Deep convolutional neural networks have performed remarkably well on many Computer Vision tasks. However, these networks are heavily reliant on big data to avoid overfitting. Overfitting refers to the phenomenon when a network learns a function with very high variance such as to perfectly model the training data. Unfortunately, many application domains do not have access to big data, such as medical image analysis. This survey focuses on Data Augmentation, a data-space solution to the problem of limited data. Data Augmentation encompasses a suite of techniques that enhance the size and quality of training datasets such that better Deep Learning models can be built using them. The image augmentation algorithms discussed in this survey include geometric transformations, color space augmentations, kernel filters, mixing images, random erasing, feature space augmentation, adversarial training, generative adversarial networks, neural style transfer, and meta-learning. The application of augmentation methods based on GANs are heavily covered in this survey. In addition to augmentation techniques, this paper will briefly discuss other characteristics of Data Augmentation such as test-time augmentation, resolution impact, final dataset size, and curriculum learning. This survey will present existing methods for Data Augmentation, promising developments, and meta-level decisions for implementing Data Augmentation. Readers will understand how Data Augmentation can improve the performance of their models and expand limited datasets to take advantage of the capabilities of big data.

/pdf/a-survey-on-image-data-augmentation-for-deep-learning-4ru1bj2hbb.pdf

A survey on Image Data Augmentation for Deep Learning

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.

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

Generative adversarial network in medical imaging: A review.

Small-bowel capsule endoscopy (SBCE)
 1 ESGE recommends that prior to SBCE patients ingest a purgative (2 L of polyethylene glycol [PEG]) for better visualization. Strong recommendation, high quality evidence. However, the optimal timing for taking purgatives is yet to be established. 2 ESGE recommends that SBCE should be performed as an outpatient procedure if possible, since completion rates are higher in outpatients than in inpatients. Strong recommendation, moderate quality evidence. 3 ESGE recommends that patients with pacemakers can safely undergo SBCE without special precautions. Strong recommendation, low quality evidence. 4 ESGE suggests that SBCE can also be safely performed in patients with implantable cardioverter defibrillators and left ventricular assist devices. Weak recommendation, low quality evidence. 5 ESGE recommends the acceptance of qualified nurses and trained technicians as prereaders of capsule endoscopy studies as their competency in identifying pathology is similar to that of medically qualified readers. The responsibility of establishing a diagnosis must however remain with the attending physician. Strong recommendation, moderate quality evidence. 6 ESGE recommends observation in cases of asymptomatic capsule retention. Strong recommendation, moderate quality evidence. In cases where capsule retrieval is indicated, ESGE recommends the use of device-assisted enteroscopy as the method of choice. Strong recommendation, moderate quality evidence. Device-assisted enteroscopy (DAE)
 1 ESGE recommends performing diagnostic DAE as a day-case procedure in patients without significant underlying co-morbidities; in patients with co-morbidities and/or those undergoing a therapeutic procedure, an inpatient stay is recommended. Strong recommendation, low quality evidence The choice between different settings also depends on sedation protocols. Strong recommendation, low quality evidence. 2 ESGE suggests that conscious sedation, deep sedation, and general anesthesia are all acceptable alternatives: the choice between them should be governed by procedure complexity, clinical factors, and local organizational protocols. Weak recommendation, low quality evidence. 3 ESGE recommends that the findings of previous diagnostic investigations should guide the choice of insertion route. Strong recommendation, moderate quality evidence. If the location of the small-bowel lesion is unknown or uncertain, ESGE recommends that the antegrade route should be generally preferred. Strong recommendation, low quality evidence. In the setting of massive overt bleeding, ESGE recommends an initial antegrade approach. Strong recommendation, low quality evidence. 4 ESGE recommends that, for balloon-assisted enteroscopy (i. e., single-balloon enteroscopy [SBE] and double-balloon enteroscopy [DBE]), small-bowel insertion depth should be estimated by counting net advancement of the enteroscope during the insertion phase, with confirmation of this estimate during withdrawal. Strong recommendation, low quality evidence. ESGE recommends that, for spiral enteroscopy, insertion depth should be estimated during withdrawal. Strong recommendation, moderate quality evidence. Since the calculated insertion depth is only a rough estimate, ESGE recommends placing a tattoo to mark the identified lesion and/or the deepest point of insertion. Strong recommendation, low quality evidence. 5 ESGE recommends that all endoscopic therapeutic procedures can be undertaken at the time of DAE. Strong recommendation, moderate quality evidence. Moreover, when therapeutic interventions are performed, additional specific safety measures are needed to prevent complications. Strong recommendation, high quality evidence.

