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

Deep Learning in Medical Image Analysis

This review covers computer-assisted analysis of images in the field of medical imaging. Recent advances in machine learning, especially with regard to deep learning, are helping to identify, classify, and quantify patterns in medical images. At the core of these advances is the ability to exploit hierarchical feature representations learned solely from data, instead of features designed by hand according to domain-specific knowledge. Deep learning is rapidly becoming the state of the art, leading to enhanced performance in various medical applications. We introduce the fundamentals of deep learning methods and review their successes in image registration, detection of anatomical and cellular structures, tissue segmentation, computer-aided disease diagnosis and prognosis, and so on. We conclude by discussing research issues and suggesting future directions for further improvement.

Low-Grade Glioma Segmentation Based on CNN with Fully Connected CRF

This work proposed a novel automatic three-dimensional (3D) magnetic resonance imaging (MRI) segmentation method which would be widely used in the clinical diagnosis of the most common and aggressive brain tumor, namely, glioma. The method combined a multipathway convolutional neural network (CNN) and fully connected conditional random field (CRF). Firstly, 3D information was introduced into the CNN which makes more accurate recognition of glioma with low contrast. Then, fully connected CRF was added as a postprocessing step which purposed more delicate delineation of glioma boundary. The method was applied to T2flair MRI images of 160 low-grade glioma patients. With 59 cases of data training and manual segmentation as the ground truth, the Dice similarity coefficient (DSC) of our method was 0.85 for the test set of 101 MRI images. The results of our method were better than those of another state-of-the-art CNN method, which gained the DSC of 0.76 for the same dataset. It proved that our method could produce better results for the segmentation of low-grade gliomas.

Lung Cancer Pathological Image Analysis Using a Hidden Potts Model

Nowadays, many biological data are acquired via images. In this article, we study the pathological images scanned from 205 patients with lung cancer with the goal to find out the relationship between the survival time and the spatial distribution of different types of cells, including lymphocyte, stroma, and tumor cells. Toward this goal, we model the spatial distribution of different types of cells using a modified Potts model for which the parameters represent interactions between different types of cells and estimate the parameters of the Potts model using the double Metropolis-Hastings algorithm. The double Metropolis-Hastings algorithm allows us to simulate samples approximately from a distribution with an intractable normalizing constant. Our numerical results indicate that the spatial interaction between the lymphocyte and tumor cells is significantly associated with the patient's survival time, and it can be used together with the cell count information to predict the survival of the patients.

Multi-level deep supervised networks for retinal vessel segmentation

PURPOSE

Changes in the appearance of retinal blood vessels are an important indicator for various ophthalmologic and cardiovascular diseases, including diabetes, hypertension, arteriosclerosis, and choroidal neovascularization. Vessel segmentation from retinal images is very challenging because of low blood vessel contrast, intricate vessel topology, and the presence of pathologies such as microaneurysms and hemorrhages. To overcome these challenges, we propose a neural network-based method for vessel segmentation.

METHODS

A deep supervised fully convolutional network is developed by leveraging multi-level hierarchical features of the deep networks. To improve the discriminative capability of features in lower layers of the deep network and guide the gradient back propagation to overcome gradient vanishing, deep supervision with auxiliary classifiers is incorporated in some intermediate layers of the network. Moreover, the transferred knowledge learned from other domains is used to alleviate the issue of insufficient medical training data. The proposed approach does not rely on hand-crafted features and needs no problem-specific preprocessing or postprocessing, which reduces the impact of subjective factors.

RESULTS

We evaluate the proposed method on three publicly available databases, the DRIVE, STARE, and CHASE_DB1 databases. Extensive experiments demonstrate that our approach achieves better or comparable performance to state-of-the-art methods with a much faster processing speed, making it suitable for real-world clinical applications. The results of cross-training experiments demonstrate its robustness with respect to the training set.

CONCLUSIONS

The proposed approach segments retinal vessels accurately with a much faster processing speed and can be easily applied to other biomedical segmentation tasks.

