there is a need for very large image datasets having documented segmentations for both computer algorithm training and evaluation. This paper presents a method and implementation for facilitating such datasets that addresses the critical issue of size scaling for algorithm validation and evaluation; current evaluation methods that are usually used in academic studies do not scale to large datasets. This method includes protocols for the documentation of many regions in very large image datasets; the documentation may be incrementally updated by new image data and by improved algorithm outcomes. This method has been used for 5 years in the context of chest health biomarkers from low-dose chest CT images that are now being used with increasing frequency in lung cancer screening practice. The lung scans are segmented into over 100 different anatomical regions, and the method has been applied to a dataset of over 20,000 chest CT images. Using this framework, the computer algorithms have been developed to achieve over 90% acceptable image segmentation on the complete dataset.

Automatic quality assessment of apical four-chamber echocardiograms using deep convolutional neural networks

NASA Astrophysics Data System (ADS)

Abdi, Amir H.; Luong, Christina; Tsang, Teresa; Allan, Gregory; Nouranian, Saman; Jue, John; Hawley, Dale; Fleming, Sarah; Gin, Ken; Swift, Jody; Rohling, Robert; Abolmaesumi, Purang

2017-02-01

Echocardiography (echo) is the most common test for diagnosis and management of patients with cardiac condi- tions. While most medical imaging modalities benefit from a relatively automated procedure, this is not the case for echo and the quality of the final echo view depends on the competency and experience of the sonographer. It is not uncommon that the sonographer does not have adequate experience to adjust the transducer and acquire a high quality echo, which may further affect the clinical diagnosis. In this work, we aim to aid the operator during image acquisition by automatically assessing the quality of the echo and generating the Automatic Echo Score (AES). This quality assessment method is based on a deep convolutional neural network, trained in an end-to-end fashion on a large dataset of apical four-chamber (A4C) echo images. For this project, an expert car- diologist went through 2,904 A4C images obtained from independent studies and assessed their condition based on a 6-scale grading system. The scores assigned by the expert ranged from 0 to 5. The distribution of scores among the 6 levels were almost uniform. The network was then trained on 80% of the data (2,345 samples). The average absolute error of the trained model in calculating the AES was 0.8 +/- 0:72. The computation time of the GPU implementation of the neural network was estimated at 5 ms per frame, which is sufficient for real-time deployment.

Prediction of crime occurrence from multi-modal data using deep learning

PubMed Central

Kang, Hyeon-Woo

2017-01-01

In recent years, various studies have been conducted on the prediction of crime occurrences. This predictive capability is intended to assist in crime prevention by facilitating effective implementation of police patrols. Previous studies have used data from multiple domains such as demographics, economics, and education. Their prediction models treat data from different domains equally. These methods have problems in crime occurrence prediction, such as difficulty in discovering highly nonlinear relationships, redundancies, and dependencies between multiple datasets. In order to enhance crime prediction models, we consider environmental context information, such as broken windows theory and crime prevention through environmental design. In this paper, we propose a feature-level data fusion method with environmental context based on a deep neural network (DNN). Our dataset consists of data collected from various online databases of crime statistics, demographic and meteorological data, and images in Chicago, Illinois. Prior to generating training data, we select crime-related data by conducting statistical analyses. Finally, we train our DNN, which consists of the following four kinds of layers: spatial, temporal, environmental context, and joint feature representation layers. Coupled with crucial data extracted from various domains, our fusion DNN is a product of an efficient decision-making process that statistically analyzes data redundancy. Experimental performance results show that our DNN model is more accurate in predicting crime occurrence than other prediction models. PMID:28437486

Prediction of crime occurrence from multi-modal data using deep learning.

PubMed

Kang, Hyeon-Woo; Kang, Hang-Bong

2017-01-01

In recent years, various studies have been conducted on the prediction of crime occurrences. This predictive capability is intended to assist in crime prevention by facilitating effective implementation of police patrols. Previous studies have used data from multiple domains such as demographics, economics, and education. Their prediction models treat data from different domains equally. These methods have problems in crime occurrence prediction, such as difficulty in discovering highly nonlinear relationships, redundancies, and dependencies between multiple datasets. In order to enhance crime prediction models, we consider environmental context information, such as broken windows theory and crime prevention through environmental design. In this paper, we propose a feature-level data fusion method with environmental context based on a deep neural network (DNN). Our dataset consists of data collected from various online databases of crime statistics, demographic and meteorological data, and images in Chicago, Illinois. Prior to generating training data, we select crime-related data by conducting statistical analyses. Finally, we train our DNN, which consists of the following four kinds of layers: spatial, temporal, environmental context, and joint feature representation layers. Coupled with crucial data extracted from various domains, our fusion DNN is a product of an efficient decision-making process that statistically analyzes data redundancy. Experimental performance results show that our DNN model is more accurate in predicting crime occurrence than other prediction models.

Deep-sea Lebensspuren of the Australian continental margins

NASA Astrophysics Data System (ADS)

Przeslawski, Rachel; Dundas, Kate; Radke, Lynda; Anderson, Tara J.

Much of the deep sea comprises soft-sediment habitats dominated by comparatively low abundances of species-rich macrofauna and meiofauna. Although often not observed, these animals bioturbate the sediment during feeding and burrowing, leaving signs of their activities called Lebensspuren ('life traces'). In this study, we use still images to quantify Lebensspuren from the eastern (1921 images, 13 stations, 1300-2200 m depth) and western (1008 images, 11 stations, 1500-4400 m depth) Australian margins using a univariate measure of trace richness and a multivariate measure of Lebensspuren assemblages. A total of 46 Lebensspuren types were identified, including those matching named trace fossils and modern Lebensspuren found elsewhere in the world. Most traces could be associated with waste, crawling, dwellings, organism tests, feeding, or resting, but the origin of 15% of trace types remains unknown. Assemblages were significantly different between the two regions and depth profiles, with five Lebensspuren types accounting for over 95% of the differentiation (ovoid pinnate trace, crater row, spider trace, matchstick trace, mesh trace). Lebensspuren richness showed no strong relationships with depth, total organic carbon, or mud, although there was a positive correlation to chlorin index (i.e., organic freshness) in the eastern margin, with richness increasing with organic freshness. Lebensspuren richness was not related to epifauna either, indicating that epifauna may not be the primary source of Lebensspuren. Despite the abundance and distinctiveness of several traces both in the current and previous studies (e.g., ovoid pinnate, mesh, spider), their origin and distribution remains a mystery. We discuss this and several other considerations in the identification and quantification of Lebensspuren. This study represents the first comprehensive catalogue of deep-sea Lebensspuren in Australian waters and highlights the potential of Lebensspuren as valuable and often untapped deep-sea datasets that can be used for biogeographical, evolutionary, behavioural, and ecological studies.

A deep learning model observer for use in alterative forced choice virtual clinical trials

NASA Astrophysics Data System (ADS)

Alnowami, M.; Mills, G.; Awis, M.; Elangovanr, P.; Patel, M.; Halling-Brown, M.; Young, K. C.; Dance, D. R.; Wells, K.

2018-03-01

Virtual clinical trials (VCTs) represent an alternative assessment paradigm that overcomes issues of dose, high cost and delay encountered in conventional clinical trials for breast cancer screening. However, to fully utilize the potential benefits of VCTs requires a machine-based observer that can rapidly and realistically process large numbers of experimental conditions. To address this, a Deep Learning Model Observer (DLMO) was developed and trained to identify lesion targets from normal tissue in small (200 x 200 pixel) image segments, as used in Alternative Forced Choice (AFC) studies. The proposed network consists of 5 convolutional layers with 2x2 kernels and ReLU (Rectified Linear Unit) activations, followed by max pooling with size equal to the size of the final feature maps and three dense layers. The class outputs weights from the final fully connected dense layer are used to consider sets of n images in an n-AFC paradigm to determine the image most likely to contain a target. To examine the DLMO performance on clinical data, a training set of 2814 normal and 2814 biopsy-confirmed malignant mass targets were used. This produced a sensitivity of 0.90 and a specificity of 0.92 when presented with a test data set of 800 previously unseen clinical images. To examine the DLMOs minimum detectable contrast, a second dataset of 630 simulated backgrounds and 630 images with simulated lesion and spherical targets (4mm and 6mm diameter), produced contrast thresholds equivalent to/better than human observer performance for spherical targets, and comparable (12 % difference) for lesion targets.

Presentation Attack Detection for Iris Recognition System Using NIR Camera Sensor

PubMed Central

Nguyen, Dat Tien; Baek, Na Rae; Pham, Tuyen Danh; Park, Kang Ryoung

2018-01-01

Among biometric recognition systems such as fingerprint, finger-vein, or face, the iris recognition system has proven to be effective for achieving a high recognition accuracy and security level. However, several recent studies have indicated that an iris recognition system can be fooled by using presentation attack images that are recaptured using high-quality printed images or by contact lenses with printed iris patterns. As a result, this potential threat can reduce the security level of an iris recognition system. In this study, we propose a new presentation attack detection (PAD) method for an iris recognition system (iPAD) using a near infrared light (NIR) camera image. To detect presentation attack images, we first localized the iris region of the input iris image using circular edge detection (CED). Based on the result of iris localization, we extracted the image features using deep learning-based and handcrafted-based methods. The input iris images were then classified into real and presentation attack categories using support vector machines (SVM). Through extensive experiments with two public datasets, we show that our proposed method effectively solves the iris recognition presentation attack detection problem and produces detection accuracy superior to previous studies. PMID:29695113

Presentation Attack Detection for Iris Recognition System Using NIR Camera Sensor.

PubMed

Nguyen, Dat Tien; Baek, Na Rae; Pham, Tuyen Danh; Park, Kang Ryoung

2018-04-24

Among biometric recognition systems such as fingerprint, finger-vein, or face, the iris recognition system has proven to be effective for achieving a high recognition accuracy and security level. However, several recent studies have indicated that an iris recognition system can be fooled by using presentation attack images that are recaptured using high-quality printed images or by contact lenses with printed iris patterns. As a result, this potential threat can reduce the security level of an iris recognition system. In this study, we propose a new presentation attack detection (PAD) method for an iris recognition system (iPAD) using a near infrared light (NIR) camera image. To detect presentation attack images, we first localized the iris region of the input iris image using circular edge detection (CED). Based on the result of iris localization, we extracted the image features using deep learning-based and handcrafted-based methods. The input iris images were then classified into real and presentation attack categories using support vector machines (SVM). Through extensive experiments with two public datasets, we show that our proposed method effectively solves the iris recognition presentation attack detection problem and produces detection accuracy superior to previous studies.

Internal solitary waves on the Saya de Malha bank of the Mascarene Plateau: SAR observations and interpretation

NASA Astrophysics Data System (ADS)

New, A. L.; Magalhaes, J. M.; da Silva, J. C. B.

2013-09-01

Energetic Internal Solitary Waves (ISWs) were recently discovered radiating from the central region of the Mascarene Plateau in the south-western Indian Ocean (da Silva et al., 2011). SAR imagery revealed the two-dimensional structure of the waves which propagated for several hundred kilometres in deep water both to the east and west of a sill, located near 12.5°S, 61°E between the Saya de Malha and Nazareth banks. These waves were presumed to originate from the disintegration of a large lee wave formed on the western side of the sill at the time of maximum barotropic flow to the west. In the present paper we focus instead on ISWs propagating in the shallow water above the Saya da Malha (SM) bank (to the north of the sill), rather than on those propagating in deep water (here denominated as type-I or -II waves if propagating to the west or east respectively). Analysis of an extended SAR image dataset reveals strong sea surface signatures of complex patterns of ISWs propagating over the SM bank arising from different sources. We identify three distinct types of waves, and propose suitable generation mechanisms for them using synergy from different remotely sensed datasets, together with analyses of linear phase speeds (resulting from local stratification and bathymetry). In particular, we find a family of ISWs (termed here A-type waves) which results from the disintegration of a lee wave which forms on the western slopes of SM. We also identify two further wave trains (B- and C-type waves) which we suggest result from refraction of the deep water type-I and -II waves onto the SM bank. Therefore, both B- and C-type waves can be considered to result from the same generation source as the type-I and -II waves. Finally, we consider the implications of the ISWs for mixing and biological production over the SM bank, and provide direct evidence, from ocean colour satellite images, of enhanced surface chlorophyll over a shallow topographic feature on the bank, which is consistent with the breaking of the ISWs.

A Multi-modal, Discriminative and Spatially Invariant CNN for RGB-D Object Labeling.

PubMed

Asif, Umar; Bennamoun, Mohammed; Sohel, Ferdous

2017-08-30

While deep convolutional neural networks have shown a remarkable success in image classification, the problems of inter-class similarities, intra-class variances, the effective combination of multimodal data, and the spatial variability in images of objects remain to be major challenges. To address these problems, this paper proposes a novel framework to learn a discriminative and spatially invariant classification model for object and indoor scene recognition using multimodal RGB-D imagery. This is achieved through three postulates: 1) spatial invariance - this is achieved by combining a spatial transformer network with a deep convolutional neural network to learn features which are invariant to spatial translations, rotations, and scale changes, 2) high discriminative capability - this is achieved by introducing Fisher encoding within the CNN architecture to learn features which have small inter-class similarities and large intra-class compactness, and 3) multimodal hierarchical fusion - this is achieved through the regularization of semantic segmentation to a multi-modal CNN architecture, where class probabilities are estimated at different hierarchical levels (i.e., imageand pixel-levels), and fused into a Conditional Random Field (CRF)- based inference hypothesis, the optimization of which produces consistent class labels in RGB-D images. Extensive experimental evaluations on RGB-D object and scene datasets, and live video streams (acquired from Kinect) show that our framework produces superior object and scene classification results compared to the state-of-the-art methods.

Face Search at Scale.

PubMed

Wang, Dayong; Otto, Charles; Jain, Anil K

2017-06-01

Given the prevalence of social media websites, one challenge facing computer vision researchers is to devise methods to search for persons of interest among the billions of shared photos on these websites. Despite significant progress in face recognition, searching a large collection of unconstrained face images remains a difficult problem. To address this challenge, we propose a face search system which combines a fast search procedure, coupled with a state-of-the-art commercial off the shelf (COTS) matcher, in a cascaded framework. Given a probe face, we first filter the large gallery of photos to find the top- k most similar faces using features learned by a convolutional neural network. The k retrieved candidates are re-ranked by combining similarities based on deep features and those output by the COTS matcher. We evaluate the proposed face search system on a gallery containing 80 million web-downloaded face images. Experimental results demonstrate that while the deep features perform worse than the COTS matcher on a mugshot dataset (93.7 percent versus 98.6 percent TAR@FAR of 0.01 percent), fusing the deep features with the COTS matcher improves the overall performance ( 99.5 percent TAR@FAR of 0.01 percent). This shows that the learned deep features provide complementary information over representations used in state-of-the-art face matchers. On the unconstrained face image benchmarks, the performance of the learned deep features is competitive with reported accuracies. LFW database: 98.20 percent accuracy under the standard protocol and 88.03 percent TAR@FAR of 0.1 percent under the BLUFR protocol; IJB-A benchmark: 51.0 percent TAR@FAR of 0.1 percent (verification), rank 1 retrieval of 82.2 percent (closed-set search), 61.5 percent FNIR@FAR of 1 percent (open-set search). The proposed face search system offers an excellent trade-off between accuracy and scalability on galleries with millions of images. Additionally, in a face search experiment involving photos of the Tsarnaev brothers, convicted of the Boston Marathon bombing, the proposed cascade face search system could find the younger brother's (Dzhokhar Tsarnaev) photo at rank 1 in 1 second on a 5 M gallery and at rank 8 in 7 seconds on an 80 M gallery.

Fast, Simple and Accurate Handwritten Digit Classification by Training Shallow Neural Network Classifiers with the ‘Extreme Learning Machine’ Algorithm

PubMed Central

McDonnell, Mark D.; Tissera, Migel D.; Vladusich, Tony; van Schaik, André; Tapson, Jonathan

2015-01-01

Recent advances in training deep (multi-layer) architectures have inspired a renaissance in neural network use. For example, deep convolutional networks are becoming the default option for difficult tasks on large datasets, such as image and speech recognition. However, here we show that error rates below 1% on the MNIST handwritten digit benchmark can be replicated with shallow non-convolutional neural networks. This is achieved by training such networks using the ‘Extreme Learning Machine’ (ELM) approach, which also enables a very rapid training time (∼ 10 minutes). Adding distortions, as is common practise for MNIST, reduces error rates even further. Our methods are also shown to be capable of achieving less than 5.5% error rates on the NORB image database. To achieve these results, we introduce several enhancements to the standard ELM algorithm, which individually and in combination can significantly improve performance. The main innovation is to ensure each hidden-unit operates only on a randomly sized and positioned patch of each image. This form of random ‘receptive field’ sampling of the input ensures the input weight matrix is sparse, with about 90% of weights equal to zero. Furthermore, combining our methods with a small number of iterations of a single-batch backpropagation method can significantly reduce the number of hidden-units required to achieve a particular performance. Our close to state-of-the-art results for MNIST and NORB suggest that the ease of use and accuracy of the ELM algorithm for designing a single-hidden-layer neural network classifier should cause it to be given greater consideration either as a standalone method for simpler problems, or as the final classification stage in deep neural networks applied to more difficult problems. PMID:26262687

"Whoa! We're Going Deep in the Trees!": Patterns of Collaboration around an Interactive Information Visualization Exhibit

ERIC Educational Resources Information Center

Davis, Pryce; Horn, Michael; Block, Florian; Phillips, Brenda; Evans, E. Margaret; Diamond, Judy; Shen, Chia

2015-01-01

In this paper we present a qualitative analysis of natural history museum visitor interaction around a multi-touch tabletop exhibit called "DeepTree" that we designed around concepts of evolution and common descent. DeepTree combines several large scientific datasets and an innovative visualization technique to display a phylogenetic…

Large-scale image region documentation for fully automated image biomarker algorithm development and evaluation

PubMed Central

Reeves, Anthony P.; Xie, Yiting; Liu, Shuang

2017-01-01

Abstract. With the advent of fully automated image analysis and modern machine learning methods, there is a need for very large image datasets having documented segmentations for both computer algorithm training and evaluation. This paper presents a method and implementation for facilitating such datasets that addresses the critical issue of size scaling for algorithm validation and evaluation; current evaluation methods that are usually used in academic studies do not scale to large datasets. This method includes protocols for the documentation of many regions in very large image datasets; the documentation may be incrementally updated by new image data and by improved algorithm outcomes. This method has been used for 5 years in the context of chest health biomarkers from low-dose chest CT images that are now being used with increasing frequency in lung cancer screening practice. The lung scans are segmented into over 100 different anatomical regions, and the method has been applied to a dataset of over 20,000 chest CT images. Using this framework, the computer algorithms have been developed to achieve over 90% acceptable image segmentation on the complete dataset. PMID:28612037

Image fusion using sparse overcomplete feature dictionaries

DOEpatents

Brumby, Steven P.; Bettencourt, Luis; Kenyon, Garrett T.; Chartrand, Rick; Wohlberg, Brendt

2015-10-06

Approaches for deciding what individuals in a population of visual system "neurons" are looking for using sparse overcomplete feature dictionaries are provided. A sparse overcomplete feature dictionary may be learned for an image dataset and a local sparse representation of the image dataset may be built using the learned feature dictionary. A local maximum pooling operation may be applied on the local sparse representation to produce a translation-tolerant representation of the image dataset. An object may then be classified and/or clustered within the translation-tolerant representation of the image dataset using a supervised classification algorithm and/or an unsupervised clustering algorithm.

Deep Seismic Imaging of the Hellenic Subduction Zone with New MCS Data of the SISMED Project

NASA Astrophysics Data System (ADS)

Becel, A.; Mireille, L.; Hussni, S.; Dessa, J. X.; Schenini, L.; Sachpazi, M.; Vitard, C.

2016-12-01

The southwestern segment of the Hellenic subduction zone has generated a M>8 tsunamigenic earthquake in the past (365 AD), the largest event ever reported in Europe, but fundamental questions remain about the deep geometry and characteristics of the interplate fault and connected splay faults in the overriding plate that might be rooted in the megathrust. In the Fall 2012, the ULYSSE seismic program acquired deep penetration multichannel seismic (MCS) and OBS refraction profiles across a 300-km-wide section of the forearc domain. MCS data were acquired with a 4.5 km-long streamer on board the R/V Le Pourquoi Pas? from the French IFREMER facilities. The two 240 km-long seismic reflection dip profiles reveal a large and rough topography of the top of the forearc crust in both the outer and inner domains, including a several km thick forearc basin. Despite the thick Messinian evaporites at shallow depths, the 11000 cu.in airgun source reveal several discontinuous arcward-dipping reflections at 15 km depth beneath the outer forearc domain that could be related to the top of the subducting oceanic crust. Unfortunately, the 4.5 km-long streamer is too short for improving their lateral continuity and getting more detailed constraints on their geometry. In the Fall 2015, we chartered the R/V Marcus Langseth equipped with unmatched seismic facilities in the European academic fleet by means of a strong mobilization of the French and American involved laboratories (Géoazur, LDEO, ISTEP, ENS-Paris, EOST, LDO, Pau Univ.) and their research agencies (CNRS, NSF, OCA, and UCA). During the SISMED survey (Seismic Imaging inveStigation in MEDiterranean Sea for deep seismogenic faults), we collected with the R/V Marcus Langseth a 210 km-long profile coincident with the eastern ULYSSE transect with the 8 km-long streamer and a 6600 cu.in tuned airgun array shot every 50 meters. The source and the streamer were towed at a depth of 12 m to maximize low frequencies and deep imaging. Here, we will present the preliminary results of the newly acquired high-quality, high-resolution and deep-penetration data and we will provide a comparison of the two datasets collected with different acquisition parameters.

Large-scale time-lapse microscopy of Oct4 expression in human embryonic stem cell colonies.

PubMed

Bhadriraju, Kiran; Halter, Michael; Amelot, Julien; Bajcsy, Peter; Chalfoun, Joe; Vandecreme, Antoine; Mallon, Barbara S; Park, Kye-Yoon; Sista, Subhash; Elliott, John T; Plant, Anne L

2016-07-01

Identification and quantification of the characteristics of stem cell preparations is critical for understanding stem cell biology and for the development and manufacturing of stem cell based therapies. We have developed image analysis and visualization software that allows effective use of time-lapse microscopy to provide spatial and dynamic information from large numbers of human embryonic stem cell colonies. To achieve statistically relevant sampling, we examined >680 colonies from 3 different preparations of cells over 5days each, generating a total experimental dataset of 0.9 terabyte (TB). The 0.5 Giga-pixel images at each time point were represented by multi-resolution pyramids and visualized using the Deep Zoom Javascript library extended to support viewing Giga-pixel images over time and extracting data on individual colonies. We present a methodology that enables quantification of variations in nominally-identical preparations and between colonies, correlation of colony characteristics with Oct4 expression, and identification of rare events. Copyright © 2016. Published by Elsevier B.V.

A Hybrid Spectral Clustering and Deep Neural Network Ensemble Algorithm for Intrusion Detection in Sensor Networks

PubMed Central

Ma, Tao; Wang, Fen; Cheng, Jianjun; Yu, Yang; Chen, Xiaoyun

2016-01-01

The development of intrusion detection systems (IDS) that are adapted to allow routers and network defence systems to detect malicious network traffic disguised as network protocols or normal access is a critical challenge. This paper proposes a novel approach called SCDNN, which combines spectral clustering (SC) and deep neural network (DNN) algorithms. First, the dataset is divided into k subsets based on sample similarity using cluster centres, as in SC. Next, the distance between data points in a testing set and the training set is measured based on similarity features and is fed into the deep neural network algorithm for intrusion detection. Six KDD-Cup99 and NSL-KDD datasets and a sensor network dataset were employed to test the performance of the model. These experimental results indicate that the SCDNN classifier not only performs better than backpropagation neural network (BPNN), support vector machine (SVM), random forest (RF) and Bayes tree models in detection accuracy and the types of abnormal attacks found. It also provides an effective tool of study and analysis of intrusion detection in large networks. PMID:27754380

A Hybrid Spectral Clustering and Deep Neural Network Ensemble Algorithm for Intrusion Detection in Sensor Networks.

