Video Capsule Endoscopy Images Research Articles

ViTCA-Net: a framework for disease detection in video capsule endoscopy images using a vision transformer and convolutional neural network with a specific attention mechanism

ViTCA-Net: a framework for disease detection in video capsule endoscopy images using a vision transformer and convolutional neural network with a specific attention mechanism

Enhanced accuracy for classification of video capsule endoscopy images using multiple deep learning convolutional neural networks

Background and aimsVideo capsule endoscopy (VCE) is widely used in the detection of abnormalities in the small intestine. However, there remains the challenge to correctly identify a limited number of possible abnormal images from tens of thousands of total images, and this impediment has limited the expansion of the technology. More recently artificial intelligence (AI) technology has been used in classifying VCE images from patients, but a clinical-grade diagnostic accuracy (greater than 99%) has not been achieved. MethodsThis study proposes a system for the automatic classification of a number of categories of unbounded VCE images with high accuracy by means of a transfer learning approach using multiple convolutional neural networks (CNNs). With this new approach, it is not necessary to implement image segmentation, so that the feature extraction becomes automatic and the existing models can be fine-tuned to obtain specific classifiers. ResultsOver 16,000 VCE gastrointestinal (GI) images from normal individuals including those with normal clean mucosa, the pylorus, the ileocecal valve, reduced mucosal view due to luminal contents and lymphangiectasia (a normal variant), and patients with five pathological states including angioectasia, bleeding, erosion/s, ulcers and foreign bodies, were obtained from a publicly available data set. These were used in building, testing and validating AI models for evaluating the diagnostic accuracy of our combined 17-CNN deep learning approach. Compared to a single CNN approach used by other research groups, our AI method, using 17 CNNs, achieved an overall diagnostic accuracy of 99.79%, with an accuracy of 100% for identifying bleeding and foreign bodies. The high accuracy was further demonstrated in the confusion matrices, precision, recall, and F1 score. ConclusionsWe have developed accurate AI deep learning models for unbounded VCE image classification of various medical conditions in medical practice.

Deep Learning in Coeliac Disease: A Systematic Review on Novel Diagnostic Approaches to Disease Diagnosis

Background: Coeliac disease affects approximately 1% of the global population with the diagnosis often relying on invasive and time-demanding methods. Deep learning, a powerful tool in medical science, shows potential for non-invasive, accurate coeliac disease diagnosis, though challenges remain. Objective: This systematic review aimed to evaluate the current state of deep-learning applications in coeliac disease diagnosis and identify potential areas for future research that could enhance diagnostic accuracy, sensitivity, and specificity. Methods: A systematic review was conducted using the following databases: PubMed, Embase, Web of Science, and Scopus. PRISMA guidelines were applied. Two independent reviewers identified research articles using deep learning for coeliac disease diagnosis and severity assessment. Only original research articles with performance metrics data were included. The quality of the diagnostic accuracy studies was assessed using the QUADAS-2 tool, categorizing studies based on risk of bias and concerns about applicability. Due to heterogeneity, a narrative synthesis was conducted to describe the applications and efficacy of the deep-learning techniques (DLT) in coeliac disease diagnosis. Results: The initial search across four databases yielded 417 studies with 195 being removed due to duplicity. Finally, eight studies were found to be suitable for inclusion after rigorous evaluation. They were all published between 2017 and 2023 and focused on using DLT for coeliac disease diagnosis or assessing disease severity. Different deep-learning architectures were applied. Accuracy levels ranged from 84% to 95.94% with the GoogLeNet model achieving 100% sensitivity and specificity for video capsule endoscopy images. Conclusions: DLT hold substantial potential in coeliac disease diagnosis. They offer improved accuracy and the prospect of mitigating clinician bias. However, key challenges persist, notably the requirement for more extensive and diverse datasets, especially to detect milder forms of coeliac disease. These methods are in their nascent stages, underscoring the need of integrating multiple data sources to achieve comprehensive coeliac disease diagnosis.