A two-step convolutional neural network based computer-aided detection scheme for automatically segmenting adipose tissue volume depicting on CT images

Accurately assessment of adipose tissue volume inside a human body plays an important role in predicting disease or cancer risk, diagnosis and prognosis. In order to overcome limitation of using only one subjectively selected CT image slice to estimate size of fat areas, this study aims to develop and test a computer-aided detection (CAD) scheme based on deep learning technique to automatically segment subcutaneous fat areas (SFA) and visceral fat areas (VFA) depicting on volumetric CT images. A retrospectively collected CT image dataset was divided into two independent training and testing groups. The proposed CAD framework consisted of two steps with two convolution neural networks (CNNs) namely, Selection-CNN and Segmentation-CNN. The first CNN was trained using 2,240 CT slices to select abdominal CT slices depicting SFA and VFA. The second CNN was trained with 84,000pixel patches and applied to the selected CT slices to identify fat-related pixels and assign them into SFA and VFA classes. Comparing to the manual CT slice selection and fat pixel segmentation results, the accuracy of CT slice selection using the Selection-CNN yielded 95.8%, while the accuracy of fat pixel segmentation using the Segmentation-CNN was 96.8%. This study demonstrated the feasibility of applying a new deep learning based CAD scheme to automatically recognize abdominal section of human body from CT scans and segment SFA and VFA from volumetric CT data with high accuracy or agreement with the manual segmentation results.

Ultrasound Standard Plane Detection Using a Composite Neural Network Framework

Ultrasound (US) imaging is a widely used screening tool for obstetric examination and diagnosis. Accurate acquisition of fetal standard planes with key anatomical structures is very crucial for substantial biometric measurement and diagnosis. However, the standard plane acquisition is a labor-intensive task and requires operator equipped with a thorough knowledge of fetal anatomy. Therefore, automatic approaches are highly demanded in clinical practice to alleviate the workload and boost the examination efficiency. The automatic detection of standard planes from US videos remains a challenging problem due to the high intraclass and low interclass variations of standard planes, and the relatively low image quality. Unlike previous studies which were specifically designed for individual anatomical standard planes, respectively, we present a general framework for the automatic identification of different standard planes from US videos. Distinct from conventional way that devises hand-crafted visual features for detection, our framework explores in- and between-plane feature learning with a novel composite framework of the convolutional and recurrent neural networks. To further address the issue of limited training data, a multitask learning framework is implemented to exploit common knowledge across detection tasks of distinctive standard planes for the augmentation of feature learning. Extensive experiments have been conducted on hundreds of US fetus videos to corroborate the better efficacy of the proposed framework on the difficult standard plane detection problem.

DeepPap: Deep Convolutional Networks for Cervical Cell Classification

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.

Spotting L3 slice in CT scans using deep convolutional network and transfer learning

In this article, we present a complete automated system for spotting a particular slice in a complete 3D Computed Tomography exam (CT scan). Our approach does not require any assumptions on which part of the patient's body is covered by the scan. It relies on an original machine learning regression approach. Our models are learned using the transfer learning trick by exploiting deep architectures that have been pre-trained on imageNet database, and therefore it requires very little annotation for its training. The whole pipeline consists of three steps: i) conversion of the CT scans into Maximum Intensity Projection (MIP) images, ii) prediction from a Convolutional Neural Network (CNN) applied in a sliding window fashion over the MIP image, and iii) robust analysis of the prediction sequence to predict the height of the desired slice within the whole CT scan. Our approach is applied to the detection of the third lumbar vertebra (L3) slice that has been found to be representative to the whole body composition. Our system is evaluated on a database collected in our clinical center, containing 642 CT scans from different patients. We obtained an average localization error of 1.91±2.69 slices (less than 5 mm) in an average time of less than 2.5 s/CT scan, allowing integration of the proposed system into daily clinical routines.

A structured latent model for ovarian carcinoma subtyping from histopathology slides

Accurate subtyping of ovarian carcinomas is an increasingly critical and often challenging diagnostic process. This work focuses on the development of an automatic classification model for ovarian carcinoma subtyping. Specifically, we present a novel clinically inspired contextual model for histopathology image subtyping of ovarian carcinomas. A whole slide image is modelled using a collection of tissue patches extracted at multiple magnifications. An efficient and effective feature learning strategy is used for feature representation of a tissue patch. The locations of salient, discriminative tissue regions are treated as latent variables allowing the model to explicitly ignore portions of the large tissue section that are unimportant for classification. These latent variables are considered in a structured formulation to model the contextual information represented from the multi-magnification analysis of tissues. A novel, structured latent support vector machine formulation is defined and used to combine information from multiple magnifications while simultaneously operating within the latent variable framework. The structural and contextual nature of our method addresses the challenges of intra-class variation and pathologists' workload, which are prevalent in histopathology image classification. Extensive experiments on a dataset of 133 patients demonstrate the efficacy and accuracy of the proposed method against state-of-the-art approaches for histopathology image classification. We achieve an average multi-class classification accuracy of 90%, outperforming existing works while obtaining substantial agreement with six clinicians tested on the same dataset.