PubMed

Ma, Tao; Wang, Fen; Cheng, Jianjun; Yu, Yang; Chen, Xiaoyun

2016-10-13

The development of intrusion detection systems (IDS) that are adapted to allow routers and network defence systems to detect malicious network traffic disguised as network protocols or normal access is a critical challenge. This paper proposes a novel approach called SCDNN, which combines spectral clustering (SC) and deep neural network (DNN) algorithms. First, the dataset is divided into k subsets based on sample similarity using cluster centres, as in SC. Next, the distance between data points in a testing set and the training set is measured based on similarity features and is fed into the deep neural network algorithm for intrusion detection. Six KDD-Cup99 and NSL-KDD datasets and a sensor network dataset were employed to test the performance of the model. These experimental results indicate that the SCDNN classifier not only performs better than backpropagation neural network (BPNN), support vector machine (SVM), random forest (RF) and Bayes tree models in detection accuracy and the types of abnormal attacks found. It also provides an effective tool of study and analysis of intrusion detection in large networks.

VIS: the visible imager for Euclid

NASA Astrophysics Data System (ADS)

Cropper, Mark; Pottinger, S.; Niemi, S.; Azzollini, R.; Denniston, J.; Szafraniec, M.; Awan, S.; Mellier, Y.; Berthe, M.; Martignac, J.; Cara, C.; Di Giorgio, A.-M.; Sciortino, A.; Bozzo, E.; Genolet, L.; Cole, R.; Philippon, A.; Hailey, M.; Hunt, T.; Swindells, I.; Holland, A.; Gow, J.; Murray, N.; Hall, D.; Skottfelt, J.; Amiaux, J.; Laureijs, R.; Racca, G.; Salvignol, J.-C.; Short, A.; Lorenzo Alvarez, J.; Kitching, T.; Hoekstra, H.; Massey, R.; Israel, H.

2016-07-01

Euclid-VIS is the large format visible imager for the ESA Euclid space mission in their Cosmic Vision program, scheduled for launch in 2020. Together with the near infrared imaging within the NISP instrument, it forms the basis of the weak lensing measurements of Euclid. VIS will image in a single r+i+z band from 550-900 nm over a field of view of ~0.5 deg2. By combining 4 exposures with a total of 2260 sec, VIS will reach to deeper than mAB=24.5 (10σ) for sources with extent ~0.3 arcsec. The image sampling is 0.1 arcsec. VIS will provide deep imaging with a tightly controlled and stable point spread function (PSF) over a wide survey area of 15000 deg2 to measure the cosmic shear from nearly 1.5 billion galaxies to high levels of accuracy, from which the cosmological parameters will be measured. In addition, VIS will also provide a legacy dataset with an unprecedented combination of spatial resolution, depth and area covering most of the extra-Galactic sky. Here we will present the results of the study carried out by the Euclid Consortium during the period up to the Critical Design Review.

Robust multi-atlas label propagation by deep sparse representation

PubMed Central

Zu, Chen; Wang, Zhengxia; Zhang, Daoqiang; Liang, Peipeng; Shi, Yonghong; Shen, Dinggang; Wu, Guorong

2016-01-01

Recently, multi-atlas patch-based label fusion has achieved many successes in medical imaging area. The basic assumption in the current state-of-the-art approaches is that the image patch at the target image point can be represented by a patch dictionary consisting of atlas patches from registered atlas images. Therefore, the label at the target image point can be determined by fusing labels of atlas image patches with similar anatomical structures. However, such assumption on image patch representation does not always hold in label fusion since (1) the image content within the patch may be corrupted due to noise and artifact; and (2) the distribution of morphometric patterns among atlas patches might be unbalanced such that the majority patterns can dominate label fusion result over other minority patterns. The violation of the above basic assumptions could significantly undermine the label fusion accuracy. To overcome these issues, we first consider forming label-specific group for the atlas patches with the same label. Then, we alter the conventional flat and shallow dictionary to a deep multi-layer structure, where the top layer (label-specific dictionaries) consists of groups of representative atlas patches and the subsequent layers (residual dictionaries) hierarchically encode the patchwise residual information in different scales. Thus, the label fusion follows the representation consensus across representative dictionaries. However, the representation of target patch in each group is iteratively optimized by using the representative atlas patches in each label-specific dictionary exclusively to match the principal patterns and also using all residual patterns across groups collaboratively to overcome the issue that some groups might be absent of certain variation patterns presented in the target image patch. Promising segmentation results have been achieved in labeling hippocampus on ADNI dataset, as well as basal ganglia and brainstem structures, compared to other counterpart label fusion methods. PMID:27942077

Robust multi-atlas label propagation by deep sparse representation.

PubMed

Zu, Chen; Wang, Zhengxia; Zhang, Daoqiang; Liang, Peipeng; Shi, Yonghong; Shen, Dinggang; Wu, Guorong

2017-03-01

Recently, multi-atlas patch-based label fusion has achieved many successes in medical imaging area. The basic assumption in the current state-of-the-art approaches is that the image patch at the target image point can be represented by a patch dictionary consisting of atlas patches from registered atlas images. Therefore, the label at the target image point can be determined by fusing labels of atlas image patches with similar anatomical structures. However, such assumption on image patch representation does not always hold in label fusion since (1) the image content within the patch may be corrupted due to noise and artifact; and (2) the distribution of morphometric patterns among atlas patches might be unbalanced such that the majority patterns can dominate label fusion result over other minority patterns. The violation of the above basic assumptions could significantly undermine the label fusion accuracy. To overcome these issues, we first consider forming label-specific group for the atlas patches with the same label. Then, we alter the conventional flat and shallow dictionary to a deep multi-layer structure, where the top layer ( label-specific dictionaries ) consists of groups of representative atlas patches and the subsequent layers ( residual dictionaries ) hierarchically encode the patchwise residual information in different scales. Thus, the label fusion follows the representation consensus across representative dictionaries. However, the representation of target patch in each group is iteratively optimized by using the representative atlas patches in each label-specific dictionary exclusively to match the principal patterns and also using all residual patterns across groups collaboratively to overcome the issue that some groups might be absent of certain variation patterns presented in the target image patch. Promising segmentation results have been achieved in labeling hippocampus on ADNI dataset, as well as basal ganglia and brainstem structures, compared to other counterpart label fusion methods.

Predicting Virtual World User Population Fluctuations with Deep Learning

PubMed Central

Park, Nuri; Zhang, Qimeng; Kim, Jun Gi; Kang, Shin Jin; Kim, Chang Hun

2016-01-01

This paper proposes a system for predicting increases in virtual world user actions. The virtual world user population is a very important aspect of these worlds; however, methods for predicting fluctuations in these populations have not been well documented. Therefore, we attempt to predict changes in virtual world user populations with deep learning, using easily accessible online data, including formal datasets from Google Trends, Wikipedia, and online communities, as well as informal datasets collected from online forums. We use the proposed system to analyze the user population of EVE Online, one of the largest virtual worlds. PMID:27936009

Predicting Virtual World User Population Fluctuations with Deep Learning.

PubMed

Kim, Young Bin; Park, Nuri; Zhang, Qimeng; Kim, Jun Gi; Kang, Shin Jin; Kim, Chang Hun

2016-01-01

This paper proposes a system for predicting increases in virtual world user actions. The virtual world user population is a very important aspect of these worlds; however, methods for predicting fluctuations in these populations have not been well documented. Therefore, we attempt to predict changes in virtual world user populations with deep learning, using easily accessible online data, including formal datasets from Google Trends, Wikipedia, and online communities, as well as informal datasets collected from online forums. We use the proposed system to analyze the user population of EVE Online, one of the largest virtual worlds.

Scaling deep learning on GPU and knights landing clusters

DOE Office of Scientific and Technical Information (OSTI.GOV)

You, Yang; Buluc, Aydin; Demmel, James

Training neural networks has become a big bottleneck. For example, training ImageNet dataset on one Nvidia K20 GPU needs 21 days. To speed up the training process, the current deep learning systems heavily rely on the hardware accelerators. However, these accelerators have limited on-chip memory compared with CPUs. We use both self-host Intel Knights Landing (KNL) clusters and multi-GPU clusters as our target platforms. From the algorithm aspect, we focus on Elastic Averaging SGD (EASGD) to design algorithms for HPC clusters. We redesign four efficient algorithms for HPC systems to improve EASGD's poor scaling on clusters. Async EASGD, Async MEASGD,more » and Hogwild EASGD are faster than existing counter-part methods (Async SGD, Async MSGD, and Hogwild SGD) in all comparisons. Sync EASGD achieves 5.3X speedup over original EASGD on the same platform. We achieve 91.5% weak scaling efficiency on 4253 KNL cores, which is higher than the state-of-the-art implementation.« less

Comparing humans and deep learning performance for grading AMD: A study in using universal deep features and transfer learning for automated AMD analysis.

PubMed

Burlina, Philippe; Pacheco, Katia D; Joshi, Neil; Freund, David E; Bressler, Neil M

2017-03-01

When left untreated, age-related macular degeneration (AMD) is the leading cause of vision loss in people over fifty in the US. Currently it is estimated that about eight million US individuals have the intermediate stage of AMD that is often asymptomatic with regard to visual deficit. These individuals are at high risk for progressing to the advanced stage where the often treatable choroidal neovascular form of AMD can occur. Careful monitoring to detect the onset and prompt treatment of the neovascular form as well as dietary supplementation can reduce the risk of vision loss from AMD, therefore, preferred practice patterns recommend identifying individuals with the intermediate stage in a timely manner. Past automated retinal image analysis (ARIA) methods applied on fundus imagery have relied on engineered and hand-designed visual features. We instead detail the novel application of a machine learning approach using deep learning for the problem of ARIA and AMD analysis. We use transfer learning and universal features derived from deep convolutional neural networks (DCNN). We address clinically relevant 4-class, 3-class, and 2-class AMD severity classification problems. Using 5664 color fundus images from the NIH AREDS dataset and DCNN universal features, we obtain values for accuracy for the (4-, 3-, 2-) class classification problem of (79.4%, 81.5%, 93.4%) for machine vs. (75.8%, 85.0%, 95.2%) for physician grading. This study demonstrates the efficacy of machine grading based on deep universal features/transfer learning when applied to ARIA and is a promising step in providing a pre-screener to identify individuals with intermediate AMD and also as a tool that can facilitate identifying such individuals for clinical studies aimed at developing improved therapies. It also demonstrates comparable performance between computer and physician grading. Copyright © 2017 Elsevier Ltd. All rights reserved.

Converting Static Image Datasets to Spiking Neuromorphic Datasets Using Saccades.

PubMed

Orchard, Garrick; Jayawant, Ajinkya; Cohen, Gregory K; Thakor, Nitish

2015-01-01

Creating datasets for Neuromorphic Vision is a challenging task. A lack of available recordings from Neuromorphic Vision sensors means that data must typically be recorded specifically for dataset creation rather than collecting and labeling existing data. The task is further complicated by a desire to simultaneously provide traditional frame-based recordings to allow for direct comparison with traditional Computer Vision algorithms. Here we propose a method for converting existing Computer Vision static image datasets into Neuromorphic Vision datasets using an actuated pan-tilt camera platform. Moving the sensor rather than the scene or image is a more biologically realistic approach to sensing and eliminates timing artifacts introduced by monitor updates when simulating motion on a computer monitor. We present conversion of two popular image datasets (MNIST and Caltech101) which have played important roles in the development of Computer Vision, and we provide performance metrics on these datasets using spike-based recognition algorithms. This work contributes datasets for future use in the field, as well as results from spike-based algorithms against which future works can compare. Furthermore, by converting datasets already popular in Computer Vision, we enable more direct comparison with frame-based approaches.

Deep Learning in Nuclear Medicine and Molecular Imaging: Current Perspectives and Future Directions.

PubMed

Choi, Hongyoon

2018-04-01

Recent advances in deep learning have impacted various scientific and industrial fields. Due to the rapid application of deep learning in biomedical data, molecular imaging has also started to adopt this technique. In this regard, it is expected that deep learning will potentially affect the roles of molecular imaging experts as well as clinical decision making. This review firstly offers a basic overview of deep learning particularly for image data analysis to give knowledge to nuclear medicine physicians and researchers. Because of the unique characteristics and distinctive aims of various types of molecular imaging, deep learning applications can be different from other fields. In this context, the review deals with current perspectives of deep learning in molecular imaging particularly in terms of development of biomarkers. Finally, future challenges of deep learning application for molecular imaging and future roles of experts in molecular imaging will be discussed.

Software Toolbox for Low-Frequency Conductivity and Current Density Imaging Using MRI.

PubMed

Sajib, Saurav Z K; Katoch, Nitish; Kim, Hyung Joong; Kwon, Oh In; Woo, Eung Je

2017-11-01

Low-frequency conductivity and current density imaging using MRI includes magnetic resonance electrical impedance tomography (MREIT), diffusion tensor MREIT (DT-MREIT), conductivity tensor imaging (CTI), and magnetic resonance current density imaging (MRCDI). MRCDI and MREIT provide current density and isotropic conductivity images, respectively, using current-injection phase MRI techniques. DT-MREIT produces anisotropic conductivity tensor images by incorporating diffusion weighted MRI into MREIT. These current-injection techniques are finding clinical applications in diagnostic imaging and also in transcranial direct current stimulation (tDCS), deep brain stimulation (DBS), and electroporation where treatment currents can function as imaging currents. To avoid adverse effects of nerve and muscle stimulations due to injected currents, conductivity tensor imaging (CTI) utilizes B1 mapping and multi-b diffusion weighted MRI to produce low-frequency anisotropic conductivity tensor images without injecting current. This paper describes numerical implementations of several key mathematical functions for conductivity and current density image reconstructions in MRCDI, MREIT, DT-MREIT, and CTI. To facilitate experimental studies of clinical applications, we developed a software toolbox for these low-frequency conductivity and current density imaging methods. This MR-based conductivity imaging (MRCI) toolbox includes 11 toolbox functions which can be used in the MATLAB environment. The MRCI toolbox is available at http://iirc.khu.ac.kr/software.html . Its functions were tested by using several experimental datasets, which are provided together with the toolbox. Users of the toolbox can focus on experimental designs and interpretations of reconstructed images instead of developing their own image reconstruction softwares. We expect more toolbox functions to be added from future research outcomes. Low-frequency conductivity and current density imaging using MRI includes magnetic resonance electrical impedance tomography (MREIT), diffusion tensor MREIT (DT-MREIT), conductivity tensor imaging (CTI), and magnetic resonance current density imaging (MRCDI). MRCDI and MREIT provide current density and isotropic conductivity images, respectively, using current-injection phase MRI techniques. DT-MREIT produces anisotropic conductivity tensor images by incorporating diffusion weighted MRI into MREIT. These current-injection techniques are finding clinical applications in diagnostic imaging and also in transcranial direct current stimulation (tDCS), deep brain stimulation (DBS), and electroporation where treatment currents can function as imaging currents. To avoid adverse effects of nerve and muscle stimulations due to injected currents, conductivity tensor imaging (CTI) utilizes B1 mapping and multi-b diffusion weighted MRI to produce low-frequency anisotropic conductivity tensor images without injecting current. This paper describes numerical implementations of several key mathematical functions for conductivity and current density image reconstructions in MRCDI, MREIT, DT-MREIT, and CTI. To facilitate experimental studies of clinical applications, we developed a software toolbox for these low-frequency conductivity and current density imaging methods. This MR-based conductivity imaging (MRCI) toolbox includes 11 toolbox functions which can be used in the MATLAB environment. The MRCI toolbox is available at http://iirc.khu.ac.kr/software.html . Its functions were tested by using several experimental datasets, which are provided together with the toolbox. Users of the toolbox can focus on experimental designs and interpretations of reconstructed images instead of developing their own image reconstruction softwares. We expect more toolbox functions to be added from future research outcomes.

Handwritten dynamics assessment through convolutional neural networks: An application to Parkinson's disease identification.

PubMed

Pereira, Clayton R; Pereira, Danilo R; Rosa, Gustavo H; Albuquerque, Victor H C; Weber, Silke A T; Hook, Christian; Papa, João P

2018-05-01

Parkinson's disease (PD) is considered a degenerative disorder that affects the motor system, which may cause tremors, micrography, and the freezing of gait. Although PD is related to the lack of dopamine, the triggering process of its development is not fully understood yet. In this work, we introduce convolutional neural networks to learn features from images produced by handwritten dynamics, which capture different information during the individual's assessment. Additionally, we make available a dataset composed of images and signal-based data to foster the research related to computer-aided PD diagnosis. The proposed approach was compared against raw data and texture-based descriptors, showing suitable results, mainly in the context of early stage detection, with results nearly to 95%. The analysis of handwritten dynamics using deep learning techniques showed to be useful for automatic Parkinson's disease identification, as well as it can outperform handcrafted features. Copyright © 2018 Elsevier B.V. All rights reserved.

Automated Detection of Diabetic Retinopathy using Deep Learning.

PubMed

Lam, Carson; Yi, Darvin; Guo, Margaret; Lindsey, Tony

2018-01-01

Diabetic retinopathy is a leading cause of blindness among working-age adults. Early detection of this condition is critical for good prognosis. In this paper, we demonstrate the use of convolutional neural networks (CNNs) on color fundus images for the recognition task of diabetic retinopathy staging. Our network models achieved test metric performance comparable to baseline literature results, with validation sensitivity of 95%. We additionally explored multinomial classification models, and demonstrate that errors primarily occur in the misclassification of mild disease as normal due to the CNNs inability to detect subtle disease features. We discovered that preprocessing with contrast limited adaptive histogram equalization and ensuring dataset fidelity by expert verification of class labels improves recognition of subtle features. Transfer learning on pretrained GoogLeNet and AlexNet models from ImageNet improved peak test set accuracies to 74.5%, 68.8%, and 57.2% on 2-ary, 3-ary, and 4-ary classification models, respectively.

7T MRI subthalamic nucleus atlas for use with 3T MRI.

PubMed

Milchenko, Mikhail; Norris, Scott A; Poston, Kathleen; Campbell, Meghan C; Ushe, Mwiza; Perlmutter, Joel S; Snyder, Abraham Z

2018-01-01

Deep brain stimulation (DBS) of the subthalamic nucleus (STN) reduces motor symptoms in most patients with Parkinson disease (PD), yet may produce untoward effects. Investigation of DBS effects requires accurate localization of the STN, which can be difficult to identify on magnetic resonance images collected with clinically available 3T scanners. The goal of this study is to develop a high-quality STN atlas that can be applied to standard 3T images. We created a high-definition STN atlas derived from seven older participants imaged at 7T. This atlas was nonlinearly registered to a standard template representing 56 patients with PD imaged at 3T. This process required development of methodology for nonlinear multimodal image registration. We demonstrate mm-scale STN localization accuracy by comparison of our 3T atlas with a publicly available 7T atlas. We also demonstrate less agreement with an earlier histological atlas. STN localization error in the 56 patients imaged at 3T was less than 1 mm on average. Our methodology enables accurate STN localization in individuals imaged at 3T. The STN atlas and underlying 3T average template in MNI space are freely available to the research community. The image registration methodology developed in the course of this work may be generally applicable to other datasets.

Comparing MODIS C6 'Deep Blue' and 'Dark Target' Aerosol Data

NASA Technical Reports Server (NTRS)

Hsu, N. C.; Sayer, A. M.; Bettenhausen, C.; Lee, J.; Levy, R. C.; Mattoo, S.; Munchak, L. A.; Kleidman, R.

2014-01-01

The MODIS Collection 6 Atmospheres product suite includes refined versions of both 'Deep Blue' (DB) and 'Dark Target' (DT) aerosol algorithms, with the DB dataset now expanded to include coverage over vegetated land surfaces. This means that, over much of the global land surface, users will have both DB and DT data to choose from. A 'merged' dataset is also provided, primarily for visualization purposes, which takes retrievals from either or both algorithms based on regional and seasonal climatologies of normalized difference vegetation index (NDVI). This poster present some comparisons of these two C6 aerosol algorithms, focusing on AOD at 550 nm derived from MODIS Aqua measurements, with each other and with Aerosol Robotic Network (AERONET) data, with the intent to facilitate user decisions about the suitability of the two datasets for their desired applications.

Thermodynamic Data Rescue and Informatics for Deep Carbon Science

NASA Astrophysics Data System (ADS)

Zhong, H.; Ma, X.; Prabhu, A.; Eleish, A.; Pan, F.; Parsons, M. A.; Ghiorso, M. S.; West, P.; Zednik, S.; Erickson, J. S.; Chen, Y.; Wang, H.; Fox, P. A.

2017-12-01

A large number of legacy datasets are contained in geoscience literature published between 1930 and 1980 and not expressed external to the publication text in digitized formats. Extracting, organizing, and reusing these "dark" datasets is highly valuable for many within the Earth and planetary science community. As a part of the Deep Carbon Observatory (DCO) data legacy missions, the DCO Data Science Team and Extreme Physics and Chemistry community identified thermodynamic datasets related to carbon, or more specifically datasets about the enthalpy and entropy of chemicals, as a proof of principle analysis. The data science team endeavored to develop a semi-automatic workflow, which includes identifying relevant publications, extracting contained datasets using OCR methods, collaborative reviewing, and registering the datasets via the DCO Data Portal where the 'Linked Data' feature of the data portal provides a mechanism for connecting rescued datasets beyond their individual data sources, to research domains, DCO Communities, and more, making data discovery and retrieval more effective.To date, the team has successfully rescued, deposited and registered additional datasets from publications with thermodynamic sources. These datasets contain 3 main types of data: (1) heat content or enthalpy data determined for a given compound as a function of temperature using high-temperature calorimetry, (2) heat content or enthalpy data determined for a given compound as a function of temperature using adiabatic calorimetry, and (3) direct determination of heat capacity of a compound as a function of temperature using differential scanning calorimetry. The data science team integrated these datasets and delivered a spectrum of data analytics including visualizations, which will lead to a comprehensive characterization of the thermodynamics of carbon and carbon-related materials.

A Deep Convolutional Neural Network for segmenting and classifying epithelial and stromal regions in histopathological images

PubMed Central

Xu, Jun; Luo, Xiaofei; Wang, Guanhao; Gilmore, Hannah; Madabhushi, Anant

2016-01-01

Epithelial (EP) and stromal (ST) are two types of tissues in histological images. Automated segmentation or classification of EP and ST tissues is important when developing computerized system for analyzing the tumor microenvironment. In this paper, a Deep Convolutional Neural Networks (DCNN) based feature learning is presented to automatically segment or classify EP and ST regions from digitized tumor tissue microarrays (TMAs). Current approaches are based on handcraft feature representation, such as color, texture, and Local Binary Patterns (LBP) in classifying two regions. Compared to handcrafted feature based approaches, which involve task dependent representation, DCNN is an end-to-end feature extractor that may be directly learned from the raw pixel intensity value of EP and ST tissues in a data driven fashion. These high-level features contribute to the construction of a supervised classifier for discriminating the two types of tissues. In this work we compare DCNN based models with three handcraft feature extraction based approaches on two different datasets which consist of 157 Hematoxylin and Eosin (H&E) stained images of breast cancer and 1376 immunohistological (IHC) stained images of colorectal cancer, respectively. The DCNN based feature learning approach was shown to have a F1 classification score of 85%, 89%, and 100%, accuracy (ACC) of 84%, 88%, and 100%, and Matthews Correlation Coefficient (MCC) of 86%, 77%, and 100% on two H&E stained (NKI and VGH) and IHC stained data, respectively. Our DNN based approach was shown to outperform three handcraft feature extraction based approaches in terms of the classification of EP and ST regions. PMID:28154470

A Deep Convolutional Neural Network for segmenting and classifying epithelial and stromal regions in histopathological images.