Semi-supervised Learning for Segmentation of Bleeding Regions in Video Capsule Endoscopy

In the realm of modern diagnostic technology, video capsule endoscopy (VCE) is a standout for its high efficacy and non-invasive nature in diagnosing various gastrointestinal (GI) conditions, including obscure bleeding. Importantly, for the successful diagnosis and treatment of these conditions, accurate recognition of bleeding regions in VCE images is crucial. While deep learning-based methods have emerged as powerful tools for the automated analysis of VCE images, they often demand large training datasets with comprehensive annotations. Acquiring these labeled datasets tends to be time-consuming, costly, and requires significant domain expertise. To mitigate this issue, we have embraced a semi-supervised learning (SSL) approach for the bleeding regions segmentation within VCE. By adopting the ‘Mean Teacher’ method, we construct a student U-Net equipped with an scSE attention block, alongside a teacher model of the same architecture. These models’ parameters are alternately updated throughout the training process. We use the Kvasir-Capsule dataset for our experiments, which encompasses various GI bleeding conditions. Notably, we develop the segmentation annotations for this dataset ourselves. The findings from our experiments endorse the efficacy of the SSL-based segmentation strategy, demonstrating its capacity to reduce reliance on large volumes of annotations for model training, without compromising on the accuracy of identification.

Clean mucosal area detection of gastroenterologists versus artificial intelligence in small bowel capsule endoscopy.

Studies comparing the detection of clean mucosal areas in capsule endoscopy (CE) using human judgment versus artificial intelligence (AI) are rare. This study statistically analyzed gastroenterologist judgments and AI results. Three hundred CE video clips (100 patients) were prepared. Five gastroenterologists classified the video clips into 3 groups (≥75% [high], 50%-75% [middle], and < 50% [low]) according to their subjective judgment of cleanliness. Visualization scores were calculated using an AI algorithm based on the predicted visible area, and the 5 gastroenterologists' judgments and AI results were compared. The 5 gastroenterologists evaluated CE clip video quality as "high" in 10.7% to 36.7% and as "low" in 28.7% to 60.3% and 29.7% of cases, respectively. The AI evaluated CE clip video quality as "high" in 27.7% and as "low" in 29.7% of cases. Repeated-measures analysis of variance (ANOVA) revealed significant differences in the 6 evaluation indicators (5 gastroenterologists and 1 AI) (P < .001). Among the 300 judgments, 90 (30%) were consistent with 5 gastroenterologists' judgments, and 82 (91.1%) agreed with the AI judgments. The "high" and "low" judgments of the gastroenterologists and AI agreed in 95.0% and 94.9% of cases, respectively. Bonferroni's multiple comparison test showed no significant difference between 3 gastroenterologists and AI (P = .0961, P = 1.0000, and P = .0676, respectively) but a significant difference between the other 2 with AI (P < .0001). When evaluating CE images for cleanliness, the judgments of 5 gastroenterologists were relatively diverse. The AI produced a relatively universal judgment that was consistent with the gastroenterologists' judgements.

DOP75 Video capsule endoscopy may detect active small bowel Crohn’s disease requiring change of treatment despite unremarkable ileo-colonic anastomosis

Abstract Background Previous studies have shown benefit in performing video capsule endoscopy (VCE) in Crohn’s disease (CD) diagnosis, especially in cases of an unremarkable ileo-colonoscopy, however data on the role of VCE in postsurgical follow-up of CD patients is lacking. We aimed to evaluate CD endoscopic activity in the small bowel (SB) of patients presenting with disease remission at the ileo-colonic anastomosis. Methods We retrospectively obtained data on patients with CD older than 18 years of age, who underwent ileo-colonoscopy between the dates 1/10/2012 through 31/09/2022. Patients in whom VCE was performed within 8 months of ileo-colonoscopy were eligible for analysis. Lewis and Rutgeerts scores were calculated based on VCE images and endoscopy reports, respectively. Disease remission was deemed as Rutgeerts score of i0-i1 at the ileo-colonic anastomosis, and Lewis score (LS) of&amp;lt;135 on VCE. Results A total of 1,334 ileo-colonoscopies were performed in CD patients with ileo-colonic anastomoses during the study period, of whom 31 patients (male gender 58.06%; mean age 41.77±14.19 years) were referred to VCE within 8 months of ileo-colonoscopy, (median 124.81 days). Twenty patients exhibited endoscopic remission at the anastomosis, 3 of them failed passage of a patency capsule and 17 successfully completed VCE. Only 5 patients (29.4%) had a normal LS, while 12 (70.6%) exhibited an active SB disease (LS≥135, mean 791.3±1254.9). Patients with a discrepancy between the two scores, were characterized by a shorter disease duration (P=0.011) and showed a trend towards a shorter time since surgery (P=0.06). No correlation was found between SB disease activity and concurrent medical therapy (P=0.898). Treatment modification was applied in 7 patients (41.2%) and was associated with a higher LS (biserial correlation coefficient 0.51, P=0.037). Conclusion Endoscopic remission at the ileo-colonic anastomosis is not an accurate indicator for CD remission in the SB, especially in a new-onset disease and soon after the surgery. VCE has an added value to ileo-colonoscopy in postsurgical follow-up and treatment adjustment in patients with CD.