FUIQA: Fetal Ultrasound Image Quality Assessment With Deep Convolutional Networks

The quality of ultrasound (US) images for the obstetric examination is crucial for accurate biometric measurement. However, manual quality control is a labor intensive process and often impractical in a clinical setting. To improve the efficiency of examination and alleviate the measurement error caused by improper US scanning operation and slice selection, a computerized fetal US image quality assessment (FUIQA) scheme is proposed to assist the implementation of US image quality control in the clinical obstetric examination. The proposed FUIQA is realized with two deep convolutional neural network models, which are denoted as L-CNN and C-CNN, respectively. The L-CNN aims to find the region of interest (ROI) of the fetal abdominal region in the US image. Based on the ROI found by the L-CNN, the C-CNN evaluates the image quality by assessing the goodness of depiction for the key structures of stomach bubble and umbilical vein. To further boost the performance of the L-CNN, we augment the input sources of the neural network with the local phase features along with the original US data. It will be shown that the heterogeneous input sources will help to improve the performance of the L-CNN. The performance of the proposed FUIQA is compared with the subjective image quality evaluation results from three medical doctors. With comprehensive experiments, it will be illustrated that the computerized assessment with our FUIQA scheme can be comparable to the subjective ratings from medical doctors.

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

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.

Automating cell detection and classification in human brain fluorescent microscopy images using dictionary learning and sparse coding

BACKGROUND

Immunofluorescence (IF) plays a major role in quantifying protein expression in situ and understanding cell function. It is widely applied in assessing disease mechanisms and in drug discovery research. Automation of IF analysis can transform studies using experimental cell models. However, IF analysis of postmortem human tissue relies mostly on manual interaction, often subjected to low-throughput and prone to error, leading to low inter and intra-observer reproducibility. Human postmortem brain samples challenges neuroscientists because of the high level of autofluorescence caused by accumulation of lipofuscin pigment during aging, hindering systematic analyses. We propose a method for automating cell counting and classification in IF microscopy of human postmortem brains. Our algorithm speeds up the quantification task while improving reproducibility.

NEW METHOD

Dictionary learning and sparse coding allow for constructing improved cell representations using IF images. These models are input for detection and segmentation methods. Classification occurs by means of color distances between cells and a learned set.

RESULTS

Our method successfully detected and classified cells in 49 human brain images. We evaluated our results regarding true positive, false positive, false negative, precision, recall, false positive rate and F1 score metrics. We also measured user-experience and time saved compared to manual countings.

COMPARISON WITH EXISTING METHODS

We compared our results to four open-access IF-based cell-counting tools available in the literature. Our method showed improved accuracy for all data samples.

CONCLUSION

The proposed method satisfactorily detects and classifies cells from human postmortem brain IF images, with potential to be generalized for applications in other counting tasks.

Automated Melanoma Recognition in Dermoscopy Images via Very Deep Residual Networks

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.

Transfer learning from RF to B-mode temporal enhanced ultrasound features for prostate cancer detection

PURPOSE

We present a method for prostate cancer (PCa) detection using temporal enhanced ultrasound (TeUS) data obtained either from radiofrequency (RF) ultrasound signals or B-mode images.

METHODS

For the first time, we demonstrate that by applying domain adaptation and transfer learning methods, a tissue classification model trained on TeUS RF data (source domain) can be deployed for classification using TeUS B-mode data alone (target domain), where both data are obtained on the same ultrasound scanner. This is a critical step for clinical translation of tissue classification techniques that primarily rely on accessing RF data, since this imaging modality is not readily available on all commercial scanners in clinics. Proof of concept is provided for in vivo characterization of PCa using TeUS B-mode data, where different nonlinear processing filters in the pipeline of the RF to B-mode conversion result in a distribution shift between the two domains.

RESULTS

Our in vivo study includes data obtained in MRI-guided targeted procedure for prostate biopsy. We achieve comparable area under the curve using TeUS RF and B-mode data for medium to large cancer tumor sizes in biopsy cores (>4 mm).