PubMed

Xu, Jun; Luo, Xiaofei; Wang, Guanhao; Gilmore, Hannah; Madabhushi, Anant

2016-05-26

Epithelial (EP) and stromal (ST) are two types of tissues in histological images. Automated segmentation or classification of EP and ST tissues is important when developing computerized system for analyzing the tumor microenvironment. In this paper, a Deep Convolutional Neural Networks (DCNN) based feature learning is presented to automatically segment or classify EP and ST regions from digitized tumor tissue microarrays (TMAs). Current approaches are based on handcraft feature representation, such as color, texture, and Local Binary Patterns (LBP) in classifying two regions. Compared to handcrafted feature based approaches, which involve task dependent representation, DCNN is an end-to-end feature extractor that may be directly learned from the raw pixel intensity value of EP and ST tissues in a data driven fashion. These high-level features contribute to the construction of a supervised classifier for discriminating the two types of tissues. In this work we compare DCNN based models with three handcraft feature extraction based approaches on two different datasets which consist of 157 Hematoxylin and Eosin (H&E) stained images of breast cancer and 1376 immunohistological (IHC) stained images of colorectal cancer, respectively. The DCNN based feature learning approach was shown to have a F1 classification score of 85%, 89%, and 100%, accuracy (ACC) of 84%, 88%, and 100%, and Matthews Correlation Coefficient (MCC) of 86%, 77%, and 100% on two H&E stained (NKI and VGH) and IHC stained data, respectively. Our DNN based approach was shown to outperform three handcraft feature extraction based approaches in terms of the classification of EP and ST regions.

Convolutional networks for fast, energy-efficient neuromorphic computing

PubMed Central

Esser, Steven K.; Merolla, Paul A.; Arthur, John V.; Cassidy, Andrew S.; Appuswamy, Rathinakumar; Andreopoulos, Alexander; Berg, David J.; McKinstry, Jeffrey L.; Melano, Timothy; Barch, Davis R.; di Nolfo, Carmelo; Datta, Pallab; Amir, Arnon; Taba, Brian; Flickner, Myron D.; Modha, Dharmendra S.

2016-01-01

Deep networks are now able to achieve human-level performance on a broad spectrum of recognition tasks. Independently, neuromorphic computing has now demonstrated unprecedented energy-efficiency through a new chip architecture based on spiking neurons, low precision synapses, and a scalable communication network. Here, we demonstrate that neuromorphic computing, despite its novel architectural primitives, can implement deep convolution networks that (i) approach state-of-the-art classification accuracy across eight standard datasets encompassing vision and speech, (ii) perform inference while preserving the hardware’s underlying energy-efficiency and high throughput, running on the aforementioned datasets at between 1,200 and 2,600 frames/s and using between 25 and 275 mW (effectively >6,000 frames/s per Watt), and (iii) can be specified and trained using backpropagation with the same ease-of-use as contemporary deep learning. This approach allows the algorithmic power of deep learning to be merged with the efficiency of neuromorphic processors, bringing the promise of embedded, intelligent, brain-inspired computing one step closer. PMID:27651489

Convolutional networks for fast, energy-efficient neuromorphic computing.

PubMed

Esser, Steven K; Merolla, Paul A; Arthur, John V; Cassidy, Andrew S; Appuswamy, Rathinakumar; Andreopoulos, Alexander; Berg, David J; McKinstry, Jeffrey L; Melano, Timothy; Barch, Davis R; di Nolfo, Carmelo; Datta, Pallab; Amir, Arnon; Taba, Brian; Flickner, Myron D; Modha, Dharmendra S

2016-10-11

Deep networks are now able to achieve human-level performance on a broad spectrum of recognition tasks. Independently, neuromorphic computing has now demonstrated unprecedented energy-efficiency through a new chip architecture based on spiking neurons, low precision synapses, and a scalable communication network. Here, we demonstrate that neuromorphic computing, despite its novel architectural primitives, can implement deep convolution networks that (i) approach state-of-the-art classification accuracy across eight standard datasets encompassing vision and speech, (ii) perform inference while preserving the hardware's underlying energy-efficiency and high throughput, running on the aforementioned datasets at between 1,200 and 2,600 frames/s and using between 25 and 275 mW (effectively >6,000 frames/s per Watt), and (iii) can be specified and trained using backpropagation with the same ease-of-use as contemporary deep learning. This approach allows the algorithmic power of deep learning to be merged with the efficiency of neuromorphic processors, bringing the promise of embedded, intelligent, brain-inspired computing one step closer.

Searching for prostate cancer by fully automated magnetic resonance imaging classification: deep learning versus non-deep learning.

PubMed

Wang, Xinggang; Yang, Wei; Weinreb, Jeffrey; Han, Juan; Li, Qiubai; Kong, Xiangchuang; Yan, Yongluan; Ke, Zan; Luo, Bo; Liu, Tao; Wang, Liang

2017-11-13

Prostate cancer (PCa) is a major cause of death since ancient time documented in Egyptian Ptolemaic mummy imaging. PCa detection is critical to personalized medicine and varies considerably under an MRI scan. 172 patients with 2,602 morphologic images (axial 2D T2-weighted imaging) of the prostate were obtained. A deep learning with deep convolutional neural network (DCNN) and a non-deep learning with SIFT image feature and bag-of-word (BoW), a representative method for image recognition and analysis, were used to distinguish pathologically confirmed PCa patients from prostate benign conditions (BCs) patients with prostatitis or prostate benign hyperplasia (BPH). In fully automated detection of PCa patients, deep learning had a statistically higher area under the receiver operating characteristics curve (AUC) than non-deep learning (P = 0.0007 < 0.001). The AUCs were 0.84 (95% CI 0.78-0.89) for deep learning method and 0.70 (95% CI 0.63-0.77) for non-deep learning method, respectively. Our results suggest that deep learning with DCNN is superior to non-deep learning with SIFT image feature and BoW model for fully automated PCa patients differentiation from prostate BCs patients. Our deep learning method is extensible to image modalities such as MR imaging, CT and PET of other organs.

Blood vessels segmentation of hatching eggs based on fully convolutional networks

NASA Astrophysics Data System (ADS)

Geng, Lei; Qiu, Ling; Wu, Jun; Xiao, Zhitao

2018-04-01

FCN, trained end-to-end, pixels-to-pixels, predict result of each pixel. It has been widely used for semantic segmentation. In order to realize the blood vessels segmentation of hatching eggs, a method based on FCN is proposed in this paper. The training datasets are composed of patches extracted from very few images to augment data. The network combines with lower layer and deconvolution to enables precise segmentation. The proposed method frees from the problem that training deep networks need large scale samples. Experimental results on hatching eggs demonstrate that this method can yield more accurate segmentation outputs than previous researches. It provides a convenient reference for fertility detection subsequently.

Analyzing structural variations along strike in a deep-water thrust belt

NASA Astrophysics Data System (ADS)

Totake, Yukitsugu; Butler, Robert W. H.; Bond, Clare E.; Aziz, Aznan

2018-03-01

We characterize a deep-water fold-thrust arrays imaged by a high-resolution 3D seismic dataset in the offshore NW Borneo, Malaysia, to understand the kinematics behind spatial arrangement of structural variations throughout the fold-thrust system. The seismic volume used covers two sub-parallel fold trains associated with a series of fore-thrusts and back-thrusts. We measured fault heave, shortening value, fold geometries (forelimb dip, interlimb angle and crest depth) along strike in individual fold trains. Heave plot on strike projection allows to identify individual thrust segments showing semi-elliptical to triangular to bimodal patterns, and linkages of these segments. The linkage sites are marked by local minima in cumulative heave. These local heave minima are compensated by additional structures, such as small imbricate thrusts and tight folds indicated by large forelimb dip and small interlimb angle. Complementary profiles of the shortening amount for the two fold trains result in smoother gradient of total shortening across the structures. We interpret this reflects kinematic interaction between two fold-thrust trains. This type of along-strike variation analysis provides comprehensive understanding of a fold-thrust system and may provide an interpretative strategy for inferring the presence of complex multiple faults in less well-imaged parts of seismic volumes.

A deep 3D residual CNN for false-positive reduction in pulmonary nodule detection.

PubMed

Jin, Hongsheng; Li, Zongyao; Tong, Ruofeng; Lin, Lanfen

2018-05-01

The automatic detection of pulmonary nodules using CT scans improves the efficiency of lung cancer diagnosis, and false-positive reduction plays a significant role in the detection. In this paper, we focus on the false-positive reduction task and propose an effective method for this task. We construct a deep 3D residual CNN (convolution neural network) to reduce false-positive nodules from candidate nodules. The proposed network is much deeper than the traditional 3D CNNs used in medical image processing. Specifically, in the network, we design a spatial pooling and cropping (SPC) layer to extract multilevel contextual information of CT data. Moreover, we employ an online hard sample selection strategy in the training process to make the network better fit hard samples (e.g., nodules with irregular shapes). Our method is evaluated on 888 CT scans from the dataset of the LUNA16 Challenge. The free-response receiver operating characteristic (FROC) curve shows that the proposed method achieves a high detection performance. Our experiments confirm that our method is robust and that the SPC layer helps increase the prediction accuracy. Additionally, the proposed method can easily be extended to other 3D object detection tasks in medical image processing. © 2018 American Association of Physicists in Medicine.

ESO imaging survey: optical deep public survey

NASA Astrophysics Data System (ADS)

Mignano, A.; Miralles, J.-M.; da Costa, L.; Olsen, L. F.; Prandoni, I.; Arnouts, S.; Benoist, C.; Madejsky, R.; Slijkhuis, R.; Zaggia, S.

2007-02-01

This paper presents new five passbands (UBVRI) optical wide-field imaging data accumulated as part of the DEEP Public Survey (DPS) carried out as a public survey by the ESO Imaging Survey (EIS) project. Out of the 3 square degrees originally proposed, the survey covers 2.75 square degrees, in at least one band (normally R), and 1.00 square degrees in five passbands. The median seeing, as measured in the final stacked images, is 0.97 arcsec, ranging from 0.75 arcsec to 2.0 arcsec. The median limiting magnitudes (AB system, 2´´ aperture, 5σ detection limit) are UAB=25.65, BAB=25.54, VAB=25.18, RAB = 24.8 and IAB =24.12 mag, consistent with those proposed in the original survey design. The paper describes the observations and data reduction using the EIS Data Reduction System and its associated EIS/MVM library. The quality of the individual images were inspected, bad images discarded and the remaining used to produce final image stacks in each passband, from which sources have been extracted. Finally, the scientific quality of these final images and associated catalogs was assessed qualitatively by visual inspection and quantitatively by comparison of statistical measures derived from these data with those of other authors as well as model predictions, and from direct comparison with the results obtained from the reduction of the same dataset using an independent (hands-on) software system. Finally to illustrate one application of this survey, the results of a preliminary effort to identify sub-mJy radio sources are reported. To the limiting magnitude reached in the R and I passbands the success rate ranges from 66 to 81% (depending on the fields). These data are publicly available at CDS. Based on observations carried out at the European Southern Observatory, La Silla, Chile under program Nos. 164.O-0561, 169.A-0725, and 267.A-5729. Appendices A, B and C are only available in electronic form at http://www.aanda.org

Overview of deep learning in medical imaging.

PubMed

Suzuki, Kenji

2017-09-01

The use of machine learning (ML) has been increasing rapidly in the medical imaging field, including computer-aided diagnosis (CAD), radiomics, and medical image analysis. Recently, an ML area called deep learning emerged in the computer vision field and became very popular in many fields. It started from an event in late 2012, when a deep-learning approach based on a convolutional neural network (CNN) won an overwhelming victory in the best-known worldwide computer vision competition, ImageNet Classification. Since then, researchers in virtually all fields, including medical imaging, have started actively participating in the explosively growing field of deep learning. In this paper, the area of deep learning in medical imaging is overviewed, including (1) what was changed in machine learning before and after the introduction of deep learning, (2) what is the source of the power of deep learning, (3) two major deep-learning models: a massive-training artificial neural network (MTANN) and a convolutional neural network (CNN), (4) similarities and differences between the two models, and (5) their applications to medical imaging. This review shows that ML with feature input (or feature-based ML) was dominant before the introduction of deep learning, and that the major and essential difference between ML before and after deep learning is the learning of image data directly without object segmentation or feature extraction; thus, it is the source of the power of deep learning, although the depth of the model is an important attribute. The class of ML with image input (or image-based ML) including deep learning has a long history, but recently gained popularity due to the use of the new terminology, deep learning. There are two major models in this class of ML in medical imaging, MTANN and CNN, which have similarities as well as several differences. In our experience, MTANNs were substantially more efficient in their development, had a higher performance, and required a lesser number of training cases than did CNNs. "Deep learning", or ML with image input, in medical imaging is an explosively growing, promising field. It is expected that ML with image input will be the mainstream area in the field of medical imaging in the next few decades.

CPTAC Releases Largest-Ever Ovarian Cancer Proteome Dataset from Previously Genome Characterized Tumors | Office of Cancer Clinical Proteomics Research

Cancer.gov

National Cancer Institute (NCI) Clinical Proteomic Tumor Analysis Consortium (CPTAC) scientists have just released a comprehensive dataset of the proteomic analysis of high grade serous ovarian tumor samples, previously genomically analyzed by The Cancer Genome Atlas (TCGA).  This is one of the largest public datasets covering the proteome, phosphoproteome and glycoproteome with complementary deep genomic sequencing data on the same tumor.

What does fault tolerant Deep Learning need from MPI?

DOE Office of Scientific and Technical Information (OSTI.GOV)

Amatya, Vinay C.; Vishnu, Abhinav; Siegel, Charles M.

Deep Learning (DL) algorithms have become the {\\em de facto} Machine Learning (ML) algorithm for large scale data analysis. DL algorithms are computationally expensive -- even distributed DL implementations which use MPI require days of training (model learning) time on commonly studied datasets. Long running DL applications become susceptible to faults -- requiring development of a fault tolerant system infrastructure, in addition to fault tolerant DL algorithms. This raises an important question: {\\em What is needed from MPI for designing fault tolerant DL implementations?} In this paper, we address this problem for permanent faults. We motivate the need for amore » fault tolerant MPI specification by an in-depth consideration of recent innovations in DL algorithms and their properties, which drive the need for specific fault tolerance features. We present an in-depth discussion on the suitability of different parallelism types (model, data and hybrid); a need (or lack thereof) for check-pointing of any critical data structures; and most importantly, consideration for several fault tolerance proposals (user-level fault mitigation (ULFM), Reinit) in MPI and their applicability to fault tolerant DL implementations. We leverage a distributed memory implementation of Caffe, currently available under the Machine Learning Toolkit for Extreme Scale (MaTEx). We implement our approaches by extending MaTEx-Caffe for using ULFM-based implementation. Our evaluation using the ImageNet dataset and AlexNet neural network topology demonstrates the effectiveness of the proposed fault tolerant DL implementation using OpenMPI based ULFM.« less

Extraction of skin lesions from non-dermoscopic images for surgical excision of melanoma.

PubMed

Jafari, M Hossein; Nasr-Esfahani, Ebrahim; Karimi, Nader; Soroushmehr, S M Reza; Samavi, Shadrokh; Najarian, Kayvan

2017-06-01

Computerized prescreening of suspicious moles and lesions for malignancy is of great importance for assessing the need and the priority of the removal surgery. Detection can be done by images captured by standard cameras, which are more preferable due to low cost and availability. One important step in computerized evaluation is accurate detection of lesion's region, i.e., segmentation of an image into two regions as lesion and normal skin. In this paper, a new method based on deep neural networks is proposed for accurate extraction of a lesion region. The input image is preprocessed, and then, its patches are fed to a convolutional neural network. Local texture and global structure of the patches are processed in order to assign pixels to lesion or normal classes. A method for effective selection of training patches is proposed for more accurate detection of a lesion's border. Our results indicate that the proposed method could reach the accuracy of 98.7% and the sensitivity of 95.2% in segmentation of lesion regions over the dataset of clinical images. The experimental results of qualitative and quantitative evaluations demonstrate that our method can outperform other state-of-the-art algorithms exist in the literature.

Deep learning of the sectional appearances of 3D CT images for anatomical structure segmentation based on an FCN voting method.

PubMed

Zhou, Xiangrong; Takayama, Ryosuke; Wang, Song; Hara, Takeshi; Fujita, Hiroshi

2017-10-01

We propose a single network trained by pixel-to-label deep learning to address the general issue of automatic multiple organ segmentation in three-dimensional (3D) computed tomography (CT) images. Our method can be described as a voxel-wise multiple-class classification scheme for automatically assigning labels to each pixel/voxel in a 2D/3D CT image. We simplify the segmentation algorithms of anatomical structures (including multiple organs) in a CT image (generally in 3D) to a majority voting scheme over the semantic segmentation of multiple 2D slices drawn from different viewpoints with redundancy. The proposed method inherits the spirit of fully convolutional networks (FCNs) that consist of "convolution" and "deconvolution" layers for 2D semantic image segmentation, and expands the core structure with 3D-2D-3D transformations to adapt to 3D CT image segmentation. All parameters in the proposed network are trained pixel-to-label from a small number of CT cases with human annotations as the ground truth. The proposed network naturally fulfills the requirements of multiple organ segmentations in CT cases of different sizes that cover arbitrary scan regions without any adjustment. The proposed network was trained and validated using the simultaneous segmentation of 19 anatomical structures in the human torso, including 17 major organs and two special regions (lumen and content inside of stomach). Some of these structures have never been reported in previous research on CT segmentation. A database consisting of 240 (95% for training and 5% for testing) 3D CT scans, together with their manually annotated ground-truth segmentations, was used in our experiments. The results show that the 19 structures of interest were segmented with acceptable accuracy (88.1% and 87.9% voxels in the training and testing datasets, respectively, were labeled correctly) against the ground truth. We propose a single network based on pixel-to-label deep learning to address the challenging issue of anatomical structure segmentation in 3D CT cases. The novelty of this work is the policy of deep learning of the different 2D sectional appearances of 3D anatomical structures for CT cases and the majority voting of the 3D segmentation results from multiple crossed 2D sections to achieve availability and reliability with better efficiency, generality, and flexibility than conventional segmentation methods, which must be guided by human expertise. © 2017 The Authors. Medical Physics published by Wiley Periodicals, Inc. on behalf of American Association of Physicists in Medicine.

CIFAR10-DVS: An Event-Stream Dataset for Object Classification

PubMed Central

Li, Hongmin; Liu, Hanchao; Ji, Xiangyang; Li, Guoqi; Shi, Luping

2017-01-01

Neuromorphic vision research requires high-quality and appropriately challenging event-stream datasets to support continuous improvement of algorithms and methods. However, creating event-stream datasets is a time-consuming task, which needs to be recorded using the neuromorphic cameras. Currently, there are limited event-stream datasets available. In this work, by utilizing the popular computer vision dataset CIFAR-10, we converted 10,000 frame-based images into 10,000 event streams using a dynamic vision sensor (DVS), providing an event-stream dataset of intermediate difficulty in 10 different classes, named as “CIFAR10-DVS.” The conversion of event-stream dataset was implemented by a repeated closed-loop smooth (RCLS) movement of frame-based images. Unlike the conversion of frame-based images by moving the camera, the image movement is more realistic in respect of its practical applications. The repeated closed-loop image movement generates rich local intensity changes in continuous time which are quantized by each pixel of the DVS camera to generate events. Furthermore, a performance benchmark in event-driven object classification is provided based on state-of-the-art classification algorithms. This work provides a large event-stream dataset and an initial benchmark for comparison, which may boost algorithm developments in even-driven pattern recognition and object classification. PMID:28611582

CIFAR10-DVS: An Event-Stream Dataset for Object Classification.

PubMed

Li, Hongmin; Liu, Hanchao; Ji, Xiangyang; Li, Guoqi; Shi, Luping

2017-01-01

Neuromorphic vision research requires high-quality and appropriately challenging event-stream datasets to support continuous improvement of algorithms and methods. However, creating event-stream datasets is a time-consuming task, which needs to be recorded using the neuromorphic cameras. Currently, there are limited event-stream datasets available. In this work, by utilizing the popular computer vision dataset CIFAR-10, we converted 10,000 frame-based images into 10,000 event streams using a dynamic vision sensor (DVS), providing an event-stream dataset of intermediate difficulty in 10 different classes, named as "CIFAR10-DVS." The conversion of event-stream dataset was implemented by a repeated closed-loop smooth (RCLS) movement of frame-based images. Unlike the conversion of frame-based images by moving the camera, the image movement is more realistic in respect of its practical applications. The repeated closed-loop image movement generates rich local intensity changes in continuous time which are quantized by each pixel of the DVS camera to generate events. Furthermore, a performance benchmark in event-driven object classification is provided based on state-of-the-art classification algorithms. This work provides a large event-stream dataset and an initial benchmark for comparison, which may boost algorithm developments in even-driven pattern recognition and object classification.

Multi-sourced, 3D geometric characterization of volcanogenic karst features: Integrating lidar, sonar, and geophysical datasets (Invited)

NASA Astrophysics Data System (ADS)

Sharp, J. M.; Gary, M. O.; Reyes, R.; Halihan, T.; Fairfield, N.; Stone, W. C.

2009-12-01

Karstic aquifers can form very complex hydrogeological systems and 3-D mapping has been difficult, but Lidar, phased array sonar, and improved earth resistivity techniques show promise in this and in linking metadata to models. Zacatón, perhaps the Earth’s deepest cenote, has a sub-aquatic void space exceeding 7.5 x 106 cubic m3. It is the focus of this study which has created detailed 3D maps of the system. These maps include data from above and beneath the the water table and within the rock matrix to document the extent of the immense karst features and to interpret the geologic processes that formed them. Phase 1 used high resolution (20 mm) Lidar scanning of surficial features of four large cenotes. Scan locations, selected to achieve full feature coverage once registered, were established atop surface benchmarks with UTM coordinates established using GPS and Total Stations. The combined datasets form a geo-registered mesh of surface features down to water level in the cenotes. Phase 2 conducted subsurface imaging using Earth Resistivity Imaging (ERI) geophysics. ERI identified void spaces isolated from open flow conduits. A unique travertine morphology exists in which some cenotes are dry or contain shallow lakes with flat travertine floors; some water-filled cenotes have flat floors without the cone of collapse material; and some have collapse cones. We hypothesize that the floors may have large water-filled voids beneath them. Three separate flat travertine caps were imaged: 1) La Pilita, which is partially open, exposing cap structure over a deep water-filled shaft; 2) Poza Seca, which is dry and vegetated; and 3) Tule, which contains a shallow (<1 m) lake. A fourth line was run adjacent to cenote Verde. La Pilita ERI, verified by SCUBA, documented the existence of large water-filled void zones ERI at Poza Seca showed a thin cap overlying a conductive zone extending to at least 25 m depth beneath the cap with no lower boundary of this zone evident. Verde ERI indicate a deep water-filled cavity below the 45 m deep floor. Phase 3 acquired high-resolution imagery of the underwater voids. Because of the great depths (In 1994, Bowden descended to a record 289 meters), unmanned exploration is required to explore these systems. Supported by NASA, the DEPTHX (DEep Phreatic THermal eXplorer) robotic mapper was designed. DEPTHX conducted 3-D underwater mapping missions of 4 cenotes. These found no lateral tunnels connecting the cenotes. The cenote, El Zacatón was shown to be 319 meters deep making it is the deepest underwater vertical shaft and second deepest underwater cave in the world. Spatial geochemical data collected during DEPTHX mapping indicate that water in the 3 of the cenotes is homogeneous, but the 4th (Verde) displays typical lacustrine chemoclines. The data collected by DEPTHX data are being combined with other geologic information to study the nature of the hypogenic karst processes that formed this system.

A novel microaneurysms detection approach based on convolutional neural networks with reinforcement sample learning algorithm.

PubMed

Budak, Umit; Şengür, Abdulkadir; Guo, Yanhui; Akbulut, Yaman

2017-12-01

Microaneurysms (MAs) are known as early signs of diabetic-retinopathy which are called red lesions in color fundus images. Detection of MAs in fundus images needs highly skilled physicians or eye angiography. Eye angiography is an invasive and expensive procedure. Therefore, an automatic detection system to identify the MAs locations in fundus images is in demand. In this paper, we proposed a system to detect the MAs in colored fundus images. The proposed method composed of three stages. In the first stage, a series of pre-processing steps are used to make the input images more convenient for MAs detection. To this end, green channel decomposition, Gaussian filtering, median filtering, back ground determination, and subtraction operations are applied to input colored fundus images. After pre-processing, a candidate MAs extraction procedure is applied to detect potential regions. A five-stepped procedure is adopted to get the potential MA locations. Finally, deep convolutional neural network (DCNN) with reinforcement sample learning strategy is used to train the proposed system. The DCNN is trained with color image patches which are collected from ground-truth MA locations and non-MA locations. We conducted extensive experiments on ROC dataset to evaluate of our proposal. The results are encouraging.