S1559 Reduction of Video Capsule Endoscopy Reading Times Using Artificial Intelligence

Introduction: Video capsule endoscopy (VCE) is an innovation that has revolutionized care within the field of gastroenterology, but the time needed to read the studies generated has often been cited as an area for improvement. The purpose of this study is to create a machine learning model capable of significantly reducing capsule endoscopy reading times. Methods: In this study, we have trained a convolutional neural network, ResNet50, using videos from the KID dataset to confidently exclude normal images on VCE, while retaining abnormal ones. The CNN was exposed to 3 full length capsule endoscopy videos A threefold cross-validation scheme was employed whereby the model was trained using 2 of the aforementioned videos, tested on the third and this was repeated for all possible video combinations. We trained our model for 9 epochs and further improved the predictions of our model by adding rotated versions of abnormal segments to our training data. Results: Our study identified abnormal frames in 3 KID videos as outlined in the Table. We were able to reduce the video length by 47%, on average, and captured frames from 118 of the 119 abnormal segments labeled by the expert physician. We were able to develop an algorithm that successfully detected 99% of abnormal segments while reducing the reading time for a physician by over 43%. Conclusion: Our model demonstrated high levels of accuracy with significant reduction in physician reading time. Our results are reassuring and demonstrate the benefit of CNN in processing VCE images. We believe our study lays an excellent foundation for further validation in large multicenter trials. Table 1. - Description of the various characteristics of the of the 3 KID videos in addition to the reduced number of frames produced and abnormal segments detected by the ResNet50 model Video Name Total Number of Frames Number of Abnormal Segments Reduced Number of Frames by ResNet50 Number of Abnormal Segments Detected by ResNet50 KID Video 1 28480 (96 min) 22 14828 (49 min) 22 KID Video 2 117565 (391 min) 86 84672 (282 min) 86 KID Video 3 74762 (249 min) 11 27099 (90 min) 11

Automated detection of celiac disease using Machine Learning Algorithms

Celiac disease is a disorder of the immune system that mainly affects the small intestine but can also affect the skeletal system. The diagnosis relies on histological assessment of duodenal biopsies acquired by upper digestive endoscopy. Immunological tests involve collecting a blood sample to detect if the antibodies have been produced in the body. Endoscopy is invasive and histology is time-consuming. In recent years there have been various algorithms that use artificial intelligence (AI) and neural convolutions (CNN, Convolutional Neural Network) to process images from capsule endoscopy, a non-invasive endoscopy approach, that provides magnified, high qualitative images of the small bowel mucosa, to quickly establish a diagnosis. The proposed innovative approach do not use complex learning algorithms, instead it find some artefacts in the endoscopies using kernels and use classified machine learning algorithms. Each used artefacts have a psychical meaning: atrophies of the mucosa with a visible submucosal vascular pattern; the presence of cracks (depressions) that have an appearance similar to that of dry land; reduction or complete loss of folds in the duodenum; the presence of a submerged appearance at the Kerckring folds and a low number of villi. The results obtained for video capsule endoscopy images processing reveal an accuracy of 94.1% and F1 score of 94%, which is competitive with other complex algorithms. The main goal of the present research was to demonstrate that computer-aided diagnosis of celiac disease is possible even without the use of very complex algorithms, which require expensive hardware and a lot of processing time. The use of the proposed automated images processing acquired noninvasively by capsule endoscopy would be assistive in detecting the subtle presence of villous atrophy not evident by visual inspection. It may also be useful to assess the degree of improvement of celiac. Patients on a gluten-free diet, the main treatment method for stopping the autoimmune process and improving the state of the small intestinal villi. The novelty of the work is that the algorithm uses two modified filters to properly analyse the intestine wall texture. It is proved that using the right filters, the proper diagnostic can be obtained by image processing, without the use of a complicated machine learning algorithm.