CONCLUSION

Our result suggests that the proposed adaptation technique is successful in reducing the divergence between TeUS RF and B-mode data.

Multiscale Rotation-Invariant Convolutional Neural Networks for Lung Texture Classification

We propose a new multiscale rotation-invariant convolutional neural network (MRCNN) model for classifying various lung tissue types on high-resolution computed tomography. MRCNN employs Gabor-local binary pattern that introduces a good property in image analysis-invariance to image scales and rotations. In addition, we offer an approach to deal with the problems caused by imbalanced number of samples between different classes in most of the existing works, accomplished by changing the overlapping size between the adjacent patches. Experimental results on a public interstitial lung disease database show a superior performance of the proposed method to state of the art.

Mass detection in digital breast tomosynthesis: Deep convolutional neural network with transfer learning from mammography

PURPOSE

Develop a computer-aided detection (CAD) system for masses in digital breast tomosynthesis (DBT) volume using a deep convolutional neural network (DCNN) with transfer learning from mammograms.

METHODS

A data set containing 2282 digitized film and digital mammograms and 324 DBT volumes were collected with IRB approval. The mass of interest on the images was marked by an experienced breast radiologist as reference standard. The data set was partitioned into a training set (2282 mammograms with 2461 masses and 230 DBT views with 228 masses) and an independent test set (94 DBT views with 89 masses). For DCNN training, the region of interest (ROI) containing the mass (true positive) was extracted from each image. False positive (FP) ROIs were identified at prescreening by their previously developed CAD systems. After data augmentation, a total of 45 072 mammographic ROIs and 37 450 DBT ROIs were obtained. Data normalization and reduction of non-uniformity in the ROIs across heterogeneous data was achieved using a background correction method applied to each ROI. A DCNN with four convolutional layers and three fully connected (FC) layers was first trained on the mammography data. Jittering and dropout techniques were used to reduce overfitting. After training with the mammographic ROIs, all weights in the first three convolutional layers were frozen, and only the last convolution layer and the FC layers were randomly initialized again and trained using the DBT training ROIs. The authors compared the performances of two CAD systems for mass detection in DBT: one used the DCNN-based approach and the other used their previously developed feature-based approach for FP reduction. The prescreening stage was identical in both systems, passing the same set of mass candidates to the FP reduction stage. For the feature-based CAD system, 3D clustering and active contour method was used for segmentation; morphological, gray level, and texture features were extracted and merged with a linear discriminant classifier to score the detected masses. For the DCNN-based CAD system, ROIs from five consecutive slices centered at each candidate were passed through the trained DCNN and a mass likelihood score was generated. The performances of the CAD systems were evaluated using free-response ROC curves and the performance difference was analyzed using a non-parametric method.

RESULTS

Before transfer learning, the DCNN trained only on mammograms with an AUC of 0.99 classified DBT masses with an AUC of 0.81 in the DBT training set. After transfer learning with DBT, the AUC improved to 0.90. For breast-based CAD detection in the test set, the sensitivity for the feature-based and the DCNN-based CAD systems was 83% and 91%, respectively, at 1 FP/DBT volume. The difference between the performances for the two systems was statistically significant (p-value < 0.05).

CONCLUSIONS

The image patterns learned from the mammograms were transferred to the mass detection on DBT slices through the DCNN. This study demonstrated that large data sets collected from mammography are useful for developing new CAD systems for DBT, alleviating the problem and effort of collecting entirely new large data sets for the new modality.

Automatic Segmentation and Quantification of White and Brown Adipose Tissues from PET/CT Scans

In this paper, we investigate the automatic detection of white and brown adipose tissues using Positron Emission Tomography/Computed Tomography (PET/CT) scans, and develop methods for the quantification of these tissues at the whole-body and body-region levels. We propose a patient-specific automatic adiposity analysis system with two modules. In the first module, we detect white adipose tissue (WAT) and its two sub-types from CT scans: Visceral Adipose Tissue (VAT) and Subcutaneous Adipose Tissue (SAT). This process relies conventionally on manual or semi-automated segmentation, leading to inefficient solutions. Our novel framework addresses this challenge by proposing an unsupervised learning method to separate VAT from SAT in the abdominal region for the clinical quantification of central obesity. This step is followed by a context driven label fusion algorithm through sparse 3D Conditional Random Fields (CRF) for volumetric adiposity analysis. In the second module, we automatically detect, segment, and quantify brown adipose tissue (BAT) using PET scans because unlike WAT, BAT is metabolically active. After identifying BAT regions using PET, we perform a co-segmentation procedure utilizing asymmetric complementary information from PET and CT. Finally, we present a new probabilistic distance metric for differentiating BAT from non-BAT regions. Both modules are integrated via an automatic body-region detection unit based on one-shot learning. Experimental evaluations conducted on 151 PET/CT scans achieve state-of-the-art performances in both central obesity as well as brown adiposity quantification.