A sparse autoencoder-based deep neural network for protein solvent accessibility and contact number prediction.

PubMed

Deng, Lei; Fan, Chao; Zeng, Zhiwen

2017-12-28

Direct prediction of the three-dimensional (3D) structures of proteins from one-dimensional (1D) sequences is a challenging problem. Significant structural characteristics such as solvent accessibility and contact number are essential for deriving restrains in modeling protein folding and protein 3D structure. Thus, accurately predicting these features is a critical step for 3D protein structure building. In this study, we present DeepSacon, a computational method that can effectively predict protein solvent accessibility and contact number by using a deep neural network, which is built based on stacked autoencoder and a dropout method. The results demonstrate that our proposed DeepSacon achieves a significant improvement in the prediction quality compared with the state-of-the-art methods. We obtain 0.70 three-state accuracy for solvent accessibility, 0.33 15-state accuracy and 0.74 Pearson Correlation Coefficient (PCC) for the contact number on the 5729 monomeric soluble globular protein dataset. We also evaluate the performance on the CASP11 benchmark dataset, DeepSacon achieves 0.68 three-state accuracy and 0.69 PCC for solvent accessibility and contact number, respectively. We have shown that DeepSacon can reliably predict solvent accessibility and contact number with stacked sparse autoencoder and a dropout approach.

CPTAC Releases Largest-Ever Breast Cancer Proteome Dataset from Previously Genome Characterized Tumors | Office of Cancer Clinical Proteomics Research

Cancer.gov

National Cancer Institute (NCI) Clinical Proteomic Tumor Analysis Consortium (CPTAC) scientists have released a dataset of proteins and  phosphopeptides identified through deep proteomic and phosphoproteomic analysis of breast tumor samples, previously genomically analyzed by The Cancer Genome Atlas (TCGA).

Slab seismicity in the Western Hellenic Subduction Zone: Constraints from tomography and double-difference relocation

NASA Astrophysics Data System (ADS)

Halpaap, Felix; Rondenay, Stéphane; Ottemöller, Lars

2016-04-01

The Western Hellenic subduction zone is characterized by a transition from oceanic to continental subduction. In the southern oceanic portion of the system, abundant seismicity reaches intermediate depths of 100-120 km, while the northern continental portion rarely exhibits deep earthquakes. Our study aims to investigate how this oceanic-continental transition affects fluid release and related seismicity along strike, by focusing on the distribution of intermediate depth earthquakes. To obtain a detailed image of the seismicity, we carry out a tomographic inversion for P- and S-velocities and double-difference earthquake relocation using a dataset of unprecedented spatial coverage in this area. Here we present results of these analyses in conjunction with high-resolution profiles from migrated receiver function images obtained from the MEDUSA experiment. We generate tomographic models by inverting data from 237 manually picked, well locatable events recorded at up to 130 stations. Stations from the permanent Greek network and the EGELADOS experiment supplement the 3-D coverage of the modeled domain, which covers a large part of mainland Greece and surrounding offshore areas. Corrections for the sphericity of the Earth and our update to the SIMULR16 package, which now allows S-inversion, help improve our previous models. Flexible gridding focusses the inversion on the domains of highest gradient around the slab, and we evaluate the resolution with checker board tests. We use the resulting velocity model to relocate earthquakes via the Double-Difference method, using a large dataset of differential traveltimes obtained by crosscorrelation of seismograms. Tens of earthquakes align along two planes forming a double seismic zone in the southern, oceanic portion of the subduction zone. With increasing subduction depth, the earthquakes appear closer to the center of the slab, outlining probable deserpentinization of the slab and concomitant eclogitization of dry crustal rocks. Against expectations, we relocate one robust deep event at ≈70 km depth in the northern, continental part of the subduction zone.

Dataset variability leverages white-matter lesion segmentation performance with convolutional neural network

NASA Astrophysics Data System (ADS)

Ravnik, Domen; Jerman, Tim; Pernuš, Franjo; Likar, Boštjan; Å piclin, Žiga

2018-03-01

Performance of a convolutional neural network (CNN) based white-matter lesion segmentation in magnetic resonance (MR) brain images was evaluated under various conditions involving different levels of image preprocessing and augmentation applied and different compositions of the training dataset. On images of sixty multiple sclerosis patients, half acquired on one and half on another scanner of different vendor, we first created a highly accurate multi-rater consensus based lesion segmentations, which were used in several experiments to evaluate the CNN segmentation result. First, the CNN was trained and tested without preprocessing the images and by using various combinations of preprocessing techniques, namely histogram-based intensity standardization, normalization by whitening, and train dataset augmentation by flipping the images across the midsagittal plane. Then, the CNN was trained and tested on images of the same, different or interleaved scanner datasets using a cross-validation approach. The results indicate that image preprocessing has little impact on performance in a same-scanner situation, while between-scanner performance benefits most from intensity standardization and normalization, but also further by incorporating heterogeneous multi-scanner datasets in the training phase. Under such conditions the between-scanner performance of the CNN approaches that of the ideal situation, when the CNN is trained and tested on the same scanner dataset.

DeepGene: an advanced cancer type classifier based on deep learning and somatic point mutations.

PubMed

Yuan, Yuchen; Shi, Yi; Li, Changyang; Kim, Jinman; Cai, Weidong; Han, Zeguang; Feng, David Dagan

2016-12-23

With the developments of DNA sequencing technology, large amounts of sequencing data have become available in recent years and provide unprecedented opportunities for advanced association studies between somatic point mutations and cancer types/subtypes, which may contribute to more accurate somatic point mutation based cancer classification (SMCC). However in existing SMCC methods, issues like high data sparsity, small volume of sample size, and the application of simple linear classifiers, are major obstacles in improving the classification performance. To address the obstacles in existing SMCC studies, we propose DeepGene, an advanced deep neural network (DNN) based classifier, that consists of three steps: firstly, the clustered gene filtering (CGF) concentrates the gene data by mutation occurrence frequency, filtering out the majority of irrelevant genes; secondly, the indexed sparsity reduction (ISR) converts the gene data into indexes of its non-zero elements, thereby significantly suppressing the impact of data sparsity; finally, the data after CGF and ISR is fed into a DNN classifier, which extracts high-level features for accurate classification. Experimental results on our curated TCGA-DeepGene dataset, which is a reformulated subset of the TCGA dataset containing 12 selected types of cancer, show that CGF, ISR and DNN all contribute in improving the overall classification performance. We further compare DeepGene with three widely adopted classifiers and demonstrate that DeepGene has at least 24% performance improvement in terms of testing accuracy. Based on deep learning and somatic point mutation data, we devise DeepGene, an advanced cancer type classifier, which addresses the obstacles in existing SMCC studies. Experiments indicate that DeepGene outperforms three widely adopted existing classifiers, which is mainly attributed to its deep learning module that is able to extract the high level features between combinatorial somatic point mutations and cancer types.

Image classification using multiscale information fusion based on saliency driven nonlinear diffusion filtering.

PubMed

Hu, Weiming; Hu, Ruiguang; Xie, Nianhua; Ling, Haibin; Maybank, Stephen

2014-04-01

In this paper, we propose saliency driven image multiscale nonlinear diffusion filtering. The resulting scale space in general preserves or even enhances semantically important structures such as edges, lines, or flow-like structures in the foreground, and inhibits and smoothes clutter in the background. The image is classified using multiscale information fusion based on the original image, the image at the final scale at which the diffusion process converges, and the image at a midscale. Our algorithm emphasizes the foreground features, which are important for image classification. The background image regions, whether considered as contexts of the foreground or noise to the foreground, can be globally handled by fusing information from different scales. Experimental tests of the effectiveness of the multiscale space for the image classification are conducted on the following publicly available datasets: 1) the PASCAL 2005 dataset; 2) the Oxford 102 flowers dataset; and 3) the Oxford 17 flowers dataset, with high classification rates.

Object-Oriented Image Clustering Method Using UAS Photogrammetric Imagery

NASA Astrophysics Data System (ADS)

Lin, Y.; Larson, A.; Schultz-Fellenz, E. S.; Sussman, A. J.; Swanson, E.; Coppersmith, R.

2016-12-01

Unmanned Aerial Systems (UAS) have been used widely as an imaging modality to obtain remotely sensed multi-band surface imagery, and are growing in popularity due to their efficiency, ease of use, and affordability. Los Alamos National Laboratory (LANL) has employed the use of UAS for geologic site characterization and change detection studies at a variety of field sites. The deployed UAS equipped with a standard visible band camera to collect imagery datasets. Based on the imagery collected, we use deep sparse algorithmic processing to detect and discriminate subtle topographic features created or impacted by subsurface activities. In this work, we develop an object-oriented remote sensing imagery clustering method for land cover classification. To improve the clustering and segmentation accuracy, instead of using conventional pixel-based clustering methods, we integrate the spatial information from neighboring regions to create super-pixels to avoid salt-and-pepper noise and subsequent over-segmentation. To further improve robustness of our clustering method, we also incorporate a custom digital elevation model (DEM) dataset generated using a structure-from-motion (SfM) algorithm together with the red, green, and blue (RGB) band data for clustering. In particular, we first employ an agglomerative clustering to create an initial segmentation map, from where every object is treated as a single (new) pixel. Based on the new pixels obtained, we generate new features to implement another level of clustering. We employ our clustering method to the RGB+DEM datasets collected at the field site. Through binary clustering and multi-object clustering tests, we verify that our method can accurately separate vegetation from non-vegetation regions, and are also able to differentiate object features on the surface.

Can a CNN recognize Catalan diet?

NASA Astrophysics Data System (ADS)

Herruzo, P.; Bolaños, M.; Radeva, P.

2016-10-01

Nowadays, we can find several diseases related to the unhealthy diet habits of the population, such as diabetes, obesity, anemia, bulimia and anorexia. In many cases, these diseases are related to the food consumption of people. Mediterranean diet is scientifically known as a healthy diet that helps to prevent many metabolic diseases. In particular, our work focuses on the recognition of Mediterranean food and dishes. The development of this methodology would allow to analise the daily habits of users with wearable cameras, within the topic of lifelogging. By using automatic mechanisms we could build an objective tool for the analysis of the patient's behavior, allowing specialists to discover unhealthy food patterns and understand the user's lifestyle. With the aim to automatically recognize a complete diet, we introduce a challenging multi-labeled dataset related to Mediter-ranean diet called FoodCAT. The first type of label provided consists of 115 food classes with an average of 400 images per dish, and the second one consists of 12 food categories with an average of 3800 pictures per class. This dataset will serve as a basis for the development of automatic diet recognition. In this context, deep learning and more specifically, Convolutional Neural Networks (CNNs), currently are state-of-the-art methods for automatic food recognition. In our work, we compare several architectures for image classification, with the purpose of diet recognition. Applying the best model for recognising food categories, we achieve a top-1 accuracy of 72.29%, and top-5 of 97.07%. In a complete diet recognition of dishes from Mediterranean diet, enlarged with the Food-101 dataset for international dishes recognition, we achieve a top-1 accuracy of 68.07%, and top-5 of 89.53%, for a total of 115+101 food classes.

Research on image retrieval using deep convolutional neural network combining L1 regularization and PRelu activation function

NASA Astrophysics Data System (ADS)

QingJie, Wei; WenBin, Wang

2017-06-01

In this paper, the image retrieval using deep convolutional neural network combined with regularization and PRelu activation function is studied, and improves image retrieval accuracy. Deep convolutional neural network can not only simulate the process of human brain to receive and transmit information, but also contains a convolution operation, which is very suitable for processing images. Using deep convolutional neural network is better than direct extraction of image visual features for image retrieval. However, the structure of deep convolutional neural network is complex, and it is easy to over-fitting and reduces the accuracy of image retrieval. In this paper, we combine L1 regularization and PRelu activation function to construct a deep convolutional neural network to prevent over-fitting of the network and improve the accuracy of image retrieval

DNCON2: improved protein contact prediction using two-level deep convolutional neural networks.

PubMed

Adhikari, Badri; Hou, Jie; Cheng, Jianlin

2018-05-01

Significant improvements in the prediction of protein residue-residue contacts are observed in the recent years. These contacts, predicted using a variety of coevolution-based and machine learning methods, are the key contributors to the recent progress in ab initio protein structure prediction, as demonstrated in the recent CASP experiments. Continuing the development of new methods to reliably predict contact maps is essential to further improve ab initio structure prediction. In this paper we discuss DNCON2, an improved protein contact map predictor based on two-level deep convolutional neural networks. It consists of six convolutional neural networks-the first five predict contacts at 6, 7.5, 8, 8.5 and 10 Å distance thresholds, and the last one uses these five predictions as additional features to predict final contact maps. On the free-modeling datasets in CASP10, 11 and 12 experiments, DNCON2 achieves mean precisions of 35, 50 and 53.4%, respectively, higher than 30.6% by MetaPSICOV on CASP10 dataset, 34% by MetaPSICOV on CASP11 dataset and 46.3% by Raptor-X on CASP12 dataset, when top L/5 long-range contacts are evaluated. We attribute the improved performance of DNCON2 to the inclusion of short- and medium-range contacts into training, two-level approach to prediction, use of the state-of-the-art optimization and activation functions, and a novel deep learning architecture that allows each filter in a convolutional layer to access all the input features of a protein of arbitrary length. The web server of DNCON2 is at http://sysbio.rnet.missouri.edu/dncon2/ where training and testing datasets as well as the predictions for CASP10, 11 and 12 free-modeling datasets can also be downloaded. Its source code is available at https://github.com/multicom-toolbox/DNCON2/. chengji@missouri.edu. Supplementary data are available at Bioinformatics online.

Organic carbon accumulation in modern sediments of the Angola basin influenced by the Congo deep-sea fan

NASA Astrophysics Data System (ADS)

Baudin, François; Martinez, Philippe; Dennielou, Bernard; Charlier, Karine; Marsset, Tania; Droz, Laurence; Rabouille, Christophe

2017-08-01

Geochemical data (total organic carbon-TOC content, δ13Corg, C:N, Rock-Eval analyses) were obtained on 150 core tops from the Angola basin, with a special focus on the Congo deep-sea fan. Combined with the previously published data, the resulting dataset (322 stations) shows a good spatial and bathymetric representativeness. TOC content and δ13Corg maps of the Angola basin were generated using this enhanced dataset. The main difference in our map with previously published ones is the high terrestrial organic matter content observed downslope along the active turbidite channel of the Congo deep-sea fan till the distal lobe complex near 5000 m of water-depth. Interpretation of downslope trends in TOC content and organic matter composition indicates that lateral particle transport by turbidity currents is the primary mechanism controlling supply and burial of organic matter in the bathypelagic depths.

Performing data analytics on information obtained from various sensors on an OSUS compliant system

NASA Astrophysics Data System (ADS)

Cashion, Kelly; Landoll, Darian; Klawon, Kevin; Powar, Nilesh

2017-05-01

The Open Standard for Unattended Sensors (OSUS) was developed by DIA and ARL to provide a plug-n-play platform for sensor interoperability. Our objective is to use the standardized data produced by OSUS in performing data analytics on information obtained from various sensors. Data analytics can be integrated in one of three ways: within an asset itself; as an independent plug-in designed for one type of asset (i.e. camera or seismic sensor); or as an independent plug-in designed to incorporate data from multiple assets. As a proof-of-concept, we develop a model that can be used in the second of these types - an independent component for camera images. The dataset used was collected as part of a demonstration and test of OSUS capabilities. The image data includes images of empty outdoor scenes and scenes with human or vehicle activity. We design, test, and train a convolution neural network (CNN) to analyze these images and assess the presence of activity in the image. The resulting classifier labels input images as empty or activity with 86.93% accuracy, demonstrating the promising opportunities for deep learning, machine learning, and predictive analytics as an extension of OSUS's already robust suite of capabilities.

DeepNAT: Deep convolutional neural network for segmenting neuroanatomy.

PubMed

Wachinger, Christian; Reuter, Martin; Klein, Tassilo

2018-04-15

We introduce DeepNAT, a 3D Deep convolutional neural network for the automatic segmentation of NeuroAnaTomy in T1-weighted magnetic resonance images. DeepNAT is an end-to-end learning-based approach to brain segmentation that jointly learns an abstract feature representation and a multi-class classification. We propose a 3D patch-based approach, where we do not only predict the center voxel of the patch but also neighbors, which is formulated as multi-task learning. To address a class imbalance problem, we arrange two networks hierarchically, where the first one separates foreground from background, and the second one identifies 25 brain structures on the foreground. Since patches lack spatial context, we augment them with coordinates. To this end, we introduce a novel intrinsic parameterization of the brain volume, formed by eigenfunctions of the Laplace-Beltrami operator. As network architecture, we use three convolutional layers with pooling, batch normalization, and non-linearities, followed by fully connected layers with dropout. The final segmentation is inferred from the probabilistic output of the network with a 3D fully connected conditional random field, which ensures label agreement between close voxels. The roughly 2.7million parameters in the network are learned with stochastic gradient descent. Our results show that DeepNAT compares favorably to state-of-the-art methods. Finally, the purely learning-based method may have a high potential for the adaptation to young, old, or diseased brains by fine-tuning the pre-trained network with a small training sample on the target application, where the availability of larger datasets with manual annotations may boost the overall segmentation accuracy in the future. Copyright © 2017 Elsevier Inc. All rights reserved.

MIDG-Emerging grid technologies for multi-site preclinical molecular imaging research communities.

PubMed

Lee, Jasper; Documet, Jorge; Liu, Brent; Park, Ryan; Tank, Archana; Huang, H K

2011-03-01

Molecular imaging is the visualization and identification of specific molecules in anatomy for insight into metabolic pathways, tissue consistency, and tracing of solute transport mechanisms. This paper presents the Molecular Imaging Data Grid (MIDG) which utilizes emerging grid technologies in preclinical molecular imaging to facilitate data sharing and discovery between preclinical molecular imaging facilities and their collaborating investigator institutions to expedite translational sciences research. Grid-enabled archiving, management, and distribution of animal-model imaging datasets help preclinical investigators to monitor, access and share their imaging data remotely, and promote preclinical imaging facilities to share published imaging datasets as resources for new investigators. The system architecture of the Molecular Imaging Data Grid is described in a four layer diagram. A data model for preclinical molecular imaging datasets is also presented based on imaging modalities currently used in a molecular imaging center. The MIDG system components and connectivity are presented. And finally, the workflow steps for grid-based archiving, management, and retrieval of preclincial molecular imaging data are described. Initial performance tests of the Molecular Imaging Data Grid system have been conducted at the USC IPILab using dedicated VMware servers. System connectivity, evaluated datasets, and preliminary results are presented. The results show the system's feasibility, limitations, direction of future research. Translational and interdisciplinary research in medicine is increasingly interested in cellular and molecular biology activity at the preclinical levels, utilizing molecular imaging methods on animal models. The task of integrated archiving, management, and distribution of these preclinical molecular imaging datasets at preclinical molecular imaging facilities is challenging due to disparate imaging systems and multiple off-site investigators. A Molecular Imaging Data Grid design, implementation, and initial evaluation is presented to demonstrate the secure and novel data grid solution for sharing preclinical molecular imaging data across the wide-area-network (WAN).

Breast mass detection in mammography and tomosynthesis via fully convolutional network-based heatmap regression

NASA Astrophysics Data System (ADS)

Zhang, Jun; Cain, Elizabeth Hope; Saha, Ashirbani; Zhu, Zhe; Mazurowski, Maciej A.

2018-02-01

Breast mass detection in mammography and digital breast tomosynthesis (DBT) is an essential step in computerized breast cancer analysis. Deep learning-based methods incorporate feature extraction and model learning into a unified framework and have achieved impressive performance in various medical applications (e.g., disease diagnosis, tumor detection, and landmark detection). However, these methods require large-scale accurately annotated data. Unfortunately, it is challenging to get precise annotations of breast masses. To address this issue, we propose a fully convolutional network (FCN) based heatmap regression method for breast mass detection, using only weakly annotated mass regions in mammography images. Specifically, we first generate heat maps of masses based on human-annotated rough regions for breast masses. We then develop an FCN model for end-to-end heatmap regression with an F-score loss function, where the mammography images are regarded as the input and heatmaps for breast masses are used as the output. Finally, the probability map of mass locations can be estimated with the trained model. Experimental results on a mammography dataset with 439 subjects demonstrate the effectiveness of our method. Furthermore, we evaluate whether we can use mammography data to improve detection models for DBT, since mammography shares similar structure with tomosynthesis. We propose a transfer learning strategy by fine-tuning the learned FCN model from mammography images. We test this approach on a small tomosynthesis dataset with only 40 subjects, and we show an improvement in the detection performance as compared to training the model from scratch.

Discriminating Projections for Estimating Face Age in Wild Images

DOE Office of Scientific and Technical Information (OSTI.GOV)

Tokola, Ryan A; Bolme, David S; Ricanek, Karl

2014-01-01

We introduce a novel approach to estimating the age of a human from a single uncontrolled image. Current face age estimation algorithms work well in highly controlled images, and some are robust to changes in illumination, but it is usually assumed that images are close to frontal. This bias is clearly seen in the datasets that are commonly used to evaluate age estimation, which either entirely or mostly consist of frontal images. Using pose-specific projections, our algorithm maps image features into a pose-insensitive latent space that is discriminative with respect to age. Age estimation is then performed using a multi-classmore » SVM. We show that our approach outperforms other published results on the Images of Groups dataset, which is the only age-related dataset with a non-trivial number of off-axis face images, and that we are competitive with recent age estimation algorithms on the mostly-frontal FG-NET dataset. We also experimentally demonstrate that our feature projections introduce insensitivity to pose.« less

a Novel Framework for Remote Sensing Image Scene Classification

NASA Astrophysics Data System (ADS)

Jiang, S.; Zhao, H.; Wu, W.; Tan, Q.

2018-04-01

High resolution remote sensing (HRRS) images scene classification aims to label an image with a specific semantic category. HRRS images contain more details of the ground objects and their spatial distribution patterns than low spatial resolution images. Scene classification can bridge the gap between low-level features and high-level semantics. It can be applied in urban planning, target detection and other fields. This paper proposes a novel framework for HRRS images scene classification. This framework combines the convolutional neural network (CNN) and XGBoost, which utilizes CNN as feature extractor and XGBoost as a classifier. Then, this framework is evaluated on two different HRRS images datasets: UC-Merced dataset and NWPU-RESISC45 dataset. Our framework achieved satisfying accuracies on two datasets, which is 95.57 % and 83.35 % respectively. From the experiments result, our framework has been proven to be effective for remote sensing images classification. Furthermore, we believe this framework will be more practical for further HRRS scene classification, since it costs less time on training stage.

Using deep learning in image hyper spectral segmentation, classification, and detection

NASA Astrophysics Data System (ADS)

Zhao, Xiuying; Su, Zhenyu

2018-02-01

Recent years have shown that deep learning neural networks are a valuable tool in the field of computer vision. Deep learning method can be used in applications like remote sensing such as Land cover Classification, Detection of Vehicle in Satellite Images, Hyper spectral Image classification. This paper addresses the use of the deep learning artificial neural network in Satellite image segmentation. Image segmentation plays an important role in image processing. The hue of the remote sensing image often has a large hue difference, which will result in the poor display of the images in the VR environment. Image segmentation is a pre processing technique applied to the original images and splits the image into many parts which have different hue to unify the color. Several computational models based on supervised, unsupervised, parametric, probabilistic region based image segmentation techniques have been proposed. Recently, one of the machine learning technique known as, deep learning with convolution neural network has been widely used for development of efficient and automatic image segmentation models. In this paper, we focus on study of deep neural convolution network and its variants for automatic image segmentation rather than traditional image segmentation strategies.