Prelimenary results of red lesion segmentation in WCE images

Wireless capsule endoscopy (WCE) is a novel imaging technique that can view the entire small bowel in human body. Thus, it is presented as an excellent diagnostic tool for evaluation of gastrointestinal diseases compared with traditional endoscopies. However, the diagnosis by the physicians is tedious since it requires reviewing the video extracted from the capsule and analysing all of its frames. This tedious task has encouraged the researchers to provide automated diagnostic technics for WCE frameworks to detect symptoms of gastrointestinal illness. In this paper, we present the prelimenary results of red lesion detection in WCE images using Dense-Unet deep learning segmentation model. To this end, we have used a dataset containing two subsets of anonymized video capsule endoscopy images with annotated red lesions. The first set, used in this work, has 3,295 non-sequential frames and their corresponding annotated masks. The results obtained by the proposed scheme are promising.

S1311 CNN Analysis of Video Capsule Endoscopy Frames Reduces Miss Rate of Relevant Lesions

INTRODUCTION: Human reads of small bowel video capsule endoscopy (VCE) have a surprisingly high miss-rate for known lesions.1 We previously reported development of a convolutional neural networks (CNN) with very high sensitivity and specificity for detection of lesions on VCE images.2 We aimed to determine if this CNN can detect relevant lesions missed by human readers of VCE. METHODS: A total of 146 VCE studies (not previously exposed to CNN) were evaluated for this retrospective study. Each video was de-identified and analyzed by the CNN to select frames predicted to contain abnormalities. The frames saved by the original reader (‘OR’) were verified to include examples of all relevant lesions mentioned in the de-identified report. Abnormalities identified in ‘OR’ frames and CNN frames were reviewed and classified by 3 experts as clinically relevant, possibly relevant, or innocent. Each abnormality found by ‘OR’ and CNN were aligned by timestamp to determine if found by CNN only, ‘OR’ only, or by both CNN and reader. RESULTS: From these 146 VCEs, there were a total of 542 abnormal findings (17% were relevant, 47% were possibly relevant and 36% were innocent (Table 1). Of the definitely relevant findings, 68% were found by both the ‘OR’ and the CNN, 1% was found by only the ‘OR’ (missed by CNN), and 31% were found by only the CNN (missed by ‘OR’). Of the possibly relevant findings 36% were found by both, 5% were found by only the ‘OR’, and 36% were found by only the CNN. Of the innocent findings 55% were found by both, 19% were found by only the ‘OR’, and 58% were found by only the CNN. Broken down by lesion type, miss rates of definitely relevant lesions by ‘OR’ vs CNN are shown in Table 2. Of the 94,000 mean number of frames analyzed per video, the CNN generated a mean of 200 frames with abnormalities, of which, a mean of 31 were confirmed to have abnormalities (Table 3). CONCLUSION: This retrospective comparison of lesion detection by original reader vs CNN in 146 VCE studies suggests that use of CNN may significantly reduce miss rate of clinically relevant lesions, especially AVMs, bulges, and polypoid lesions.Table 1.: Relevance of abnormalities found reader vs CNNTable 2.: Miss rates among "Definitely Relevant" lesionsTable 3.: Total frames generated by VCE and CNN compared to frames with abnormalities

Celiac disease diagnosis from videocapsule endoscopy images with residual learning and deep feature extraction