Texture analysis in radiology: Does the emperor have no clothes?

Texture analysis is more and more frequently used in radiology research. Is this a new technology, and if not, what has changed? Is texture analysis the great diagnostic and prognostic tool we have been searching for in radiology? This commentary answers these questions and places texture analysis into its proper perspective.

Deep Learning in Visual Computing and Signal Processing

Deep learning is a subfield of machine learning, which aims to learn a hierarchy of features from input data. Nowadays, researchers have intensively investigated deep learning algorithms for solving challenging problems in many areas such as image classification, speech recognition, signal processing, and natural language processing. In this study, we not only review typical deep learning algorithms in computer vision and signal processing but also provide detailed information on how to apply deep learning to specific areas such as road crack detection, fault diagnosis, and human activity detection. Besides, this study also discusses the challenges of designing and training deep neural networks.

Deep Learning for Health Informatics

With a massive influx of multimodality data, the role of data analytics in health informatics has grown rapidly in the last decade. This has also prompted increasing interests in the generation of analytical, data driven models based on machine learning in health informatics. Deep learning, a technique with its foundation in artificial neural networks, is emerging in recent years as a powerful tool for machine learning, promising to reshape the future of artificial intelligence. Rapid improvements in computational power, fast data storage, and parallelization have also contributed to the rapid uptake of the technology in addition to its predictive power and ability to generate automatically optimized high-level features and semantic interpretation from the input data. This article presents a comprehensive up-to-date review of research employing deep learning in health informatics, providing a critical analysis of the relative merit, and potential pitfalls of the technique as well as its future outlook. The paper mainly focuses on key applications of deep learning in the fields of translational bioinformatics, medical imaging, pervasive sensing, medical informatics, and public health.

Integrating Online and Offline Three-Dimensional Deep Learning for Automated Polyp Detection in Colonoscopy Videos

Automated polyp detection in colonoscopy videos has been demonstrated to be a promising way for colorectal cancer prevention and diagnosis. Traditional manual screening is time consuming, operator dependent, and error prone; hence, automated detection approach is highly demanded in clinical practice. However, automated polyp detection is very challenging due to high intraclass variations in polyp size, color, shape, and texture, and low interclass variations between polyps and hard mimics. In this paper, we propose a novel offline and online three-dimensional (3-D) deep learning integration framework by leveraging the 3-D fully convolutional network (3D-FCN) to tackle this challenging problem. Compared with the previous methods employing hand-crafted features or 2-D convolutional neural network, the 3D-FCN is capable of learning more representative spatio-temporal features from colonoscopy videos, and hence has more powerful discrimination capability. More importantly, we propose a novel online learning scheme to deal with the problem of limited training data by harnessing the specific information of an input video in the learning process. We integrate offline and online learning to effectively reduce the number of false positives generated by the offline network and further improve the detection performance. Extensive experiments on the dataset of MICCAI 2015 Challenge on Polyp Detection demonstrated the better performance of our method when compared with other competitors.

Multisource Transfer Learning With Convolutional Neural Networks for Lung Pattern Analysis

Early diagnosis of interstitial lung diseases is crucial for their treatment, but even experienced physicians find it difficult, as their clinical manifestations are similar. In order to assist with the diagnosis, computer-aided diagnosis systems have been developed. These commonly rely on a fixed scale classifier that scans CT images, recognizes textural lung patterns, and generates a map of pathologies. In a previous study, we proposed a method for classifying lung tissue patterns using a deep convolutional neural network (CNN), with an architecture designed for the specific problem. In this study, we present an improved method for training the proposed network by transferring knowledge from the similar domain of general texture classification. Six publicly available texture databases are used to pretrain networks with the proposed architecture, which are then fine-tuned on the lung tissue data. The resulting CNNs are combined in an ensemble and their fused knowledge is compressed back to a network with the original architecture. The proposed approach resulted in an absolute increase of about 2% in the performance of the proposed CNN. The results demonstrate the potential of transfer learning in the field of medical image analysis, indicate the textural nature of the problem and show that the method used for training a network can be as important as designing its architecture.