A deep (learning) dive into visual search behaviour of breast radiologists

NASA Astrophysics Data System (ADS)

Mall, Suneeta; Brennan, Patrick C.; Mello-Thoms, Claudia

2018-03-01

Visual search, the process of detecting and identifying objects using the eye movements (saccades) and the foveal vision, has been studied for identification of root causes of errors in the interpretation of mammography. The aim of this study is to model visual search behaviour of radiologists and their interpretation of mammograms using deep machine learning approaches. Our model is based on a deep convolutional neural network, a biologically-inspired multilayer perceptron that simulates the visual cortex, and is reinforced with transfer learning techniques. Eye tracking data obtained from 8 radiologists (of varying experience levels in reading mammograms) reviewing 120 two-view digital mammography cases (59 cancers) have been used to train the model, which was pre-trained with the ImageNet dataset for transfer learning. Areas of the mammogram that received direct (foveally fixated), indirect (peripherally fixated) or no (never fixated) visual attention were extracted from radiologists' visual search maps (obtained by a head mounted eye tracking device). These areas, along with the radiologists' assessment (including confidence of the assessment) of suspected malignancy were used to model: 1) Radiologists' decision; 2) Radiologists' confidence on such decision; and 3) The attentional level (i.e. foveal, peripheral or none) obtained by an area of the mammogram. Our results indicate high accuracy and low misclassification in modelling such behaviours.

Galaxy Classifications with Deep Learning

NASA Astrophysics Data System (ADS)

Lukic, Vesna; Brüggen, Marcus

2017-06-01

Machine learning techniques have proven to be increasingly useful in astronomical applications over the last few years, for example in object classification, estimating redshifts and data mining. One example of object classification is classifying galaxy morphology. This is a tedious task to do manually, especially as the datasets become larger with surveys that have a broader and deeper search-space. The Kaggle Galaxy Zoo competition presented the challenge of writing an algorithm to find the probability that a galaxy belongs in a particular class, based on SDSS optical spectroscopy data. The use of convolutional neural networks (convnets), proved to be a popular solution to the problem, as they have also produced unprecedented classification accuracies in other image databases such as the database of handwritten digits (MNIST †) and large database of images (CIFAR ‡). We experiment with the convnets that comprised the winning solution, but using broad classifications. The effect of changing the number of layers is explored, as well as using a different activation function, to help in developing an intuition of how the networks function and to see how they can be applied to radio galaxy images.

Stress inversion using borehole images and geometry evolution of the fractures in the Rittershoffen EGS project (Alsace, France).

NASA Astrophysics Data System (ADS)

Azzola, Jérôme; Valley, Benoît; Schmittbuhl, Jean; Genter, Albert; Hehn, Régis

2017-04-01

In the Upper Rhine Graben, several deep geothermal projects based on the Enhanced Geothermal System (EGS) exploit local geothermal reservoirs. The principle underlying this technology consists of increasing the hydraulic performances of the reservoir by stimulating natural fractures using different methods, in order to extract hot water with commercially flow rates. In this domain, the knowledge of the in-situ stress state is of central importance to predict the response of the rock mass to stimulations. Here we characterized the stress tensor from Ultrasonic Borehole Imager (UBI) in the open hole section of the EGS doublet located in Rittershoffen, France (in the Upper Rhine Graben). Interestingly extensive logging programs were leaded at different key moments of the development of the injection well using hydraulic, thermal and chemical stimulation (Baujard et al., 2017)1. The time lapse UBI dataset consists of images of the injection well before, shortly and lastly after the stimulation. The geometry of the induced fractures in compression (breakouts) picked on the UBI images are used to determine the orientation of the in-situ stress tensor. The magnitude of the principal stresses is deduced from the drilling data. The magnitude of the maximum horizontal principal stress is evaluated using an inversion method with three failure criteria (Mohr-Coulomb, Mogi-Coulomb and modified Hoek-Brown criterion) and under the assumption of a vertical or a deviated well. Moreover, the characteristic of the dataset enables the analysis of the evolution of the borehole fracturing, as the deepening or widening of the induced fractures. The correlation of the UBI image allows firstly to determine the tool trajectory and to adapt the post and pre-stimulation images. It secondly leads to the estimation of a complete displacement field which characterizes the deformation induced by the stimulations. Even if the variable image quality deeply conditions the comparison, the considered dataset enabled a consistent evaluation of the in-situ stress field taking place in the underground of the EGS plant, as showed by the comparison to results obtained in the vicinity. Moreover, the study of the evolution of fracturing shows the influence of the stimulation process on the fractures geometry. 1(Baujard, C., Genter, A., Dalmais, E., Maurer, V., Hehn, R., Rosillette, R., Vidal, J., Schmittbuhl, J., 2017). Hydrothermal Characterization of wells GRT-1 and GRT-2 in Rittershoffen, France: Implications on the understanding of natural flow systems in the Rhine Graben, Geothermics, 65, 255-268.

Segmentation of histological images and fibrosis identification with a convolutional neural network.

PubMed

Fu, Xiaohang; Liu, Tong; Xiong, Zhaohan; Smaill, Bruce H; Stiles, Martin K; Zhao, Jichao

2018-07-01

Segmentation of histological images is one of the most crucial tasks for many biomedical analyses involving quantification of certain tissue types, such as fibrosis via Masson's trichrome staining. However, challenges are posed by the high variability and complexity of structural features in such images, in addition to imaging artifacts. Further, the conventional approach of manual thresholding is labor-intensive, and highly sensitive to inter- and intra-image intensity variations. An accurate and robust automated segmentation method is of high interest. We propose and evaluate an elegant convolutional neural network (CNN) designed for segmentation of histological images, particularly those with Masson's trichrome stain. The network comprises 11 successive convolutional - rectified linear unit - batch normalization layers. It outperformed state-of-the-art CNNs on a dataset of cardiac histological images (labeling fibrosis, myocytes, and background) with a Dice similarity coefficient of 0.947. With 100 times fewer (only 300,000) trainable parameters than the state-of-the-art, our CNN is less susceptible to overfitting, and is efficient. Additionally, it retains image resolution from input to output, captures fine-grained details, and can be trained end-to-end smoothly. To the best of our knowledge, this is the first deep CNN tailored to the problem of concern, and may potentially be extended to solve similar segmentation tasks to facilitate investigations into pathology and clinical treatment. Copyright © 2018 Elsevier Ltd. All rights reserved.

High Contrast Imaging of Exoplanets and Exoplanetary Systems with JWST

NASA Astrophysics Data System (ADS)

Hinkley, Sasha; Skemer, Andrew; Biller, Beth; Baraffe, I.; Bonnefoy, M.; Bowler, B.; Carter, A.; Chen, C.; Choquet, E.; Currie, T.; Danielski, C.; Fortney, J.; Grady, C.; Greenbaum, A.; Hines, D.; Janson, M.; Kalas, P.; Kennedy, G.; Kraus, A.; Lagrange, A.; Liu, M.; Marley, M.; Marois, C.; Matthews, B.; Mawet, D.; Metchev, S.; Meyer, M.; Millar-Blanchaer, M.; Perrin, M.; Pueyo, L.; Quanz, S.; Rameau, J.; Rodigas, T.; Sallum, S.; Sargent, B.; Schlieder, J.; Schneider, G.; Stapelfeldt, K.; Tremblin, P.; Vigan, A.; Ygouf, M.

2017-11-01

JWST will transform our ability to characterize directly imaged planets and circumstellar debris disks, including the first spectroscopic characterization of directly imaged exoplanets at wavelengths beyond 5 microns, providing a powerful diagnostic of cloud particle properties, atmospheric structure, and composition. To lay the groundwork for these science goals, we propose a 39-hour ERS program to rapidly establish optimal strategies for JWST high contrast imaging. We will acquire: a) coronagraphic imaging of a newly discovered exoplanet companion, and a well-studied circumstellar debris disk with NIRCam & MIRI; b) spectroscopy of a wide separation planetary mass companion with NIRSPEC & MIRI; and c) deep aperture masking interferometry with NIRISS. Our primary goals are to: 1) generate representative datasets in modes to be commonly used by the exoplanet and disk imaging communities; 2) deliver science enabling products to empower a broad user base to develop successful future investigations; and 3) carry out breakthrough science by characterizing exoplanets for the first time over their full spectral range from 2-28 microns, and debris disk spectrophotometry out to 15 microns sampling the 3 micron water ice feature. Our team represents the majority of the community dedicated to exoplanet and disk imaging and has decades of experience with high contrast imaging algorithms and pipelines. We have developed a collaboration management plan and several organized working groups to ensure we can rapidly and effectively deliver high quality Science Enabling Products to the community.

Weak lensing Study in VOICE Survey I: Shear Measurement

NASA Astrophysics Data System (ADS)

Fu, Liping; Liu, Dezi; Radovich, Mario; Liu, Xiangkun; Pan, Chuzhong; Fan, Zuhui; Covone, Giovanni; Vaccari, Mattia; Amaro, Valeria; Brescia, Massimo; Capaccioli, Massimo; De Cicco, Demetra; Grado, Aniello; Limatola, Luca; Miller, Lance; Napolitano, Nicola R.; Paolillo, Maurizio; Pignata, Giuliano

2018-06-01

The VST Optical Imaging of the CDFS and ES1 Fields (VOICE) Survey is a Guaranteed Time program carried out with the ESO/VST telescope to provide deep optical imaging over two 4 deg2 patches of the sky centred on the CDFS and ES1 pointings. We present the cosmic shear measurement over the 4 deg2 covering the CDFS region in the r-band using LensFit. Each of the four tiles of 1 deg2 has more than one hundred exposures, of which more than 50 exposures passed a series of image quality selection criteria for weak lensing study. The 5σ limiting magnitude in r- band is 26.1 for point sources, which is ≳1 mag deeper than other weak lensing survey in the literature (e.g. the Kilo Degree Survey, KiDS, at VST). The photometric redshifts are estimated using the VOICE u, g, r, i together with near-infrared VIDEO data Y, J, H, Ks. The mean redshift of the shear catalogue is 0.87, considering the shear weight. The effective galaxy number density is 16.35 gal/arcmin2, which is nearly twice the one of KiDS. The performance of LensFit on such a deep dataset was calibrated using VOICE-like mock image simulations. Furthermore, we have analyzed the reliability of the shear catalogue by calculating the star-galaxy cross-correlations, the tomographic shear correlations of two redshift bins and the contaminations of the blended galaxies. As a further sanity check, we have constrained cosmological parameters by exploring the parameter space with Population Monte Carlo sampling. For a flat ΛCDM model we have obtained Σ _8 = σ _8(Ω _m/0.3)^{0.5} = 0.68^{+0.11}_{-0.15}.

Deep learning for segmentation of brain tumors: can we train with images from different institutions?

NASA Astrophysics Data System (ADS)

Paredes, David; Saha, Ashirbani; Mazurowski, Maciej A.

2017-03-01

Deep learning and convolutional neural networks (CNNs) in particular are increasingly popular tools for segmentation and classification of medical images. CNNs were shown to be successful for segmentation of brain tumors into multiple regions or labels. However, in the environment which fosters data-sharing and collection of multi-institutional datasets, a question arises: does training with data from another institution with potentially different imaging equipment, contrast protocol, and patient population impact the segmentation performance of the CNN? Our study presents preliminary data towards answering this question. Specifically, we used MRI data of glioblastoma (GBM) patients for two institutions present in The Cancer Imaging Archive. We performed a process of training and testing CNN multiple times such that half of the time the CNN was tested on data from the same institution that was used for training and half of the time it was tested on another institution, keeping the training and testing set size constant. We observed a decrease in performance as measured by Dice coefficient when the CNN was trained with data from a different institution as compared to training with data from the same institution. The changes in performance for the entire tumor and for four different labels within the tumor were: 0.72 to 0.65 (p=0.06), 0.61 to 0.58 (p=0.49), 0.54 to 0.51 (p=0.82), 0.31 to 0.24 (p<0.03), and 0.43 to 0.31(p<0.003) respectively. In summary, we found that while data across institutions can be used for development of CNNs, this might be associated with a decrease in performance.

Identification and Removal of Contaminant Sequences From Ribosomal Gene Databases: Lessons From the Census of Deep Life

PubMed Central

Sheik, Cody S.; Reese, Brandi Kiel; Twing, Katrina I.; Sylvan, Jason B.; Grim, Sharon L.; Schrenk, Matthew O.; Sogin, Mitchell L.; Colwell, Frederick S.

2018-01-01

Earth’s subsurface environment is one of the largest, yet least studied, biomes on Earth, and many questions remain regarding what microorganisms are indigenous to the subsurface. Through the activity of the Census of Deep Life (CoDL) and the Deep Carbon Observatory, an open access 16S ribosomal RNA gene sequence database from diverse subsurface environments has been compiled. However, due to low quantities of biomass in the deep subsurface, the potential for incorporation of contaminants from reagents used during sample collection, processing, and/or sequencing is high. Thus, to understand the ecology of subsurface microorganisms (i.e., the distribution, richness, or survival), it is necessary to minimize, identify, and remove contaminant sequences that will skew the relative abundances of all taxa in the sample. In this meta-analysis, we identify putative contaminants associated with the CoDL dataset, recommend best practices for removing contaminants from samples, and propose a series of best practices for subsurface microbiology sampling. The most abundant putative contaminant genera observed, independent of evenness across samples, were Propionibacterium, Aquabacterium, Ralstonia, and Acinetobacter. While the top five most frequently observed genera were Pseudomonas, Propionibacterium, Acinetobacter, Ralstonia, and Sphingomonas. The majority of the most frequently observed genera (high evenness) were associated with reagent or potential human contamination. Additionally, in DNA extraction blanks, we observed potential archaeal contaminants, including methanogens, which have not been discussed in previous contamination studies. Such contaminants would directly affect the interpretation of subsurface molecular studies, as methanogenesis is an important subsurface biogeochemical process. Utilizing previously identified contaminant genera, we found that ∼27% of the total dataset were identified as contaminant sequences that likely originate from DNA extraction and DNA cleanup methods. Thus, controls must be taken at every step of the collection and processing procedure when working with low biomass environments such as, but not limited to, portions of Earth’s deep subsurface. Taken together, we stress that the CoDL dataset is an incredible resource for the broader research community interested in subsurface life, and steps to remove contamination derived sequences must be taken prior to using this dataset. PMID:29780369

Identification and Removal of Contaminant Sequences From Ribosomal Gene Databases: Lessons From the Census of Deep Life.

PubMed

Sheik, Cody S; Reese, Brandi Kiel; Twing, Katrina I; Sylvan, Jason B; Grim, Sharon L; Schrenk, Matthew O; Sogin, Mitchell L; Colwell, Frederick S

2018-01-01

Earth's subsurface environment is one of the largest, yet least studied, biomes on Earth, and many questions remain regarding what microorganisms are indigenous to the subsurface. Through the activity of the Census of Deep Life (CoDL) and the Deep Carbon Observatory, an open access 16S ribosomal RNA gene sequence database from diverse subsurface environments has been compiled. However, due to low quantities of biomass in the deep subsurface, the potential for incorporation of contaminants from reagents used during sample collection, processing, and/or sequencing is high. Thus, to understand the ecology of subsurface microorganisms (i.e., the distribution, richness, or survival), it is necessary to minimize, identify, and remove contaminant sequences that will skew the relative abundances of all taxa in the sample. In this meta-analysis, we identify putative contaminants associated with the CoDL dataset, recommend best practices for removing contaminants from samples, and propose a series of best practices for subsurface microbiology sampling. The most abundant putative contaminant genera observed, independent of evenness across samples, were Propionibacterium , Aquabacterium , Ralstonia , and Acinetobacter . While the top five most frequently observed genera were Pseudomonas , Propionibacterium , Acinetobacter , Ralstonia , and Sphingomonas . The majority of the most frequently observed genera (high evenness) were associated with reagent or potential human contamination. Additionally, in DNA extraction blanks, we observed potential archaeal contaminants, including methanogens, which have not been discussed in previous contamination studies. Such contaminants would directly affect the interpretation of subsurface molecular studies, as methanogenesis is an important subsurface biogeochemical process. Utilizing previously identified contaminant genera, we found that ∼27% of the total dataset were identified as contaminant sequences that likely originate from DNA extraction and DNA cleanup methods. Thus, controls must be taken at every step of the collection and processing procedure when working with low biomass environments such as, but not limited to, portions of Earth's deep subsurface. Taken together, we stress that the CoDL dataset is an incredible resource for the broader research community interested in subsurface life, and steps to remove contamination derived sequences must be taken prior to using this dataset.

Discovery of bright z ≃ 7 galaxies in the UltraVISTA survey

NASA Astrophysics Data System (ADS)

Bowler, R. A. A.; Dunlop, J. S.; McLure, R. J.; McCracken, H. J.; Milvang-Jensen, B.; Furusawa, H.; Fynbo, J. P. U.; Le Fèvre, O.; Holt, J.; Ideue, Y.; Ihara, Y.; Rogers, A. B.; Taniguchi, Y.

2012-11-01

We have exploited the new, deep, near-infrared UltraVISTA imaging of the Cosmological Evolution Survey (COSMOS) field, in tandem with deep optical and mid-infrared imaging, to conduct a new search for luminous galaxies at redshifts z ≃ 7. The year-one UltraVISTA data provide contiguous Y, J, H, Ks imaging over 1.5 deg2, reaching a 5σ detection limit of Y + J ≃ 25 (AB mag, 2-arcsec-diameter aperture). The central ≃1 deg2 of this imaging coincides with the final deep optical (u*, g, r, i) data provided by the Canada-France-Hawaii Telescope (CFHT) Legacy Survey and new deep Subaru/Suprime-Cam z'-band imaging obtained specifically to enable full exploitation of UltraVISTA. It also lies within the Hubble Space Telescope (HST) I814 band and Spitzer/Infrared Array Camera imaging obtained as part of the COSMOS survey. We have utilized this unique multiwavelength dataset to select galaxy candidates at redshifts z > 6.5 by searching first for Y + J-detected objects which are undetected in the CFHT and HST optical data. This sample was then refined using a photometric redshift fitting code, enabling the rejection of lower redshift galaxy contaminants and cool galactic M, L, T dwarf stars. The final result of this process is a small sample of (at most) 10 credible galaxy candidates at z > 6.5 (from over 200 000 galaxies detected in the year-one UltraVISTA data) which we present in this paper. The first four of these appear to be robust galaxies at z > 6.5, and fitting to their stacked spectral energy distribution yields zphot = 6.98 ± 0.05 with a stellar mass M* ≃ 5 × 109 M⊙ and rest-frame ultraviolet (UV) spectral slope β ≃ -2.0 ± 0.2 (where fλ ∝ λβ). The next three are also good candidates for z > 6.5 galaxies, but the possibility that they are dwarf stars cannot be completely excluded. Our final subset of three additional candidates is afflicted not only by potential dwarf star contamination, but also contains objects likely to lie at redshifts just below z = 6.5. We show that the three even-brighter z ≳ 7 galaxy candidates reported in the COSMOS field by Capak et al. are in fact all lower redshift galaxies at z ≃ 1.5-3.5. Consequently the new z ≃ 7 galaxies reported here are the first credible z ≃ 7 Lyman-break galaxies discovered in the COSMOS field and, as the most UV luminous discovered to date at these redshifts, are prime targets for deep follow-up spectroscopy. We explore their physical properties, and briefly consider the implications of their inferred number density for the form of the galaxy luminosity function at z ≃ 7.

Automatic labeling of MR brain images through extensible learning and atlas forests.

PubMed

Xu, Lijun; Liu, Hong; Song, Enmin; Yan, Meng; Jin, Renchao; Hung, Chih-Cheng

2017-12-01

Multiatlas-based method is extensively used in MR brain images segmentation because of its simplicity and robustness. This method provides excellent accuracy although it is time consuming and limited in terms of obtaining information about new atlases. In this study, an automatic labeling of MR brain images through extensible learning and atlas forest is presented to address these limitations. We propose an extensible learning model which allows the multiatlas-based framework capable of managing the datasets with numerous atlases or dynamic atlas datasets and simultaneously ensure the accuracy of automatic labeling. Two new strategies are used to reduce the time and space complexity and improve the efficiency of the automatic labeling of brain MR images. First, atlases are encoded to atlas forests through random forest technology to reduce the time consumed for cross-registration between atlases and target image, and a scatter spatial vector is designed to eliminate errors caused by inaccurate registration. Second, an atlas selection method based on the extensible learning model is used to select atlases for target image without traversing the entire dataset and then obtain the accurate labeling. The labeling results of the proposed method were evaluated in three public datasets, namely, IBSR, LONI LPBA40, and ADNI. With the proposed method, the dice coefficient metric values on the three datasets were 84.17 ± 4.61%, 83.25 ± 4.29%, and 81.88 ± 4.53% which were 5% higher than those of the conventional method, respectively. The efficiency of the extensible learning model was evaluated by state-of-the-art methods for labeling of MR brain images. Experimental results showed that the proposed method could achieve accurate labeling for MR brain images without traversing the entire datasets. In the proposed multiatlas-based method, extensible learning and atlas forests were applied to control the automatic labeling of brain anatomies on large atlas datasets or dynamic atlas datasets and obtain accurate results. © 2017 American Association of Physicists in Medicine.

DeepSynergy: predicting anti-cancer drug synergy with Deep Learning

PubMed Central

Preuer, Kristina; Lewis, Richard P I; Hochreiter, Sepp; Bender, Andreas; Bulusu, Krishna C; Klambauer, Günter

2018-01-01

Abstract Motivation While drug combination therapies are a well-established concept in cancer treatment, identifying novel synergistic combinations is challenging due to the size of combinatorial space. However, computational approaches have emerged as a time- and cost-efficient way to prioritize combinations to test, based on recently available large-scale combination screening data. Recently, Deep Learning has had an impact in many research areas by achieving new state-of-the-art model performance. However, Deep Learning has not yet been applied to drug synergy prediction, which is the approach we present here, termed DeepSynergy. DeepSynergy uses chemical and genomic information as input information, a normalization strategy to account for input data heterogeneity, and conical layers to model drug synergies. Results DeepSynergy was compared to other machine learning methods such as Gradient Boosting Machines, Random Forests, Support Vector Machines and Elastic Nets on the largest publicly available synergy dataset with respect to mean squared error. DeepSynergy significantly outperformed the other methods with an improvement of 7.2% over the second best method at the prediction of novel drug combinations within the space of explored drugs and cell lines. At this task, the mean Pearson correlation coefficient between the measured and the predicted values of DeepSynergy was 0.73. Applying DeepSynergy for classification of these novel drug combinations resulted in a high predictive performance of an AUC of 0.90. Furthermore, we found that all compared methods exhibit low predictive performance when extrapolating to unexplored drugs or cell lines, which we suggest is due to limitations in the size and diversity of the dataset. We envision that DeepSynergy could be a valuable tool for selecting novel synergistic drug combinations. Availability and implementation DeepSynergy is available via www.bioinf.jku.at/software/DeepSynergy. Contact klambauer@bioinf.jku.at Supplementary information Supplementary data are available at Bioinformatics online. PMID:29253077

Reference-tissue correction of T2-weighted signal intensity for prostate cancer detection

NASA Astrophysics Data System (ADS)

Peng, Yahui; Jiang, Yulei; Oto, Aytekin

2014-03-01

The purpose of this study was to investigate whether correction with respect to reference tissue of T2-weighted MRimage signal intensity (SI) improves its effectiveness for classification of regions of interest (ROIs) as prostate cancer (PCa) or normal prostatic tissue. Two image datasets collected retrospectively were used in this study: 71 cases acquired with GE scanners (dataset A), and 59 cases acquired with Philips scanners (dataset B). Through a consensus histology- MR correlation review, 175 PCa and 108 normal-tissue ROIs were identified and drawn manually. Reference-tissue ROIs were selected in each case from the levator ani muscle, urinary bladder, and pubic bone. T2-weighted image SI was corrected as the ratio of the average T2-weighted image SI within an ROI to that of a reference-tissue ROI. Area under the receiver operating characteristic curve (AUC) was used to evaluate the effectiveness of T2-weighted image SIs for differentiation of PCa from normal-tissue ROIs. AUC (+/- standard error) for uncorrected T2-weighted image SIs was 0.78+/-0.04 (datasets A) and 0.65+/-0.05 (datasets B). AUC for corrected T2-weighted image SIs with respect to muscle, bladder, and bone reference was 0.77+/-0.04 (p=1.0), 0.77+/-0.04 (p=1.0), and 0.75+/-0.04 (p=0.8), respectively, for dataset A; and 0.81+/-0.04 (p=0.002), 0.78+/-0.04 (p<0.001), and 0.79+/-0.04 (p<0.001), respectively, for dataset B. Correction in reference to the levator ani muscle yielded the most consistent results between GE and Phillips images. Correction of T2-weighted image SI in reference to three types of extra-prostatic tissue can improve its effectiveness for differentiation of PCa from normal-tissue ROIs, and correction in reference to the levator ani muscle produces consistent T2-weighted image SIs between GE and Phillips MR images.