/pdf/small-bowel-capsule-endoscopy-and-device-assisted-4dltzyt3z1.pdf

Small-bowel capsule endoscopy and device-assisted enteroscopy for diagnosis and treatment of small-bowel disorders: European Society of Gastrointestinal Endoscopy (ESGE) Clinical Guideline.

Chest X-ray is the first imaging technique that plays an important role in the diagnosis of COVID-19 disease. Due to the high availability of large-scale annotated image datasets, great success has been achieved using convolutional neural networks (CNN s) for image recognition and classification. However, due to the limited availability of annotated medical images, the classification of medical images remains the biggest challenge in medical diagnosis. Thanks to transfer learning, an effective mechanism that can provide a promising solution by transferring knowledge from generic object recognition tasks to domain-specific tasks. In this paper, we validate and a deep CNN, called Decompose, Transfer, and Compose (DeTraC), for the classification of COVID-19 chest X-ray images. DeTraC can deal with any irregularities in the image dataset by investigating its class boundaries using a class decomposition mechanism. The experimental results showed the capability of DeTraC in the detection of COVID-19 cases from a comprehensive image dataset collected from several hospitals around the world. High accuracy of 93.1% (with a sensitivity of 100%) was achieved by DeTraC in the detection of COVID-19 X-ray images from normal, and severe acute respiratory syndrome cases.

/pdf/classification-of-covid-19-in-chest-x-ray-images-using-2jfcx6h435.pdf

Classification of COVID-19 in chest X-ray images using DeTraC deep convolutional neural network.

The National Library of Medicine (NLM) is developing a digital chest X-ray (CXR) screening system for deployment in resource constrained communities and developing countries worldwide with a focus on early detection of tuberculosis. A critical component in the computer-aided diagnosis of digital CXRs is the automatic detection of the lung regions. In this paper, we present a nonrigid registration-driven robust lung segmentation method using image retrieval-based patient specific adaptive lung models that detects lung boundaries, surpassing state-of-the-art performance. The method consists of three main stages: 1) a content-based image retrieval approach for identifying training images (with masks) most similar to the patient CXR using a partial Radon transform and Bhattacharyya shape similarity measure, 2) creating the initial patient-specific anatomical model of lung shape using SIFT-flow for deformable registration of training masks to the patient CXR, and 3) extracting refined lung boundaries using a graph cuts optimization approach with a customized energy function. Our average accuracy of 95.4% on the public JSRT database is the highest among published results. A similar degree of accuracy of 94.1% and 91.7% on two new CXR datasets from Montgomery County, MD, USA, and India, respectively, demonstrates the robustness of our lung segmentation approach.

/pdf/lung-segmentation-in-chest-radiographs-using-anatomical-1tnky71ryk.pdf

Lung Segmentation in Chest Radiographs Using Anatomical Atlases With Nonrigid Registration

Tuberculosis is a major health threat in many regions of the world. Opportunistic infections in immunocompromised HIV/AIDS patients and multi-drug-resistant bacterial strains have exacerbated the problem, while diagnosing tuberculosis still remains a challenge. When left undiagnosed and thus untreated, mortality rates of patients with tuberculosis are high. Standard diagnostics still rely on methods developed in the last century. They are slow and often unreliable. In an effort to reduce the burden of the disease, this paper presents our automated approach for detecting tuberculosis in conventional posteroanterior chest radiographs. We first extract the lung region using a graph cut segmentation method. For this lung region, we compute a set of texture and shape features, which enable the X-rays to be classified as normal or abnormal using a binary classifier. We measure the performance of our system on two datasets: a set collected by the tuberculosis control program of our local county's health department in the United States, and a set collected by Shenzhen Hospital, China. The proposed computer-aided diagnostic system for TB screening, which is ready for field deployment, achieves a performance that approaches the performance of human experts. We achieve an area under the ROC curve (AUC) of 87% (78.3% accuracy) for the first set, and an AUC of 90% (84% accuracy) for the second set. For the first set, we compare our system performance with the performance of radiologists. When trying not to miss any positive cases, radiologists achieve an accuracy of about 82% on this set, and their false positive rate is about half of our system's rate.