An Ensemble of Fine-Tuned Convolutional Neural Networks for Medical Image Classification

The availability of medical imaging data from clinical archives, research literature, and clinical manuals, coupled with recent advances in computer vision offer the opportunity for image-based diagnosis, teaching, and biomedical research. However, the content and semantics of an image can vary depending on its modality and as such the identification of image modality is an important preliminary step. The key challenge for automatically classifying the modality of a medical image is due to the visual characteristics of different modalities: some are visually distinct while others may have only subtle differences. This challenge is compounded by variations in the appearance of images based on the diseases depicted and a lack of sufficient training data for some modalities. In this paper, we introduce a new method for classifying medical images that uses an ensemble of different convolutional neural network (CNN) architectures. CNNs are a state-of-the-art image classification technique that learns the optimal image features for a given classification task. We hypothesise that different CNN architectures learn different levels of semantic image representation and thus an ensemble of CNNs will enable higher quality features to be extracted. Our method develops a new feature extractor by fine-tuning CNNs that have been initialized on a large dataset of natural images. The fine-tuning process leverages the generic image features from natural images that are fundamental for all images and optimizes them for the variety of medical imaging modalities. These features are used to train numerous multiclass classifiers whose posterior probabilities are fused to predict the modalities of unseen images. Our experiments on the ImageCLEF 2016 medical image public dataset (30 modalities; 6776 training images, and 4166 test images) show that our ensemble of fine-tuned CNNs achieves a higher accuracy than established CNNs. Our ensemble also achieves a higher accuracy than methods in the literature evaluated on the same benchmark dataset and is only overtaken by those methods that source additional training data.

Automatic Detection and Classification of Colorectal Polyps by Transferring Low-Level CNN Features From Nonmedical Domain

Colorectal cancer (CRC) is a leading cause of cancer deaths worldwide. Although polypectomy at early stage reduces CRC incidence, 90% of the polyps are small and diminutive, where removal of them poses risks to patients that may outweigh the benefits. Correctly detecting and predicting polyp type during colonoscopy allows endoscopists to resect and discard the tissue without submitting it for histology, saving time, and costs. Nevertheless, human visual observation of early stage polyps varies. Therefore, this paper aims at developing a fully automatic algorithm to detect and classify hyperplastic and adenomatous colorectal polyps. Adenomatous polyps should be removed, whereas distal diminutive hyperplastic polyps are considered clinically insignificant and may be left in situ . A novel transfer learning application is proposed utilizing features learned from big nonmedical datasets with 1.4-2.5 million images using deep convolutional neural network. The endoscopic images we collected for experiment were taken under random lighting conditions, zooming and optical magnification, including 1104 endoscopic nonpolyp images taken under both white-light and narrowband imaging (NBI) endoscopy and 826 NBI endoscopic polyp images, of which 263 images were hyperplasia and 563 were adenoma as confirmed by histology. The proposed method identified polyp images from nonpolyp images in the beginning followed by predicting the polyp histology. When compared with visual inspection by endoscopists, the results of this study show that the proposed method has similar precision (87.3% versus 86.4%) but a higher recall rate (87.6% versus 77.0%) and a higher accuracy (85.9% versus 74.3%). In conclusion, automatic algorithms can assist endoscopists in identifying polyps that are adenomatous but have been incorrectly judged as hyperplasia and, therefore, enable timely resection of these polyps at an early stage before they develop into invasive cancer.

Deep learning based classification of breast tumors with shear-wave elastography

This study aims to build a deep learning (DL) architecture for automated extraction of learned-from-data image features from the shear-wave elastography (SWE), and to evaluate the DL architecture in differentiation between benign and malignant breast tumors. We construct a two-layer DL architecture for SWE feature extraction, comprised of the point-wise gated Boltzmann machine (PGBM) and the restricted Boltzmann machine (RBM). The PGBM contains task-relevant and task-irrelevant hidden units, and the task-relevant units are connected to the RBM. Experimental evaluation was performed with five-fold cross validation on a set of 227 SWE images, 135 of benign tumors and 92 of malignant tumors, from 121 patients. The features learned with our DL architecture were compared with the statistical features quantifying image intensity and texture. Results showed that the DL features achieved better classification performance with an accuracy of 93.4%, a sensitivity of 88.6%, a specificity of 97.1%, and an area under the receiver operating characteristic curve of 0.947. The DL-based method integrates feature learning with feature selection on SWE. It may be potentially used in clinical computer-aided diagnosis of breast cancer.