Automated generation of massive image knowledge collections using Microsoft Live Labs Pivot to promote neuroimaging and translational research.

PubMed

Viangteeravat, Teeradache; Anyanwu, Matthew N; Ra Nagisetty, Venkateswara; Kuscu, Emin

2011-07-15

Massive datasets comprising high-resolution images, generated in neuro-imaging studies and in clinical imaging research, are increasingly challenging our ability to analyze, share, and filter such images in clinical and basic translational research. Pivot collection exploratory analysis provides each user the ability to fully interact with the massive amounts of visual data to fully facilitate sufficient sorting, flexibility and speed to fluidly access, explore or analyze the massive image data sets of high-resolution images and their associated meta information, such as neuro-imaging databases from the Allen Brain Atlas. It is used in clustering, filtering, data sharing and classifying of the visual data into various deep zoom levels and meta information categories to detect the underlying hidden pattern within the data set that has been used. We deployed prototype Pivot collections using the Linux CentOS running on the Apache web server. We also tested the prototype Pivot collections on other operating systems like Windows (the most common variants) and UNIX, etc. It is demonstrated that the approach yields very good results when compared with other approaches used by some researchers for generation, creation, and clustering of massive image collections such as the coronal and horizontal sections of the mouse brain from the Allen Brain Atlas. Pivot visual analytics was used to analyze a prototype of dataset Dab2 co-expressed genes from the Allen Brain Atlas. The metadata along with high-resolution images were automatically extracted using the Allen Brain Atlas API. It is then used to identify the hidden information based on the various categories and conditions applied by using options generated from automated collection. A metadata category like chromosome, as well as data for individual cases like sex, age, and plan attributes of a particular gene, is used to filter, sort and to determine if there exist other genes with a similar characteristics to Dab2. And online access to the mouse brain pivot collection can be viewed using the link http://edtech-dev.uthsc.edu/CTSI/teeDev1/unittest/PaPa/collection.html (user name: tviangte and password: demome) Our proposed algorithm has automated the creation of large image Pivot collections; this will enable investigators of clinical research projects to easily and quickly analyse the image collections through a perspective that is useful for making critical decisions about the image patterns discovered.

Bayesian convolutional neural network based MRI brain extraction on nonhuman primates.

PubMed

Zhao, Gengyan; Liu, Fang; Oler, Jonathan A; Meyerand, Mary E; Kalin, Ned H; Birn, Rasmus M

2018-07-15

Brain extraction or skull stripping of magnetic resonance images (MRI) is an essential step in neuroimaging studies, the accuracy of which can severely affect subsequent image processing procedures. Current automatic brain extraction methods demonstrate good results on human brains, but are often far from satisfactory on nonhuman primates, which are a necessary part of neuroscience research. To overcome the challenges of brain extraction in nonhuman primates, we propose a fully-automated brain extraction pipeline combining deep Bayesian convolutional neural network (CNN) and fully connected three-dimensional (3D) conditional random field (CRF). The deep Bayesian CNN, Bayesian SegNet, is used as the core segmentation engine. As a probabilistic network, it is not only able to perform accurate high-resolution pixel-wise brain segmentation, but also capable of measuring the model uncertainty by Monte Carlo sampling with dropout in the testing stage. Then, fully connected 3D CRF is used to refine the probability result from Bayesian SegNet in the whole 3D context of the brain volume. The proposed method was evaluated with a manually brain-extracted dataset comprising T1w images of 100 nonhuman primates. Our method outperforms six popular publicly available brain extraction packages and three well-established deep learning based methods with a mean Dice coefficient of 0.985 and a mean average symmetric surface distance of 0.220 mm. A better performance against all the compared methods was verified by statistical tests (all p-values < 10 -4 , two-sided, Bonferroni corrected). The maximum uncertainty of the model on nonhuman primate brain extraction has a mean value of 0.116 across all the 100 subjects. The behavior of the uncertainty was also studied, which shows the uncertainty increases as the training set size decreases, the number of inconsistent labels in the training set increases, or the inconsistency between the training set and the testing set increases. Copyright © 2018 Elsevier Inc. All rights reserved.

The Use of the Deep Convective Cloud Technique (DCCT) to Monitor On-Orbit Performance of the Geostationary Lightning Mapper (GLM): Use of Lightning Imaging Sensor (LIS) Data as Proxy

NASA Technical Reports Server (NTRS)

Buechler, Dennis E.; Christian, H. J.; Koshak, William J.; Goodman, Steve J.

2013-01-01

The Geostationary Lightning Mapper (GLM) on the next generation Geostationary Operational Environmental Satellite-R (GOES-R) will not have onboard calibration capability to monitor its performance. The Lightning Imaging Sensor (LIS) onboard the Tropical Rainfall Measuring Mission (TRMM) satellite has been providing observations of total lightning over the Earth's Tropics since 1997. The GLM design is based on LIS heritage, making it a good proxy dataset. This study examines the performance of LIS throughout its time in orbit. This was accomplished through application of the Deep Convective Cloud Technique (DCCT) (Doelling et al., 2004) to LIS background pixel radiance data. The DCCT identifies deep convective clouds by their cold Infrared (IR) brightness temperatures and using them as invariant targets in the solar reflective portion of the solar spectrum. The GLM and LIS operate in the near-IR at a wavelength of 777.4 nm. In the present study the IR data is obtained from the Visible Infrared Sensor (VIRS) which is collocated with LIS onboard the Tropical Rainfall Measuring Mission (TRMM) satellite. The DCCT is applied to LIS observations for July and August of each year from 1998-2010. The resulting distributions of LIS background DCC pixel radiance for each July August are very similar, indicating stable performance. The mean radiance of the DCCT analysis does not show a long term trend and the maximum deviation of the July August mean radiance for each year is within 0.7% of the overall mean. These results demonstrate that there has been no discernible change in LIS performance throughout its lifetime. A similar approach will used for monitoring the performance of GLM, with cold clouds identified using IR data from the Advanced Baseline Imager (ABI) which will also be located on GOES-R. Since GLM is based on LIS design heritage, the LIS results indicate that GLM should also experience stable performance over its lifetime.

Assessment of the NASA-USGS Global Land Survey (GLS) Datasets

USGS Publications Warehouse

Gutman, Garik; Huang, Chengquan; Chander, Gyanesh; Noojipady, Praveen; Masek, Jeffery G.

2013-01-01

The Global Land Survey (GLS) datasets are a collection of orthorectified, cloud-minimized Landsat-type satellite images, providing near complete coverage of the global land area decadally since the early 1970s. The global mosaics are centered on 1975, 1990, 2000, 2005, and 2010, and consist of data acquired from four sensors: Enhanced Thematic Mapper Plus, Thematic Mapper, Multispectral Scanner, and Advanced Land Imager. The GLS datasets have been widely used in land-cover and land-use change studies at local, regional, and global scales. This study evaluates the GLS datasets with respect to their spatial coverage, temporal consistency, geodetic accuracy, radiometric calibration consistency, image completeness, extent of cloud contamination, and residual gaps. In general, the three latest GLS datasets are of a better quality than the GLS-1990 and GLS-1975 datasets, with most of the imagery (85%) having cloud cover of less than 10%, the acquisition years clustered much more tightly around their target years, better co-registration relative to GLS-2000, and better radiometric absolute calibration. Probably, the most significant impediment to scientific use of the datasets is the variability of image phenology (i.e., acquisition day of year). This paper provides end-users with an assessment of the quality of the GLS datasets for specific applications, and where possible, suggestions for mitigating their deficiencies.

Three-Dimensional Anisotropic Acoustic and Elastic Full-Waveform Seismic Inversion

NASA Astrophysics Data System (ADS)

Warner, M.; Morgan, J. V.

2013-12-01

Three-dimensional full-waveform inversion is a high-resolution, high-fidelity, quantitative, seismic imaging technique that has advanced rapidly within the oil and gas industry. The method involves the iterative improvement of a starting model using a series of local linearized updates to solve the full non-linear inversion problem. During the inversion, forward modeling employs the full two-way three-dimensional heterogeneous anisotropic acoustic or elastic wave equation to predict the observed raw field data, wiggle-for-wiggle, trace-by-trace. The method is computationally demanding; it is highly parallelized, and runs on large multi-core multi-node clusters. Here, we demonstrate what can be achieved by applying this newly practical technique to several high-density 3D seismic datasets that were acquired to image four contrasting sedimentary targets: a gas cloud above an oil reservoir, a radially faulted dome, buried fluvial channels, and collapse structures overlying an evaporate sequence. We show that the resulting anisotropic p-wave velocity models match in situ measurements in deep boreholes, reproduce detailed structure observed independently on high-resolution seismic reflection sections, accurately predict the raw seismic data, simplify and sharpen reverse-time-migrated reflection images of deeper horizons, and flatten Kirchhoff-migrated common-image gathers. We also show that full-elastic 3D full-waveform inversion of pure pressure data can generate a reasonable shear-wave velocity model for one of these datasets. For two of the four datasets, the inclusion of significant transversely isotropic anisotropy with a vertical axis of symmetry was necessary in order to fit the kinematics of the field data properly. For the faulted dome, the full-waveform-inversion p-wave velocity model recovers the detailed structure of every fault that can be seen on coincident seismic reflection data. Some of the individual faults represent high-velocity zones, some represent low-velocity zones, some have more-complex internal structure, and some are visible merely as offsets between two regions with contrasting velocity. Although this has not yet been demonstrated quantitatively for this dataset, it seems likely that at least some of this fine structure in the recovered velocity model is related to the detailed lithology, strain history and fluid properties within the individual faults. We have here applied this technique to seismic data that were acquired by the extractive industries, however this inversion scheme is immediately scalable and applicable to a much wider range of problems given sufficient quality and density of observed data. Potential targets range from shallow magma chambers beneath active volcanoes, through whole-crustal sections across plate boundaries, to regional and whole-Earth models.

SiNC: Saliency-injected neural codes for representation and efficient retrieval of medical radiographs

PubMed Central

Sajjad, Muhammad; Mehmood, Irfan; Baik, Sung Wook

2017-01-01

Medical image collections contain a wealth of information which can assist radiologists and medical experts in diagnosis and disease detection for making well-informed decisions. However, this objective can only be realized if efficient access is provided to semantically relevant cases from the ever-growing medical image repositories. In this paper, we present an efficient method for representing medical images by incorporating visual saliency and deep features obtained from a fine-tuned convolutional neural network (CNN) pre-trained on natural images. Saliency detector is employed to automatically identify regions of interest like tumors, fractures, and calcified spots in images prior to feature extraction. Neuronal activation features termed as neural codes from different CNN layers are comprehensively studied to identify most appropriate features for representing radiographs. This study revealed that neural codes from the last fully connected layer of the fine-tuned CNN are found to be the most suitable for representing medical images. The neural codes extracted from the entire image and salient part of the image are fused to obtain the saliency-injected neural codes (SiNC) descriptor which is used for indexing and retrieval. Finally, locality sensitive hashing techniques are applied on the SiNC descriptor to acquire short binary codes for allowing efficient retrieval in large scale image collections. Comprehensive experimental evaluations on the radiology images dataset reveal that the proposed framework achieves high retrieval accuracy and efficiency for scalable image retrieval applications and compares favorably with existing approaches. PMID:28771497

A Deep Learning Approach to on-Node Sensor Data Analytics for Mobile or Wearable Devices.

PubMed

Ravi, Daniele; Wong, Charence; Lo, Benny; Yang, Guang-Zhong

2017-01-01

The increasing popularity of wearable devices in recent years means that a diverse range of physiological and functional data can now be captured continuously for applications in sports, wellbeing, and healthcare. This wealth of information requires efficient methods of classification and analysis where deep learning is a promising technique for large-scale data analytics. While deep learning has been successful in implementations that utilize high-performance computing platforms, its use on low-power wearable devices is limited by resource constraints. In this paper, we propose a deep learning methodology, which combines features learned from inertial sensor data together with complementary information from a set of shallow features to enable accurate and real-time activity classification. The design of this combined method aims to overcome some of the limitations present in a typical deep learning framework where on-node computation is required. To optimize the proposed method for real-time on-node computation, spectral domain preprocessing is used before the data are passed onto the deep learning framework. The classification accuracy of our proposed deep learning approach is evaluated against state-of-the-art methods using both laboratory and real world activity datasets. Our results show the validity of the approach on different human activity datasets, outperforming other methods, including the two methods used within our combined pipeline. We also demonstrate that the computation times for the proposed method are consistent with the constraints of real-time on-node processing on smartphones and a wearable sensor platform.

Corrosion-Fatigue Assessment Program

DTIC Science & Technology

2008-03-31

22 Figure 3.2.1-4 Deep -focus image of Specimen 598-7 – Crack 1...at Feature #2 .........................22 Figure 3.2.1-5 Deep -focus image of Specimen 598-7 – Crack 2 at Feature #5 .........................23...Figure 3.2.1-6 Deep -focus image of Specimen 598-7 – Crack 3 at Feature #3 .........................23 Figure 3.2.1-7 Deep -focus image of Specimen 598-7

A test-retest dataset for assessing long-term reliability of brain morphology and resting-state brain activity.

PubMed

Huang, Lijie; Huang, Taicheng; Zhen, Zonglei; Liu, Jia

2016-03-15

We present a test-retest dataset for evaluation of long-term reliability of measures from structural and resting-state functional magnetic resonance imaging (sMRI and rfMRI) scans. The repeated scan dataset was collected from 61 healthy adults in two sessions using highly similar imaging parameters at an interval of 103-189 days. However, as the imaging parameters were not completely identical, the reliability estimated from this dataset shall reflect the lower bounds of the true reliability of sMRI/rfMRI measures. Furthermore, in conjunction with other test-retest datasets, our dataset may help explore the impact of different imaging parameters on reliability of sMRI/rfMRI measures, which is especially critical for assessing datasets collected from multiple centers. In addition, intelligence quotient (IQ) was measured for each participant using Raven's Advanced Progressive Matrices. The data can thus be used for purposes other than assessing reliability of sMRI/rfMRI alone. For example, data from each single session could be used to associate structural and functional measures of the brain with the IQ metrics to explore brain-IQ association.

Stationary Wavelet Transform and AdaBoost with SVM Based Pathological Brain Detection in MRI Scanning.

PubMed

Nayak, Deepak Ranjan; Dash, Ratnakar; Majhi, Banshidhar

2017-01-01

This paper presents an automatic classification system for segregating pathological brain from normal brains in magnetic resonance imaging scanning. The proposed system employs contrast limited adaptive histogram equalization scheme to enhance the diseased region in brain MR images. Two-dimensional stationary wavelet transform is harnessed to extract features from the preprocessed images. The feature vector is constructed using the energy and entropy values, computed from the level- 2 SWT coefficients. Then, the relevant and uncorrelated features are selected using symmetric uncertainty ranking filter. Subsequently, the selected features are given input to the proposed AdaBoost with support vector machine classifier, where SVM is used as the base classifier of AdaBoost algorithm. To validate the proposed system, three standard MR image datasets, Dataset-66, Dataset-160, and Dataset- 255 have been utilized. The 5 runs of k-fold stratified cross validation results indicate the suggested scheme offers better performance than other existing schemes in terms of accuracy and number of features. The proposed system earns ideal classification over Dataset-66 and Dataset-160; whereas, for Dataset- 255, an accuracy of 99.45% is achieved. Copyright© Bentham Science Publishers; For any queries, please email at epub@benthamscience.org.

Scaling Deep Learning Workloads: NVIDIA DGX-1/Pascal and Intel Knights Landing

DOE Office of Scientific and Technical Information (OSTI.GOV)

Gawande, Nitin A.; Landwehr, Joshua B.; Daily, Jeffrey A.

Deep Learning (DL) algorithms have become ubiquitous in data analytics. As a result, major computing vendors --- including NVIDIA, Intel, AMD and IBM --- have architectural road-maps influenced by DL workloads. Furthermore, several vendors have recently advertised new computing products as accelerating DL workloads. Unfortunately, it is difficult for data scientists to quantify the potential of these different products. This paper provides a performance and power analysis of important DL workloads on two major parallel architectures: NVIDIA DGX-1 (eight Pascal P100 GPUs interconnected with NVLink) and Intel Knights Landing (KNL) CPUs interconnected with Intel Omni-Path. Our evaluation consists of amore » cross section of convolutional neural net workloads: CifarNet, CaffeNet, AlexNet and GoogleNet topologies using the Cifar10 and ImageNet datasets. The workloads are vendor optimized for each architecture. GPUs provide the highest overall raw performance. Our analysis indicates that although GPUs provide the highest overall performance, the gap can close for some convolutional networks; and KNL can be competitive when considering performance/watt. Furthermore, NVLink is critical to GPU scaling.« less

Brain tumor segmentation in multi-spectral MRI using convolutional neural networks (CNN).

PubMed

Iqbal, Sajid; Ghani, M Usman; Saba, Tanzila; Rehman, Amjad

2018-04-01

A tumor could be found in any area of the brain and could be of any size, shape, and contrast. There may exist multiple tumors of different types in a human brain at the same time. Accurate tumor area segmentation is considered primary step for treatment of brain tumors. Deep Learning is a set of promising techniques that could provide better results as compared to nondeep learning techniques for segmenting timorous part inside a brain. This article presents a deep convolutional neural network (CNN) to segment brain tumors in MRIs. The proposed network uses BRATS segmentation challenge dataset which is composed of images obtained through four different modalities. Accordingly, we present an extended version of existing network to solve segmentation problem. The network architecture consists of multiple neural network layers connected in sequential order with the feeding of Convolutional feature maps at the peer level. Experimental results on BRATS 2015 benchmark data thus show the usability of the proposed approach and its superiority over the other approaches in this area of research. © 2018 Wiley Periodicals, Inc.

Modeling EEG Waveforms with Semi-Supervised Deep Belief Nets: Fast Classification and Anomaly Measurement

PubMed Central

Wulsin, D. F.; Gupta, J. R.; Mani, R.; Blanco, J. A.; Litt, B.

2011-01-01

Clinical electroencephalography (EEG) records vast amounts of human complex data yet is still reviewed primarily by human readers. Deep Belief Nets (DBNs) are a relatively new type of multi-layer neural network commonly tested on two-dimensional image data, but are rarely applied to times-series data such as EEG. We apply DBNs in a semi-supervised paradigm to model EEG waveforms for classification and anomaly detection. DBN performance was comparable to standard classifiers on our EEG dataset, and classification time was found to be 1.7 to 103.7 times faster than the other high-performing classifiers. We demonstrate how the unsupervised step of DBN learning produces an autoencoder that can naturally be used in anomaly measurement. We compare the use of raw, unprocessed data—a rarity in automated physiological waveform analysis—to hand-chosen features and find that raw data produces comparable classification and better anomaly measurement performance. These results indicate that DBNs and raw data inputs may be more effective for online automated EEG waveform recognition than other common techniques. PMID:21525569

Object detection approach using generative sparse, hierarchical networks with top-down and lateral connections for combining texture/color detection and shape/contour detection

DOEpatents

Paiton, Dylan M.; Kenyon, Garrett T.; Brumby, Steven P.; Schultz, Peter F.; George, John S.

2015-07-28

An approach to detecting objects in an image dataset may combine texture/color detection, shape/contour detection, and/or motion detection using sparse, generative, hierarchical models with lateral and top-down connections. A first independent representation of objects in an image dataset may be produced using a color/texture detection algorithm. A second independent representation of objects in the image dataset may be produced using a shape/contour detection algorithm. A third independent representation of objects in the image dataset may be produced using a motion detection algorithm. The first, second, and third independent representations may then be combined into a single coherent output using a combinatorial algorithm.

Feature Representations for Neuromorphic Audio Spike Streams.

PubMed

Anumula, Jithendar; Neil, Daniel; Delbruck, Tobi; Liu, Shih-Chii

2018-01-01

Event-driven neuromorphic spiking sensors such as the silicon retina and the silicon cochlea encode the external sensory stimuli as asynchronous streams of spikes across different channels or pixels. Combining state-of-art deep neural networks with the asynchronous outputs of these sensors has produced encouraging results on some datasets but remains challenging. While the lack of effective spiking networks to process the spike streams is one reason, the other reason is that the pre-processing methods required to convert the spike streams to frame-based features needed for the deep networks still require further investigation. This work investigates the effectiveness of synchronous and asynchronous frame-based features generated using spike count and constant event binning in combination with the use of a recurrent neural network for solving a classification task using N-TIDIGITS18 dataset. This spike-based dataset consists of recordings from the Dynamic Audio Sensor, a spiking silicon cochlea sensor, in response to the TIDIGITS audio dataset. We also propose a new pre-processing method which applies an exponential kernel on the output cochlea spikes so that the interspike timing information is better preserved. The results from the N-TIDIGITS18 dataset show that the exponential features perform better than the spike count features, with over 91% accuracy on the digit classification task. This accuracy corresponds to an improvement of at least 2.5% over the use of spike count features, establishing a new state of the art for this dataset.

Feature Representations for Neuromorphic Audio Spike Streams

PubMed Central

Anumula, Jithendar; Neil, Daniel; Delbruck, Tobi; Liu, Shih-Chii

2018-01-01

Event-driven neuromorphic spiking sensors such as the silicon retina and the silicon cochlea encode the external sensory stimuli as asynchronous streams of spikes across different channels or pixels. Combining state-of-art deep neural networks with the asynchronous outputs of these sensors has produced encouraging results on some datasets but remains challenging. While the lack of effective spiking networks to process the spike streams is one reason, the other reason is that the pre-processing methods required to convert the spike streams to frame-based features needed for the deep networks still require further investigation. This work investigates the effectiveness of synchronous and asynchronous frame-based features generated using spike count and constant event binning in combination with the use of a recurrent neural network for solving a classification task using N-TIDIGITS18 dataset. This spike-based dataset consists of recordings from the Dynamic Audio Sensor, a spiking silicon cochlea sensor, in response to the TIDIGITS audio dataset. We also propose a new pre-processing method which applies an exponential kernel on the output cochlea spikes so that the interspike timing information is better preserved. The results from the N-TIDIGITS18 dataset show that the exponential features perform better than the spike count features, with over 91% accuracy on the digit classification task. This accuracy corresponds to an improvement of at least 2.5% over the use of spike count features, establishing a new state of the art for this dataset. PMID:29479300

Deep Reinforcement Learning of Cell Movement in the Early Stage of C. elegans Embryogenesis.

PubMed

Wang, Zi; Wang, Dali; Li, Chengcheng; Xu, Yichi; Li, Husheng; Bao, Zhirong

2018-04-25

Cell movement in the early phase of C. elegans development is regulated by a highly complex process in which a set of rules and connections are formulated at distinct scales. Previous efforts have demonstrated that agent-based, multi-scale modeling systems can integrate physical and biological rules and provide new avenues to study developmental systems. However, the application of these systems to model cell movement is still challenging and requires a comprehensive understanding of regulatory networks at the right scales. Recent developments in deep learning and reinforcement learning provide an unprecedented opportunity to explore cell movement using 3D time-lapse microscopy images. We present a deep reinforcement learning approach within an agent-based modeling system to characterize cell movement in the embryonic development of C. elegans. Our modeling system captures the complexity of cell movement patterns in the embryo and overcomes the local optimization problem encountered by traditional rule-based, agent-based modeling that uses greedy algorithms. We tested our model with two real developmental processes: the anterior movement of the Cpaaa cell via intercalation and the rearrangement of the superficial left-right asymmetry. In the first case, the model results suggested that Cpaaa's intercalation is an active directional cell movement caused by the continuous effects from a longer distance (farther than the length of two adjacent cells), as opposed to a passive movement caused by neighbor cell movements. In the second case, a leader-follower mechanism well explained the collective cell movement pattern in the asymmetry rearrangement. These results showed that our approach to introduce deep reinforcement learning into agent-based modeling can test regulatory mechanisms by exploring cell migration paths in a reverse engineering perspective. This model opens new doors to explore the large datasets generated by live imaging. Source code is available at https://github.com/zwang84/drl4cellmovement. dwang7@utk.edu, baoz@mskcc.org. Supplementary data are available at Bioinformatics online.