/pdf/automatic-tuberculosis-screening-using-chest-radiographs-4imxna1lw7.pdf

Automatic Tuberculosis Screening Using Chest Radiographs

The introduction of capsule endoscopy (CE) in clinical practice increased the interest for the study of the small-bowel. Consequently, in about 10 years, an impressive quantity of literature on indications, diagnostic yield (DY), safety profile and technical evolution of CE has been published as well as several reviews. At present time, there are 5 small-bowel capsule enteroscopy (SBCE) models in the worldwide market. Head-to-head trials have showed in the great majority of studies comparable results in terms of DY, image quality and completion rate. CE meta-analyses formed the basis of national/international guidelines; these guidelines place CE in a prime position for the diagnostic work-up of patients with obscure gastrointestinal bleeding, known and/or suspected Crohn's disease and possible small-bowel neoplasia. A 2-L polyethylene glycol-based purge, administered the day before the procedure, is the most widely practiced preparation regimen. Whether this regimen can be further improved (i.e., by further decreasing its volume, changing the timing of administration, coupling it with prokinetics and/or other factors) or if it can really affect the DY, is still under discussion. Faecal calprotectin has been used in SBCE studies in two settings: in patients taking non-steroidal anti-inflammatory drugs, to evaluate the type and extent of mucosal damage and, more importantly from a clinical point of view, in patients with known or suspected Crohn's disease for assessment of inflammation activity. Although there is still a lot of debate around the exact reasons of SBCE poor performance in various small-bowel segments, it is worth to remember that the capsule progress is non-steerable, hence more rapid in the proximal than in lower segments of the small-bowel. Capsule aspiration, a relatively unexpected complication, has been reported with increasing frequency. This is probably related with the increase in the mean age of patients undergoing CE. CE video review is a time-consuming procedure. Therefore, several attempts have been made to develop technical software features, in order to make CE video analysis easier and shorter (without jeopardizing its accuracy). Suspected Blood Indicator, QuickView and Fujinon Intelligent Chromo Endoscopy are some of the software tools that have been checked in various clinical studies to date.

Small-bowel capsule endoscopy: A ten-point contemporary review

Over the last decade, wireless capsule endoscopy (WCE) technology has become a very useful tool for diagnosing diseases within the human digestive tract. Physicians using WCE can examine the digestive tract in a minimally invasive way searching for pathological abnormalities such as bleeding, polyps, ulcers, and Crohn's disease. To improve effectiveness of WCE, researchers have developed software methods to automatically detect these diseases at a high rate of success. This paper proposes a novel synergistic methodology for automatically discovering polyps (protrusions) and perforated ulcers in WCE video frames. Finally, results of the methodology are given and statistical comparisons are also presented relevant to other works.

Detection of Small Bowel Polyps and Ulcers in Wireless Capsule Endoscopy Videos

Deep learning algorithms require large amounts of labeled data which is difficult to attain for medical imaging. Even if a particular dataset is accessible, a learned classifier struggles to maintain the same level of performance on a different medical imaging dataset from a new or never-seen data source domain. Utilizing generative adversarial networks in a semi-supervised learning architecture, we address both problems of labeled data scarcity and data domain overfitting. For cardiac abnormality classification in chest X-rays, we demonstrate that an order of magnitude less data is required with semi-supervised learning generative adversarial networks than with conventional supervised learning convolutional neural networks. In addition, we demonstrate its robustness across different datasets for similar classification tasks.

Alexandros Karargyris

Papers

Lung Segmentation in Chest Radiographs Using Anatomical Atlases With Nonrigid Registration

Automatic Tuberculosis Screening Using Chest Radiographs

Small-bowel capsule endoscopy: A ten-point contemporary review

Detection of Small Bowel Polyps and Ulcers in Wireless Capsule Endoscopy Videos

Semi-supervised learning with generative adversarial networks for chest X-ray classification with ability of data domain adaptation