Bladder Cancer Segmentation in CT for Treatment Response Assessment: Application of Deep-Learning Convolution Neural Network-A Pilot Study

Assessing the response of bladder cancer to neoadjuvant chemotherapy is crucial for reducing morbidity and increasing quality of life of patients. Changes in tumor volume during treatment is generally used to predict treatment outcome. We are developing a method for bladder cancer segmentation in CT using a pilot data set of 62 cases. 65 000 regions of interests were extracted from pre-treatment CT images to train a deep-learning convolution neural network (DL-CNN) for tumor boundary detection using leave-one-case-out cross-validation. The results were compared to our previous AI-CALS method. For all lesions in the data set, the longest diameter and its perpendicular were measured by two radiologists, and 3D manual segmentation was obtained from one radiologist. The World Health Organization (WHO) criteria and the Response Evaluation Criteria In Solid Tumors (RECIST) were calculated, and the prediction accuracy of complete response to chemotherapy was estimated by the area under the receiver operating characteristic curve (AUC). The AUCs were 0.73 ± 0.06, 0.70 ± 0.07, and 0.70 ± 0.06, respectively, for the volume change calculated using DL-CNN segmentation, the AI-CALS and the manual contours. The differences did not achieve statistical significance. The AUCs using the WHO criteria were 0.63 ± 0.07 and 0.61 ± 0.06, while the AUCs using RECIST were 0.65 ± 007 and 0.63 ± 0.06 for the two radiologists, respectively. Our results indicate that DL-CNN can produce accurate bladder cancer segmentation for calculation of tumor size change in response to treatment. The volume change performed better than the estimations from the WHO criteria and RECIST for the prediction of complete response.

DCAN: Deep contour-aware networks for object instance segmentation from histology images

In histopathological image analysis, the morphology of histological structures, such as glands and nuclei, has been routinely adopted by pathologists to assess the malignancy degree of adenocarcinomas. Accurate detection and segmentation of these objects of interest from histology images is an essential prerequisite to obtain reliable morphological statistics for quantitative diagnosis. While manual annotation is error-prone, time-consuming and operator-dependant, automated detection and segmentation of objects of interest from histology images can be very challenging due to the large appearance variation, existence of strong mimics, and serious degeneration of histological structures. In order to meet these challenges, we propose a novel deep contour-aware network (DCAN) under a unified multi-task learning framework for more accurate detection and segmentation. In the proposed network, multi-level contextual features are explored based on an end-to-end fully convolutional network (FCN) to deal with the large appearance variation. We further propose to employ an auxiliary supervision mechanism to overcome the problem of vanishing gradients when training such a deep network. More importantly, our network can not only output accurate probability maps of histological objects, but also depict clear contours simultaneously for separating clustered object instances, which further boosts the segmentation performance. Our method ranked the first in two histological object segmentation challenges, including 2015 MICCAI Gland Segmentation Challenge and 2015 MICCAI Nuclei Segmentation Challenge. Extensive experiments on these two challenging datasets demonstrate the superior performance of our method, surpassing all the other methods by a significant margin.

Exploring Deep Learning and Transfer Learning for Colonic Polyp Classification

Recently, Deep Learning, especially through Convolutional Neural Networks (CNNs) has been widely used to enable the extraction of highly representative features. This is done among the network layers by filtering, selecting, and using these features in the last fully connected layers for pattern classification. However, CNN training for automated endoscopic image classification still provides a challenge due to the lack of large and publicly available annotated databases. In this work we explore Deep Learning for the automated classification of colonic polyps using different configurations for training CNNs from scratch (or full training) and distinct architectures of pretrained CNNs tested on 8-HD-endoscopic image databases acquired using different modalities. We compare our results with some commonly used features for colonic polyp classification and the good results suggest that features learned by CNNs trained from scratch and the “off-the-shelf” CNNs features can be highly relevant for automated classification of colonic polyps. Moreover, we also show that the combination of classical features and “off-the-shelf” CNNs features can be a good approach to further improve the results.

Discrimination of Breast Cancer with Microcalcifications on Mammography by Deep Learning

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.