Deep learning for EEG-Based preference classification

NASA Astrophysics Data System (ADS)

Teo, Jason; Hou, Chew Lin; Mountstephens, James

2017-10-01

Electroencephalogram (EEG)-based emotion classification is rapidly becoming one of the most intensely studied areas of brain-computer interfacing (BCI). The ability to passively identify yet accurately correlate brainwaves with our immediate emotions opens up truly meaningful and previously unattainable human-computer interactions such as in forensic neuroscience, rehabilitative medicine, affective entertainment and neuro-marketing. One particularly useful yet rarely explored areas of EEG-based emotion classification is preference recognition [1], which is simply the detection of like versus dislike. Within the limited investigations into preference classification, all reported studies were based on musically-induced stimuli except for a single study which used 2D images. The main objective of this study is to apply deep learning, which has been shown to produce state-of-the-art results in diverse hard problems such as in computer vision, natural language processing and audio recognition, to 3D object preference classification over a larger group of test subjects. A cohort of 16 users was shown 60 bracelet-like objects as rotating visual stimuli on a computer display while their preferences and EEGs were recorded. After training a variety of machine learning approaches which included deep neural networks, we then attempted to classify the users' preferences for the 3D visual stimuli based on their EEGs. Here, we show that that deep learning outperforms a variety of other machine learning classifiers for this EEG-based preference classification task particularly in a highly challenging dataset with large inter- and intra-subject variability.

Detecting buried explosive hazards with handheld GPR and deep learning

NASA Astrophysics Data System (ADS)

Besaw, Lance E.

2016-05-01

Buried explosive hazards (BEHs), including traditional landmines and homemade improvised explosives, have proven difficult to detect and defeat during and after conflicts around the world. Despite their various sizes, shapes and construction material, ground penetrating radar (GPR) is an excellent phenomenology for detecting BEHs due to its ability to sense localized differences in electromagnetic properties. Handheld GPR detectors are common equipment for detecting BEHs because of their flexibility (in part due to the human operator) and effectiveness in cluttered environments. With modern digital electronics and positioning systems, handheld GPR sensors can sense and map variation in electromagnetic properties while searching for BEHs. Additionally, large-scale computers have demonstrated an insatiable appetite for ingesting massive datasets and extracting meaningful relationships. This is no more evident than the maturation of deep learning artificial neural networks (ANNs) for image and speech recognition now commonplace in industry and academia. This confluence of sensing, computing and pattern recognition technologies offers great potential to develop automatic target recognition techniques to assist GPR operators searching for BEHs. In this work deep learning ANNs are used to detect BEHs and discriminate them from harmless clutter. We apply these techniques to a multi-antennae, handheld GPR with centimeter-accurate positioning system that was used to collect data over prepared lanes containing a wide range of BEHs. This work demonstrates that deep learning ANNs can automatically extract meaningful information from complex GPR signatures, complementing existing GPR anomaly detection and classification techniques.

Automatic recognition of 3D GGO CT imaging signs through the fusion of hybrid resampling and layer-wise fine-tuning CNNs.

PubMed

Han, Guanghui; Liu, Xiabi; Zheng, Guangyuan; Wang, Murong; Huang, Shan

2018-06-06

Ground-glass opacity (GGO) is a common CT imaging sign on high-resolution CT, which means the lesion is more likely to be malignant compared to common solid lung nodules. The automatic recognition of GGO CT imaging signs is of great importance for early diagnosis and possible cure of lung cancers. The present GGO recognition methods employ traditional low-level features and system performance improves slowly. Considering the high-performance of CNN model in computer vision field, we proposed an automatic recognition method of 3D GGO CT imaging signs through the fusion of hybrid resampling and layer-wise fine-tuning CNN models in this paper. Our hybrid resampling is performed on multi-views and multi-receptive fields, which reduces the risk of missing small or large GGOs by adopting representative sampling panels and processing GGOs with multiple scales simultaneously. The layer-wise fine-tuning strategy has the ability to obtain the optimal fine-tuning model. Multi-CNN models fusion strategy obtains better performance than any single trained model. We evaluated our method on the GGO nodule samples in publicly available LIDC-IDRI dataset of chest CT scans. The experimental results show that our method yields excellent results with 96.64% sensitivity, 71.43% specificity, and 0.83 F1 score. Our method is a promising approach to apply deep learning method to computer-aided analysis of specific CT imaging signs with insufficient labeled images. Graphical abstract We proposed an automatic recognition method of 3D GGO CT imaging signs through the fusion of hybrid resampling and layer-wise fine-tuning CNN models in this paper. Our hybrid resampling reduces the risk of missing small or large GGOs by adopting representative sampling panels and processing GGOs with multiple scales simultaneously. The layer-wise fine-tuning strategy has ability to obtain the optimal fine-tuning model. Our method is a promising approach to apply deep learning method to computer-aided analysis of specific CT imaging signs with insufficient labeled images.

Unveiling the extreme nature of the hyper faint galaxy Virgo I

NASA Astrophysics Data System (ADS)

Crnojevic, Denija

2017-08-01

We request HST/ACS imaging to obtain a deep color-magnitude diagram of the newly discovered candidate Milky Way satellite Virgo I. With an estimated absolute magnitude of only M_V -0.8 and a Galactocentric radius of 90 kpc, Virgo I is one of the faintest and most distant dwarfs ever observed, and could be identified as a prototype ''hyper'' faint galaxy. The detailed characterization of the smallest inhabited dark matter subhalos is crucial to guide hierarchical galaxy formation models, and in particular to constrain reionization, the nature of the dark matter particle, etc. With the advent of deep wide-field, ground-based surveys, the potential of uncovering these lowest-mass galaxies is quickly turning into reality, as demonstrated by the discovery in the past two years of tens of new Local Group members in the ultra-faint regime (M_V>-8). Virgo I represents a new record in galaxy physical properties, and urges us to be prepared for the likely emergence of an entirely new class of such objects in the era of future wide-field surveys (e.g., LSST). Only high resolution HST observations can enable us to confirm the nature of Virgo I, providing significantly more accurate estimates for its distance and structural properties, when compared to the discovery Subaru/HyperSuprimeCam imaging. Our proposed dataset will constitute a fundamental step in the upcoming hunt for galaxies with similarly extreme properties.

A Case Study on Attribute Recognition of Heated Metal Mark Image Using Deep Convolutional Neural Networks.

PubMed

Mao, Keming; Lu, Duo; E, Dazhi; Tan, Zhenhua

2018-06-07

Heated metal mark is an important trace to identify the cause of fire. However, traditional methods mainly focus on the knowledge of physics and chemistry for qualitative analysis and make it still a challenging problem. This paper presents a case study on attribute recognition of the heated metal mark image using computer vision and machine learning technologies. The proposed work is composed of three parts. Material is first generated. According to national standards, actual needs and feasibility, seven attributes are selected for research. Data generation and organization are conducted, and a small size benchmark dataset is constructed. A recognition model is then implemented. Feature representation and classifier construction methods are introduced based on deep convolutional neural networks. Finally, the experimental evaluation is carried out. Multi-aspect testings are performed with various model structures, data augments, training modes, optimization methods and batch sizes. The influence of parameters, recognitio efficiency and execution time are also analyzed. The results show that with a fine-tuned model, the recognition rate of attributes metal type, heating mode, heating temperature, heating duration, cooling mode, placing duration and relative humidity are 0.925, 0.908, 0.835, 0.917, 0.928, 0.805 and 0.92, respectively. The proposed method recognizes the attribute of heated metal mark with preferable effect, and it can be used in practical application.

Connectivity Predicts Deep Brain Stimulation Outcome in Parkinson Disease

PubMed Central

Horn, Andreas; Reich, Martin; Vorwerk, Johannes; Li, Ningfei; Wenzel, Gregor; Fang, Qianqian; Schmitz-Hübsch, Tanja; Nickl, Robert; Kupsch, Andreas; Volkmann, Jens; Kühn, Andrea A.; Fox, Michael D.

2018-01-01

Objective The benefit of deep brain stimulation (DBS) for Parkinson disease (PD) may depend on connectivity between the stimulation site and other brain regions, but which regions and whether connectivity can predict outcome in patients remain unknown. Here, we identify the structural and functional connectivity profile of effective DBS to the subthalamic nucleus (STN) and test its ability to predict outcome in an independent cohort. Methods A training dataset of 51 PD patients with STN DBS was combined with publicly available human connectome data (diffusion tractography and resting state functional connectivity) to identify connections reliably associated with clinical improvement (motor score of the Unified Parkinson Disease Rating Scale [UPDRS]). This connectivity profile was then used to predict outcome in an independent cohort of 44 patients from a different center. Results In the training dataset, connectivity between the DBS electrode and a distributed network of brain regions correlated with clinical response including structural connectivity to supplementary motor area and functional anticorrelation to primary motor cortex (p<0.001). This same connectivity profile predicted response in an independent patient cohort (p<0.01). Structural and functional connectivity were independent predictors of clinical improvement (p<0.001) and estimated response in individual patients with an average error of 15% UPDRS improvement. Results were similar using connectome data from normal subjects or a connectome age, sex, and disease matched to our DBS patients. Interpretation Effective STN DBS for PD is associated with a specific connectivity profile that can predict clinical outcome across independent cohorts. This prediction does not require specialized imaging in PD patients themselves. PMID:28586141

Range and Panoramic Image Fusion Into a Textured Range Image for Culture Heritage Documentation

NASA Astrophysics Data System (ADS)

Bila, Z.; Reznicek, J.; Pavelka, K.

2013-07-01

This paper deals with a fusion of range and panoramic images, where the range image is acquired by a 3D laser scanner and the panoramic image is acquired with a digital still camera mounted on a panoramic head and tripod. The fused resulting dataset, called "textured range image", provides more reliable information about the investigated object for conservators and historians, than using both datasets separately. A simple example of fusion of a range and panoramic images, both obtained in St. Francis Xavier Church in town Opařany, is given here. Firstly, we describe the process of data acquisition, then the processing of both datasets into a proper format for following fusion and the process of fusion. The process of fusion can be divided into a two main parts: transformation and remapping. In the first, transformation, part, both images are related by matching similar features detected on both images with a proper detector, which results in transformation matrix enabling transformation of the range image onto a panoramic image. Then, the range data are remapped from the range image space into a panoramic image space and stored as an additional "range" channel. The process of image fusion is validated by comparing similar features extracted on both datasets.

Interferometric imaging of the San Andreas Fault at Parkfield Using a Massive VSP

NASA Astrophysics Data System (ADS)

Chavarria, J.; Goertz, A.; Karrenbach, M.; Paulsson, B.

2006-12-01

The San Andreas Fault Observatory at Depth (SAFOD) has the goal of investigating the physical processes controlling earthquakes in central California at Parkfield. The observatory consists of a deep well, that intersected the San Andreas Fault at seismogenic depths of ~3.2 km, where recurring microearthquake activity takes place. Previous to the drilling of Phase II, that intersected the fault at the end of Summer 2005, a long array of seismometers was deployed in the deeper part of the well to characterize the fault structure and to aid the monitoring activities of the natural seismicity. The Massive VSP array consisted of 80 three component seismometers that were deployed during April and May 2005. During this period of time we recorded active data from 13 explosions used to refine the velocity models that improved the locations of target events. In addition to this, we continuously monitored passive data from the natural earthquake activity in the area that was dominated by the Parkfield 2004 event aftershock sequence. During the recording time of this project we were able to record one of the target earthquakes of the SAFOD project as well as recently discovered non-volcanic tremor. The data recorded by this deep array of seismometers has provided important information to characterize the structure of the fault at scale that is comparable to the scale of the geologic observations. These observations, derived during drilling of the well, consist of cores and cuttings as well as geophysical logs. With the use of the borehole array of seismometers we have been able to directly correlate this lithologic information with seismic phases observed in the VSP data. This has lead to better understanding the nature of the scattering that takes place in this complex fault zone. Seismic reflections in the dataset were analyzed with Kirchhoff imaging methods to determine a partial image of the San Andreas Fault at depth. Due to the low density distribution of sources in the study area the illumination volume is limited but still shows clear signals originating from sub-vertical structures associated to the strands of the San Andreas fault inferred from high resolution earthquake location. The existing Kirchhoff imaging methods have been complemented with the use of interferometric imaging by which we extract the Green's function of the data recorded by the array. This takes place with the use of correlation analysis of both active and passive data resulting in a dataset where each downhole receiver acts as a virtual source. This analysis has provided a refined image of the structure of the San Andreas Fault at seismogenic depths with which a better understanding of the system can be achieved.

Benchmark datasets for 3D MALDI- and DESI-imaging mass spectrometry.

PubMed

Oetjen, Janina; Veselkov, Kirill; Watrous, Jeramie; McKenzie, James S; Becker, Michael; Hauberg-Lotte, Lena; Kobarg, Jan Hendrik; Strittmatter, Nicole; Mróz, Anna K; Hoffmann, Franziska; Trede, Dennis; Palmer, Andrew; Schiffler, Stefan; Steinhorst, Klaus; Aichler, Michaela; Goldin, Robert; Guntinas-Lichius, Orlando; von Eggeling, Ferdinand; Thiele, Herbert; Maedler, Kathrin; Walch, Axel; Maass, Peter; Dorrestein, Pieter C; Takats, Zoltan; Alexandrov, Theodore

2015-01-01

Three-dimensional (3D) imaging mass spectrometry (MS) is an analytical chemistry technique for the 3D molecular analysis of a tissue specimen, entire organ, or microbial colonies on an agar plate. 3D-imaging MS has unique advantages over existing 3D imaging techniques, offers novel perspectives for understanding the spatial organization of biological processes, and has growing potential to be introduced into routine use in both biology and medicine. Owing to the sheer quantity of data generated, the visualization, analysis, and interpretation of 3D imaging MS data remain a significant challenge. Bioinformatics research in this field is hampered by the lack of publicly available benchmark datasets needed to evaluate and compare algorithms. High-quality 3D imaging MS datasets from different biological systems at several labs were acquired, supplied with overview images and scripts demonstrating how to read them, and deposited into MetaboLights, an open repository for metabolomics data. 3D imaging MS data were collected from five samples using two types of 3D imaging MS. 3D matrix-assisted laser desorption/ionization imaging (MALDI) MS data were collected from murine pancreas, murine kidney, human oral squamous cell carcinoma, and interacting microbial colonies cultured in Petri dishes. 3D desorption electrospray ionization (DESI) imaging MS data were collected from a human colorectal adenocarcinoma. With the aim to stimulate computational research in the field of computational 3D imaging MS, selected high-quality 3D imaging MS datasets are provided that could be used by algorithm developers as benchmark datasets.

A uniform geostationary visible calibration approach to achieve a climate quality dataset

NASA Astrophysics Data System (ADS)

Haney, C.; Doelling, D.; Bhatt, R.; Scarino, B. R.; Gopalan, A.

2013-12-01

The geostationary (GEO) weather satellite visible and IR image record has surpassed 30 years. They have been preserved in the ISCCP-B1U 3-hourly dataset and other archives such as McIDAS, EUMETSAT, and NOAA CLASS. Since they were designed to aid in weather forecasting, long-term calibration stability was not a high priority. All GEO imagers lack onboard visible calibration and suffer from optical degradation after they are launched. In order to piece together the 35+ GEO satellite record both in time and space, a uniform calibration approach is desired to remove individual GEO temporal trends, as well as GEO spectral band differences. Otherwise, any artificial discontinuities caused by sequential GEO satellite records or spurious temporal trends caused by optical degradation may be interpreted as a change in climate. The approach relies on multiple independent methods to reduce the overall uncertainty of the GEO calibration coefficients. Consistency among methods validates the approach. During the MODIS record (2000 to the present) the GEO satellites are inter-calibrated against MODIS using ray-matched or bore-sighted radiance pairs. The MODIS and the VIIRS follow on instruments are equipped with onboard calibration thereby providing a stable calibration reference. The GEO spectral band differences are accounted for using a Spectral Band Adjustment Factor (SBAF) based on hyper-spectral SCIAMACHY data. During the pre-MODIS era, invariant earth targets of deserts and deep convective clouds (DCC) are used. Since GEO imagers have maintained their imaging scan schedules, GEO desert and DCC bidirectional reflectance distribution functions (BRDF) can be constructed and validated during the MODIS era. The BRDF models can then be applied to historical GEO imagers. Consistency among desert and DCC GEO calibration gains validates the approach. This approach has been applied to the GEO record beginning in 1985 and the results will be presented at the meeting.

BCDForest: a boosting cascade deep forest model towards the classification of cancer subtypes based on gene expression data.

PubMed

Guo, Yang; Liu, Shuhui; Li, Zhanhuai; Shang, Xuequn

2018-04-11

The classification of cancer subtypes is of great importance to cancer disease diagnosis and therapy. Many supervised learning approaches have been applied to cancer subtype classification in the past few years, especially of deep learning based approaches. Recently, the deep forest model has been proposed as an alternative of deep neural networks to learn hyper-representations by using cascade ensemble decision trees. It has been proved that the deep forest model has competitive or even better performance than deep neural networks in some extent. However, the standard deep forest model may face overfitting and ensemble diversity challenges when dealing with small sample size and high-dimensional biology data. In this paper, we propose a deep learning model, so-called BCDForest, to address cancer subtype classification on small-scale biology datasets, which can be viewed as a modification of the standard deep forest model. The BCDForest distinguishes from the standard deep forest model with the following two main contributions: First, a named multi-class-grained scanning method is proposed to train multiple binary classifiers to encourage diversity of ensemble. Meanwhile, the fitting quality of each classifier is considered in representation learning. Second, we propose a boosting strategy to emphasize more important features in cascade forests, thus to propagate the benefits of discriminative features among cascade layers to improve the classification performance. Systematic comparison experiments on both microarray and RNA-Seq gene expression datasets demonstrate that our method consistently outperforms the state-of-the-art methods in application of cancer subtype classification. The multi-class-grained scanning and boosting strategy in our model provide an effective solution to ease the overfitting challenge and improve the robustness of deep forest model working on small-scale data. Our model provides a useful approach to the classification of cancer subtypes by using deep learning on high-dimensional and small-scale biology data.

Deep learning based beat event detection in action movie franchises

NASA Astrophysics Data System (ADS)

Ejaz, N.; Khan, U. A.; Martínez-del-Amor, M. A.; Sparenberg, H.

2018-04-01

Automatic understanding and interpretation of movies can be used in a variety of ways to semantically manage the massive volumes of movies data. "Action Movie Franchises" dataset is a collection of twenty Hollywood action movies from five famous franchises with ground truth annotations at shot and beat level of each movie. In this dataset, the annotations are provided for eleven semantic beat categories. In this work, we propose a deep learning based method to classify shots and beat-events on this dataset. The training dataset for each of the eleven beat categories is developed and then a Convolution Neural Network is trained. After finding the shot boundaries, key frames are extracted for each shot and then three classification labels are assigned to each key frame. The classification labels for each of the key frames in a particular shot are then used to assign a unique label to each shot. A simple sliding window based method is then used to group adjacent shots having the same label in order to find a particular beat event. The results of beat event classification are presented based on criteria of precision, recall, and F-measure. The results are compared with the existing technique and significant improvements are recorded.

Multiclass classification of obstructive sleep apnea/hypopnea based on a convolutional neural network from a single-lead electrocardiogram.

PubMed

Urtnasan, Erdenebayar; Park, Jong-Uk; Lee, Kyoung-Joung

2018-05-24

In this paper, we propose a convolutional neural network (CNN)-based deep learning architecture for multiclass classification of obstructive sleep apnea and hypopnea (OSAH) using single-lead electrocardiogram (ECG) recordings. OSAH is the most common sleep-related breathing disorder. Many subjects who suffer from OSAH remain undiagnosed; thus, early detection of OSAH is important. In this study, automatic classification of three classes-normal, hypopnea, and apnea-based on a CNN is performed. An optimal six-layer CNN model is trained on a training dataset (45,096 events) and evaluated on a test dataset (11,274 events). The training set (69 subjects) and test set (17 subjects) were collected from 86 subjects with length of approximately 6 h and segmented into 10 s durations. The proposed CNN model reaches a mean -score of 93.0 for the training dataset and 87.0 for the test dataset. Thus, proposed deep learning architecture achieved a high performance for multiclass classification of OSAH using single-lead ECG recordings. The proposed method can be employed in screening of patients suspected of having OSAH. © 2018 Institute of Physics and Engineering in Medicine.

Multi-class segmentation of neuronal electron microscopy images using deep learning

NASA Astrophysics Data System (ADS)

Khobragade, Nivedita; Agarwal, Chirag

2018-03-01

Study of connectivity of neural circuits is an essential step towards a better understanding of functioning of the nervous system. With the recent improvement in imaging techniques, high-resolution and high-volume images are being generated requiring automated segmentation techniques. We present a pixel-wise classification method based on Bayesian SegNet architecture. We carried out multi-class segmentation on serial section Transmission Electron Microscopy (ssTEM) images of Drosophila third instar larva ventral nerve cord, labeling the four classes of neuron membranes, neuron intracellular space, mitochondria and glia / extracellular space. Bayesian SegNet was trained using 256 ssTEM images of 256 x 256 pixels and tested on 64 different ssTEM images of the same size, from the same serial stack. Due to high class imbalance, we used a class-balanced version of Bayesian SegNet by re-weighting each class based on their relative frequency. We achieved an overall accuracy of 93% and a mean class accuracy of 88% for pixel-wise segmentation using this encoder-decoder approach. On evaluating the segmentation results using similarity metrics like SSIM and Dice Coefficient, we obtained scores of 0.994 and 0.886 respectively. Additionally, we used the network trained using the 256 ssTEM images of Drosophila third instar larva for multi-class labeling of ISBI 2012 challenge ssTEM dataset.

Wishart Deep Stacking Network for Fast POLSAR Image Classification.

PubMed

Jiao, Licheng; Liu, Fang

2016-05-11

Inspired by the popular deep learning architecture - Deep Stacking Network (DSN), a specific deep model for polarimetric synthetic aperture radar (POLSAR) image classification is proposed in this paper, which is named as Wishart Deep Stacking Network (W-DSN). First of all, a fast implementation of Wishart distance is achieved by a special linear transformation, which speeds up the classification of POLSAR image and makes it possible to use this polarimetric information in the following Neural Network (NN). Then a single-hidden-layer neural network based on the fast Wishart distance is defined for POLSAR image classification, which is named as Wishart Network (WN) and improves the classification accuracy. Finally, a multi-layer neural network is formed by stacking WNs, which is in fact the proposed deep learning architecture W-DSN for POLSAR image classification and improves the classification accuracy further. In addition, the structure of WN can be expanded in a straightforward way by adding hidden units if necessary, as well as the structure of the W-DSN. As a preliminary exploration on formulating specific deep learning architecture for POLSAR image classification, the proposed methods may establish a simple but clever connection between POLSAR image interpretation and deep learning. The experiment results tested on real POLSAR image show that the fast implementation of Wishart distance is very efficient (a POLSAR image with 768000 pixels can be classified in 0.53s), and both the single-hidden-layer architecture WN and the deep learning architecture W-DSN for POLSAR image classification perform well and work efficiently.