Background and ObjectiveVideocapsule endoscopy (VCE) is a relatively new technique for evaluating the presence of villous atrophy in celiac disease patients. The diagnostic analysis of video frames is currently time-consuming and tedious. Recently, computer-aided diagnosis (CAD) systems have become an attractive research area for diagnosing celiac disease. However, the images captured from VCE are susceptible to alterations in light illumination, rotation direction, and intestinal secretions. Moreover, textural features of the mucosal villi obtained by VCE are difficult to characterize and extract. This work aims to find a novel deep learning feature learning module to assist in the diagnosis of celiac disease. MethodsIn this manuscript, we propose a novel deep learning recalibration module which shows significant gain in diagnosing celiac disease. In this recalibration module, the block-wise recalibration component is newly employed to capture the most salient feature in the local channel feature map. This learning module was embedded into ResNet50, Inception-v3 to diagnose celiac disease using a 10-time 10-fold cross-validation based upon analysis of VCE images. In addition, we employed model weights to extract feature points from training and test samples before the last fully connected layer, and then input to a support vector machine (SVM), k-nearest neighbor (KNN), and linear discriminant analysis (LDA) for differentiating celiac disease images from heathy controls. ResultsOverall, the accuracy, sensitivity and specificity of the 10-time 10-fold cross-validation were 95.94%, 97.20% and 95.63%, respectively. ConclusionsA novel deep learning recalibration module, with global response and local salient factors is proposed, and it has a high potential for utilizing deep learning networks to diagnose celiac disease using VCE images.

Color masking improves classification of celiac disease in videocapsule endoscopy images

BackgroundVideocapsule endoscopy images are useful to detect pathologic alterations, including villous atrophy, in the small intestinal mucosa, which is helpful for diagnosing celiac disease. In prior work, quantitative videocapsule analysis was found useful to classify celiac versus control images. However, the effect of dark/extraneous substances on classification efficacy requires remediation. MethodFor quantitative analysis, data from the Medtronic SB2 and SB3 systems were pooled. Videocapsule images of the distal duodenum/proximal jejunum were acquired from 13 celiac and 13 control patients. Dark regions, extraneous fluids, and air bubbles were mostly removed by utilizing color masking. Two different red-green-blue (RGB) color masks were constructed from 20 to 30 reference pixels obtained from mucosal and from extraneous regions. Each image pixel was accepted or rejected for subsequent analysis based on whether its distance was closest to a mucosal or to an extraneous reference in RGB space. Four images were then randomly selected from each videoclip for processing (52 images from each group). After masking, celiac versus control images were plotted in a three-space consisting of mean and standard deviation in pixel brightness, and surface area remaining after masking. A linear discriminant function was used for classification. The paradigm was repeated with a second random data set for validation. ResultsMasking improved classification of celiac versus control images to nearly 80% accuracy as compared to 70–77% without masking. Celiac disease patients tended to have lesser mean pixel brightness and greater variability in brightness, in accord with prior work, and more masking was needed to remove extraneous features. ConclusionsColor masking is useful to remove dim/extraneous features from videocapsule images and it results in improved classification/assessment to distinguish celiac with villous atrophy from control videocapsule image. This can be helpful to detect and map regions of pathology, to screen for celiac disease, and to determine the efficacy of a gluten free diet.

Differential Diagnosis of Jejunum and Ileum Tumours Based on Video Capsule Endoscopy Data Using Mathematical Analysis

Aim. The aim of this study was to develop an algorithm for identifying various types of jejunum and ileum tumour lesions based on video capsule endoscopy (VCE), which can be used by physicians in the process of decision making. Materials and methods. In the study, we analysed data on the examination and treatment of 65 patients (35 men and 30 women aged 18–80 years (mean age 46 ± 28 years)), who underwent VCE in the City Clinical Hospital No. 31 and in the Clinica K+31 during the period October 2008 — April 2017. The indications for VCE were a search for a reason of gastrointestinal bleeding, the anaemia of unknown etiology and suspected tumour of the small intestine. According to the VCE results, 181 cases of various changes in the jejunum and ileum were revealed. Each tumour object had been histologically verified before our study. Capsule endoscopy was performed using equipment produced by Olympus (Japan), MicroCam Intromedic (Korea), PillCam Given Imaging (Israel), OMOM Chongqing Jinshan Science &amp; Technology (China).Results. Following expert interviews, 30 signs and their gradations were identified for assessing the type of lesion in the jejunum and ileum using video capsule endoscopy images. Among them, 8 were found to be statistically significant (affecting the division of objects into groups): patient gender, gut wall/lumen deformation, path of intestinal folds, polypoid changes, vascular pattern, mucosal regularity, mucosal lobulation and colour. Using a Bayesian heterogeneous diagnostic procedure and the calculation of diagnostic factors, a three-level algorithm has been developed for the differential diagnosis of jejunum and ileum lesions.Conclusions. The application of the proposed algorithm in clinical practice will not only allow the presence or absence of the jejunum or ileum tumour lesion to be verified, but also the type of this lesion to be determined with an accuracy of more than 86%. The developed diagnostic algorithm can support decision making by the clinician within the task of differentiating jejunum or ileum tumour lesions into three main types: benign epithelial tumours, benign non-epithelial tumours and malignant tumours. Differential diagnosis of the type of jejunum or ileum tumour lesion using the proposed diagnostic algorithm facilitates not only the development of a treatment tactic for managing such patients (dynamic observation, conservative therapy, operative treatment), but also the determination of terms (emergency, urgent, planned) and methods (endometrial luminal, laparoscopic, laparotomic) of surgical treatment.