DeepSkeleton: Learning Multi-Task Scale-Associated Deep Side Outputs for Object Skeleton Extraction in Natural Images

NASA Astrophysics Data System (ADS)

Shen, Wei; Zhao, Kai; Jiang, Yuan; Wang, Yan; Bai, Xiang; Yuille, Alan

2017-11-01

Object skeletons are useful for object representation and object detection. They are complementary to the object contour, and provide extra information, such as how object scale (thickness) varies among object parts. But object skeleton extraction from natural images is very challenging, because it requires the extractor to be able to capture both local and non-local image context in order to determine the scale of each skeleton pixel. In this paper, we present a novel fully convolutional network with multiple scale-associated side outputs to address this problem. By observing the relationship between the receptive field sizes of the different layers in the network and the skeleton scales they can capture, we introduce two scale-associated side outputs to each stage of the network. The network is trained by multi-task learning, where one task is skeleton localization to classify whether a pixel is a skeleton pixel or not, and the other is skeleton scale prediction to regress the scale of each skeleton pixel. Supervision is imposed at different stages by guiding the scale-associated side outputs toward the groundtruth skeletons at the appropriate scales. The responses of the multiple scale-associated side outputs are then fused in a scale-specific way to detect skeleton pixels using multiple scales effectively. Our method achieves promising results on two skeleton extraction datasets, and significantly outperforms other competitors. Additionally, the usefulness of the obtained skeletons and scales (thickness) are verified on two object detection applications: Foreground object segmentation and object proposal detection.

Gradient-based reliability maps for ACM-based segmentation of hippocampus.

PubMed

Zarpalas, Dimitrios; Gkontra, Polyxeni; Daras, Petros; Maglaveras, Nicos

2014-04-01

Automatic segmentation of deep brain structures, such as the hippocampus (HC), in MR images has attracted considerable scientific attention due to the widespread use of MRI and to the principal role of some structures in various mental disorders. In this literature, there exists a substantial amount of work relying on deformable models incorporating prior knowledge about structures' anatomy and shape information. However, shape priors capture global shape characteristics and thus fail to model boundaries of varying properties; HC boundaries present rich, poor, and missing gradient regions. On top of that, shape prior knowledge is blended with image information in the evolution process, through global weighting of the two terms, again neglecting the spatially varying boundary properties, causing segmentation faults. An innovative method is hereby presented that aims to achieve highly accurate HC segmentation in MR images, based on the modeling of boundary properties at each anatomical location and the inclusion of appropriate image information for each of those, within an active contour model framework. Hence, blending of image information and prior knowledge is based on a local weighting map, which mixes gradient information, regional and whole brain statistical information with a multi-atlas-based spatial distribution map of the structure's labels. Experimental results on three different datasets demonstrate the efficacy and accuracy of the proposed method.

Textile Retrieval Based on Image Content from CDC and Webcam Cameras in Indoor Environments.

PubMed

García-Olalla, Oscar; Alegre, Enrique; Fernández-Robles, Laura; Fidalgo, Eduardo; Saikia, Surajit

2018-04-25

Textile based image retrieval for indoor environments can be used to retrieve images that contain the same textile, which may indicate that scenes are related. This makes up a useful approach for law enforcement agencies who want to find evidence based on matching between textiles. In this paper, we propose a novel pipeline that allows searching and retrieving textiles that appear in pictures of real scenes. Our approach is based on first obtaining regions containing textiles by using MSER on high pass filtered images of the RGB, HSV and Hue channels of the original photo. To describe the textile regions, we demonstrated that the combination of HOG and HCLOSIB is the best option for our proposal when using the correlation distance to match the query textile patch with the candidate regions. Furthermore, we introduce a new dataset, TextilTube, which comprises a total of 1913 textile regions labelled within 67 classes. We yielded 84.94% of success in the 40 nearest coincidences and 37.44% of precision taking into account just the first coincidence, which outperforms the current deep learning methods evaluated. Experimental results show that this pipeline can be used to set up an effective textile based image retrieval system in indoor environments.

Textile Retrieval Based on Image Content from CDC and Webcam Cameras in Indoor Environments

PubMed Central

García-Olalla, Oscar; Saikia, Surajit

2018-01-01

Textile based image retrieval for indoor environments can be used to retrieve images that contain the same textile, which may indicate that scenes are related. This makes up a useful approach for law enforcement agencies who want to find evidence based on matching between textiles. In this paper, we propose a novel pipeline that allows searching and retrieving textiles that appear in pictures of real scenes. Our approach is based on first obtaining regions containing textiles by using MSER on high pass filtered images of the RGB, HSV and Hue channels of the original photo. To describe the textile regions, we demonstrated that the combination of HOG and HCLOSIB is the best option for our proposal when using the correlation distance to match the query textile patch with the candidate regions. Furthermore, we introduce a new dataset, TextilTube, which comprises a total of 1913 textile regions labelled within 67 classes. We yielded 84.94% of success in the 40 nearest coincidences and 37.44% of precision taking into account just the first coincidence, which outperforms the current deep learning methods evaluated. Experimental results show that this pipeline can be used to set up an effective textile based image retrieval system in indoor environments. PMID:29693590

Querying Patterns in High-Dimensional Heterogenous Datasets

ERIC Educational Resources Information Center

Singh, Vishwakarma

2012-01-01

The recent technological advancements have led to the availability of a plethora of heterogenous datasets, e.g., images tagged with geo-location and descriptive keywords. An object in these datasets is described by a set of high-dimensional feature vectors. For example, a keyword-tagged image is represented by a color-histogram and a…

Simultenious binary hash and features learning for image retrieval

NASA Astrophysics Data System (ADS)

Frantc, V. A.; Makov, S. V.; Voronin, V. V.; Marchuk, V. I.; Semenishchev, E. A.; Egiazarian, K. O.; Agaian, S.

2016-05-01

Content-based image retrieval systems have plenty of applications in modern world. The most important one is the image search by query image or by semantic description. Approaches to this problem are employed in personal photo-collection management systems, web-scale image search engines, medical systems, etc. Automatic analysis of large unlabeled image datasets is virtually impossible without satisfactory image-retrieval technique. It's the main reason why this kind of automatic image processing has attracted so much attention during recent years. Despite rather huge progress in the field, semantically meaningful image retrieval still remains a challenging task. The main issue here is the demand to provide reliable results in short amount of time. This paper addresses the problem by novel technique for simultaneous learning of global image features and binary hash codes. Our approach provide mapping of pixel-based image representation to hash-value space simultaneously trying to save as much of semantic image content as possible. We use deep learning methodology to generate image description with properties of similarity preservation and statistical independence. The main advantage of our approach in contrast to existing is ability to fine-tune retrieval procedure for very specific application which allow us to provide better results in comparison to general techniques. Presented in the paper framework for data- dependent image hashing is based on use two different kinds of neural networks: convolutional neural networks for image description and autoencoder for feature to hash space mapping. Experimental results confirmed that our approach has shown promising results in compare to other state-of-the-art methods.

Atlas Toolkit: Fast registration of 3D morphological datasets in the absence of landmarks

PubMed Central

Grocott, Timothy; Thomas, Paul; Münsterberg, Andrea E.

2016-01-01

Image registration is a gateway technology for Developmental Systems Biology, enabling computational analysis of related datasets within a shared coordinate system. Many registration tools rely on landmarks to ensure that datasets are correctly aligned; yet suitable landmarks are not present in many datasets. Atlas Toolkit is a Fiji/ImageJ plugin collection offering elastic group-wise registration of 3D morphological datasets, guided by segmentation of the interesting morphology. We demonstrate the method by combinatorial mapping of cell signalling events in the developing eyes of chick embryos, and use the integrated datasets to predictively enumerate Gene Regulatory Network states. PMID:26864723

Atlas Toolkit: Fast registration of 3D morphological datasets in the absence of landmarks.

PubMed

Grocott, Timothy; Thomas, Paul; Münsterberg, Andrea E

2016-02-11

Image registration is a gateway technology for Developmental Systems Biology, enabling computational analysis of related datasets within a shared coordinate system. Many registration tools rely on landmarks to ensure that datasets are correctly aligned; yet suitable landmarks are not present in many datasets. Atlas Toolkit is a Fiji/ImageJ plugin collection offering elastic group-wise registration of 3D morphological datasets, guided by segmentation of the interesting morphology. We demonstrate the method by combinatorial mapping of cell signalling events in the developing eyes of chick embryos, and use the integrated datasets to predictively enumerate Gene Regulatory Network states.

Image segmentation evaluation for very-large datasets

NASA Astrophysics Data System (ADS)

Reeves, Anthony P.; Liu, Shuang; Xie, Yiting

2016-03-01

With the advent of modern machine learning methods and fully automated image analysis there is a need for very large image datasets having documented segmentations for both computer algorithm training and evaluation. Current approaches of visual inspection and manual markings do not scale well to big data. We present a new approach that depends on fully automated algorithm outcomes for segmentation documentation, requires no manual marking, and provides quantitative evaluation for computer algorithms. The documentation of new image segmentations and new algorithm outcomes are achieved by visual inspection. The burden of visual inspection on large datasets is minimized by (a) customized visualizations for rapid review and (b) reducing the number of cases to be reviewed through analysis of quantitative segmentation evaluation. This method has been applied to a dataset of 7,440 whole-lung CT images for 6 different segmentation algorithms designed to fully automatically facilitate the measurement of a number of very important quantitative image biomarkers. The results indicate that we could achieve 93% to 99% successful segmentation for these algorithms on this relatively large image database. The presented evaluation method may be scaled to much larger image databases.

Selecting AGN through Variability in SN Datasets

NASA Astrophysics Data System (ADS)

Boutsia, K.; Leibundgut, B.; Trevese, D.; Vagnetti, F.

2010-07-01

Variability is a main property of Active Galactic Nuclei (AGN) and it was adopted as a selection criterion using multi epoch surveys conducted for the detection of supernovae (SNe). We have used two SN datasets. First we selected the AXAF field of the STRESS project, centered in the Chandra Deep Field South where, besides the deep X-ray surveys also various optical catalogs exist. Our method yielded 132 variable AGN candidates. We then extended our method including the dataset of the ESSENCE project that has been active for 6 years, producing high quality light curves in the R and I bands. We obtained a sample of ˜4800 variable sources, down to R=22, in the whole 12 deg2 ESSENCE field. Among them, a subsample of ˜500 high priority AGN candidates was created using as secondary criterion the shape of the structure function. In a pilot spectroscopic run we have confirmed the AGN nature for nearly all of our candidates.

Device and methods for "gold standard" registration of clinical 3D and 2D cerebral angiograms

NASA Astrophysics Data System (ADS)

Madan, Hennadii; Likar, Boštjan; Pernuš, Franjo; Å piclin, Žiga

2015-03-01

Translation of any novel and existing 3D-2D image registration methods into clinical image-guidance systems is limited due to lack of their objective validation on clinical image datasets. The main reason is that, besides the calibration of the 2D imaging system, a reference or "gold standard" registration is very difficult to obtain on clinical image datasets. In the context of cerebral endovascular image-guided interventions (EIGIs), we present a calibration device in the form of a headband with integrated fiducial markers and, secondly, propose an automated pipeline comprising 3D and 2D image processing, analysis and annotation steps, the result of which is a retrospective calibration of the 2D imaging system and an optimal, i.e., "gold standard" registration of 3D and 2D images. The device and methods were used to create the "gold standard" on 15 datasets of 3D and 2D cerebral angiograms, whereas each dataset was acquired on a patient undergoing EIGI for either aneurysm coiling or embolization of arteriovenous malformation. The use of the device integrated seamlessly in the clinical workflow of EIGI. While the automated pipeline eliminated all manual input or interactive image processing, analysis or annotation. In this way, the time to obtain the "gold standard" was reduced from 30 to less than one minute and the "gold standard" of 3D-2D registration on all 15 datasets of cerebral angiograms was obtained with a sub-0.1 mm accuracy.

Image quality of mean temporal arterial and mean temporal portal venous phase images calculated from low dose dynamic volume perfusion CT datasets in patients with hepatocellular carcinoma and pancreatic cancer.

PubMed

Wang, X; Henzler, T; Gawlitza, J; Diehl, S; Wilhelm, T; Schoenberg, S O; Jin, Z Y; Xue, H D; Smakic, A

2016-11-01

Dynamic volume perfusion CT (dVPCT) provides valuable information on tissue perfusion in patients with hepatocellular carcinoma (HCC) and pancreatic cancer. However, currently dVPCT is often performed in addition to conventional CT acquisitions due to the limited morphologic image quality of dose optimized dVPCT protocols. The aim of this study was to prospectively compare objective and subjective image quality, lesion detectability and radiation dose between mean temporal arterial (mTA) and mean temporal portal venous (mTPV) images calculated from low dose dynamic volume perfusion CT (dVPCT) datasets with linearly blended 120-kVp arterial and portal venous datasets in patients with HCC and pancreatic cancer. All patients gave written informed consent for this institutional review board-approved HIPAA compliant study. 27 consecutive patients (18 men, 9 women, mean age, 69.1 years±9.4) with histologically proven HCC or suspected pancreatic cancer were prospectively enrolled. The study CT protocol included a dVPCT protocol performed with 70 or 80kVp tube voltage (18 spiral acquisitions, 71.2s total acquisition times) and standard dual-energy (90/150kVpSn) arterial and portal venous acquisition performed 25min after the dVPCT. The mTA and mTPV images were manually reconstructed from the 3 to 5 best visually selected single arterial and 3 to 5 best single portal venous phases dVPCT dataset. The linearly blended 120-kVp images were calculated from dual-energy CT (DECT) raw data. Image noise, SNR, and CNR of the liver, abdominal aorta (AA) and main portal vein (PV) were compared between the mTA/mTPV and the linearly blended 120-kVp dual-energy arterial and portal venous datasets, respectively. Subjective image quality was evaluated by two radiologists regarding subjective image noise, sharpness and overall diagnostic image quality using a 5-point Likert Scale. In addition, liver lesion detectability was performed for each liver segment by the two radiologists using the linearly blended120-kVp arterial and portal venous datasets as the reference standard. Image noise, SNR and CNR values of the mTA and mTPV were significantly higher when compared to the corresponding linearly blended arterial and portal venous 120-kVp datasets (all p<0.001) except for image noise within the PV in the portal venous phases (p=0.136). image quality of mTA and mTPV were rated significantly better when compared to the linearly blended 120-kVp arterial and portal venous datasets. Both readers were able to detect all liver lesions found on the linearly blended 120-kVp arterial and portal venous datasets using the mTA and mTPV datasets. The effective radiation dose of the dVPCT was 27.6mSv for the 80kVp protocol and 14.5mSv for the 70kVp protocol. The mean effective radiation dose for the linearly blended 120-kVp arterial and portal venous CT protocol together of the upper abdomen was 5.60mSv±1.48mSv. Our preliminary data suggest that subjective and objective image quality of mTA and mTPV datasets calculated from low-kVp dVPCT datasets is non-inferior when compared to linearly blended 120-kVp arterial and portal venous acquisitions in patients with HCC and pancreatic cancer. Thus, dVPCT could be used as a stand-alone imaging technique without additionally performed conventional arterial and portal venous CT acquisitions. Copyright © 2016 Elsevier Ireland Ltd. All rights reserved.

Deep Learning MR Imaging-based Attenuation Correction for PET/MR Imaging.

PubMed

Liu, Fang; Jang, Hyungseok; Kijowski, Richard; Bradshaw, Tyler; McMillan, Alan B

2018-02-01

Purpose To develop and evaluate the feasibility of deep learning approaches for magnetic resonance (MR) imaging-based attenuation correction (AC) (termed deep MRAC) in brain positron emission tomography (PET)/MR imaging. Materials and Methods A PET/MR imaging AC pipeline was built by using a deep learning approach to generate pseudo computed tomographic (CT) scans from MR images. A deep convolutional auto-encoder network was trained to identify air, bone, and soft tissue in volumetric head MR images coregistered to CT data for training. A set of 30 retrospective three-dimensional T1-weighted head images was used to train the model, which was then evaluated in 10 patients by comparing the generated pseudo CT scan to an acquired CT scan. A prospective study was carried out for utilizing simultaneous PET/MR imaging for five subjects by using the proposed approach. Analysis of covariance and paired-sample t tests were used for statistical analysis to compare PET reconstruction error with deep MRAC and two existing MR imaging-based AC approaches with CT-based AC. Results Deep MRAC provides an accurate pseudo CT scan with a mean Dice coefficient of 0.971 ± 0.005 for air, 0.936 ± 0.011 for soft tissue, and 0.803 ± 0.021 for bone. Furthermore, deep MRAC provides good PET results, with average errors of less than 1% in most brain regions. Significantly lower PET reconstruction errors were realized with deep MRAC (-0.7% ± 1.1) compared with Dixon-based soft-tissue and air segmentation (-5.8% ± 3.1) and anatomic CT-based template registration (-4.8% ± 2.2). Conclusion The authors developed an automated approach that allows generation of discrete-valued pseudo CT scans (soft tissue, bone, and air) from a single high-spatial-resolution diagnostic-quality three-dimensional MR image and evaluated it in brain PET/MR imaging. This deep learning approach for MR imaging-based AC provided reduced PET reconstruction error relative to a CT-based standard within the brain compared with current MR imaging-based AC approaches. © RSNA, 2017 Online supplemental material is available for this article.

Improving protein fold recognition by extracting fold-specific features from predicted residue-residue contacts.

PubMed

Zhu, Jianwei; Zhang, Haicang; Li, Shuai Cheng; Wang, Chao; Kong, Lupeng; Sun, Shiwei; Zheng, Wei-Mou; Bu, Dongbo

2017-12-01

Accurate recognition of protein fold types is a key step for template-based prediction of protein structures. The existing approaches to fold recognition mainly exploit the features derived from alignments of query protein against templates. These approaches have been shown to be successful for fold recognition at family level, but usually failed at superfamily/fold levels. To overcome this limitation, one of the key points is to explore more structurally informative features of proteins. Although residue-residue contacts carry abundant structural information, how to thoroughly exploit these information for fold recognition still remains a challenge. In this study, we present an approach (called DeepFR) to improve fold recognition at superfamily/fold levels. The basic idea of our approach is to extract fold-specific features from predicted residue-residue contacts of proteins using deep convolutional neural network (DCNN) technique. Based on these fold-specific features, we calculated similarity between query protein and templates, and then assigned query protein with fold type of the most similar template. DCNN has showed excellent performance in image feature extraction and image recognition; the rational underlying the application of DCNN for fold recognition is that contact likelihood maps are essentially analogy to images, as they both display compositional hierarchy. Experimental results on the LINDAHL dataset suggest that even using the extracted fold-specific features alone, our approach achieved success rate comparable to the state-of-the-art approaches. When further combining these features with traditional alignment-related features, the success rate of our approach increased to 92.3%, 82.5% and 78.8% at family, superfamily and fold levels, respectively, which is about 18% higher than the state-of-the-art approach at fold level, 6% higher at superfamily level and 1% higher at family level. An independent assessment on SCOP_TEST dataset showed consistent performance improvement, indicating robustness of our approach. Furthermore, bi-clustering results of the extracted features are compatible with fold hierarchy of proteins, implying that these features are fold-specific. Together, these results suggest that the features extracted from predicted contacts are orthogonal to alignment-related features, and the combination of them could greatly facilitate fold recognition at superfamily/fold levels and template-based prediction of protein structures. Source code of DeepFR is freely available through https://github.com/zhujianwei31415/deepfr, and a web server is available through http://protein.ict.ac.cn/deepfr. zheng@itp.ac.cn or dbu@ict.ac.cn. Supplementary data are available at Bioinformatics online. © The Author (2017). Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oup.com

GIS-based technology for marine geohazards in LW3-1 Gas Field of the South China Sea

NASA Astrophysics Data System (ADS)

Su, Tianyun; Liu, Lejun; Li, Xishuang; Hu, Guanghai; Liu, Haixing; Zhou, Lin

2013-04-01

The exploration and exploitation of deep-water oil-gas are apt to be suffered from high-risk geo-hazards such as submarine landslide, soft clay creep, shallow gas, excess pore-water pressure, mud volcano or mud diaper, salt dome and so on. Therefore, it is necessary to survey the seafloor topography, identify the unfavourable geological risks and investigate their environment and mechanism before exploiting the deep-water oil-gas. Because of complex environment, the submarine phenomenon and features, like marine geohazards, can not be recognized directly. Multi-disciplinary data are acquired and analysed comprehensively in order to get more clear understanding about the submarine processes. The data include multi-beam bathymetry data, sidescan sonar images, seismic data, shallow-bottom profiling images, boring data, etc.. Such data sets nowadays increase rapidly to large amounts, but may be heterogeneous and have different resolutions. It is difficult to make good management and utilization of such submarine data with traditional means. GIS technology can provide efficient and powerful tools or services in such aspects as spatial data management, processing, analysis and visualization. They further promote the submarine scientific research and engineering development. The Liwan 3-1 Gas Field, the first deep-water gas field in China, is located in the Zhu II Depression in the Zhujiang Basin along the continental slope of the northern South China Sea. The exploitation of this field is designed to establish subsea wellhead and to use submarine pipeline for the transportation of oil. The deep-water section of the pipeline route in the gas field is to be selected to pass through the northern continental slope of the South China Sea. To avoid huge economic loss and ecological environmental damage, it is necessary to evaluate the geo-hazards for the establishment and safe operation of the pipeline. Based on previous scientific research results, several survey cruises have been carried out with ships and AUV to collect multidisciplinary and massive submarine data such as multi-beam bathymetric data, sidescan sonar images, shallow-bottom profiling images, high-resolution multi-channel seismic data and boring test data. In order to make good use of these precious data, GIS technology is used in our research. Data model is designed to depict the structure, organization and relationship between multi disciplinary submarine data. With these data models, database is established to manage and share the attribute and spatial data effectively. The spatial datasets, such as contours, TIN models, DEM models, etc., can be generated. Some submarine characteristics, such as slope, aspects, curvature, landslide volume, etc., can be calculated and extracted with spatial analysis tools. The thematic map can be produced easily based on database and generated spatial dataset. Through thematic map, the multidisciplinary data spatial relationship can be easily established and provide helpful information for regional submarine geohazards identification, assessments and prediction. The produced thematic map of the LW3-1 Gas Field, reveal the strike of the seafloor topography to be NE to SW. Five geomorphological zones have been divided, which include the outer continental shelf margin zone with sand waves and mega-ripples, the continental slope zone with coral reefs and sand waves, the continental slope zone with a monocline shape, the continental slope zone with fault terraces and the continental slope zone with turbidity current deposits.

Object detection approach using generative sparse, hierarchical networks with top-down and lateral connections for combining texture/color detection and shape/contour detection

DOE Office of Scientific and Technical Information (OSTI.GOV)

Paiton, Dylan M.; Kenyon, Garrett T.; Brumby, Steven P.

An approach to detecting objects in an image dataset may combine texture/color detection, shape/contour detection, and/or motion detection using sparse, generative, hierarchical models with lateral and top-down connections. A first independent representation of objects in an image dataset may be produced using a color/texture detection algorithm. A second independent representation of objects in the image dataset may be produced using a shape/contour detection algorithm. A third independent representation of objects in the image dataset may be produced using a motion detection algorithm. The first, second, and third independent representations may then be combined into a single coherent output using amore » combinatorial algorithm.« less

Optoacoustic imaging in five dimensions

NASA Astrophysics Data System (ADS)

Deán-Ben, X. L.; Gottschalk, Sven; Fehm, Thomas F.; Razansky, Daniel

2015-03-01

We report on an optoacoustic imaging system capable of acquiring volumetric multispectral optoacoustic data in real time. The system is based on simultaneous acquisition of optoacoustic signals from 256 different tomographic projections by means of a spherical matrix array. Thereby, volumetric reconstructions can be done at high frame rate, only limited by the pulse repetition rate of the laser. The developed tomographic approach presents important advantages over previously reported systems that use scanning for attaining volumetric optoacoustic data. First, dynamic processes, such as the biodistribution of optical biomarkers, can be monitored in the entire volume of interest. Second, out-of-plane and motion artifacts that could degrade the image quality when imaging living specimens can be avoided. Finally, real-time 3D performance can obviously save time required for experimental and clinical observations. The feasibility of optoacoustic imaging in five dimensions, i.e. real time acquisition of volumetric datasets at multiple wavelengths, is reported. In this way, volumetric images of spectrally resolved chromophores are rendered in real time, thus offering an unparallel imaging performance among the current bio-imaging modalities. This performance is subsequently showcased by video-rate visualization of in vivo hemodynamic changes in mouse brain and handheld visualization of blood oxygenation in deep human vessels. The newly discovered capacities open new prospects for translating the optoacoustic technology into highly performing imaging modality for biomedical research and clinical practice with multiple applications envisioned, from cardiovascular and cancer diagnostics to neuroimaging and ophthalmology.