Automated diagnosis of celiac disease using DWT and nonlinear features with video capsule endoscopy images

Celiac disease is a common immune response when gluten is ingested. Over time, this response will impair the lining of the small intestine and result in malabsorption. This could bring about critical health complications. However, the symptoms of celiac disease vary and hence, it is relatively challenging to make an accurate diagnosis. This results in a high percentage of misdiagnoses. Therefore, a computer-aided detection (CAD) system is proposed to overcome the challenges. Hence, this study employed the discrete wavelet transform (DWT) to decompose the video images, after which textural and nonlinear features were extracted. Thereafter, the particle swarm optimization (PSO) was performed to choose 30 optimal features for classification. An accuracy level of 86.47%, and sensitivity and specificity of 88.43% and 84.60%, respectively, was achieved with the 10-fold cross-validation strategy. Moreover, an accuracy of 85.91% was attained with the leave-one-out cross-validation (LOOCV) technique. This methodology demonstrates potential for accurately identifying celiac disease. It can therefore be noted that the developed CAD system may improve the diagnostic performance in the detection of celiac disease, and thus reduce the number of misdiagnoses.

Coeliac disease and the videocapsule: what have we learned till now.

Celiac disease is diagnosed in part by finding areas of pathology in the small bowel (SB) mucosa. This can often be difficult because the pathologic alterations, including atrophy of the small intestinal villi, can often be sparse and subtle. Some of the quantitative methods for detecting and measuring the presence of villous atrophy from videocapsule endoscopy (VCE) images are presented and discussed. These methods consist of static features of measurement including texture, gray level, and presence and abundance of fissures contained within each acquired image. The methods also consist of dynamic measurements including spectral analysis, and determining motion from a sequence of endoscopic images as obtained from a VCE clip. Thus far, several methods have been found useful to characterize the SB mucosa of untreated celiac disease patients versus control patients lacking villous atrophy, which have revealed significant differences in texture, frequency, and motion on analysis of VCE. In untreated celiac patients undergoing endoscopy, there tends to be greater magnitude of changes and spatial differences in textural descriptors, longer periodic components, indicating slower periodic activity, and differences in feature location, suggesting alterations in motility at areas of pathology as compared to patients without villous atrophy. Improvements in the quantitative analysis of VCE imaging in celiac patients is important to detect pathology in suspected patients, so that biopsies can be obtained from pertinent regions of the small intestinal mucosa. Improvements are also necessary so that patients with celiac disease can be monitored to evaluate the progress of mucosal healing after onset of treatment.

Evaluation of performance of the Omni mode for detecting video capsule endoscopy images: A multicenter randomized controlled trial.

Background and study aims: Olympus recently developed a new algorithm called Omni mode that discards redundant video capsule endoscopy (VCE) images. The current study aimed to demonstrate the non-inferiority of the Omni mode in terms of true positives (TPs) and the superiority of the Omni mode with regard to reading time against a control (ordinary ES-10 system). Patients and methods: This multicenter prospective study included 40 patients with various small bowel diseases. VCE images were evaluated by 7 readers and 3 judging committee members. Two randomly allocated readers assessed the VCE images obtained using the 2 modalities for each patient. The order of the modalities was switched between the 2 readers and the interval between readings by the same reader was 2 weeks. The judging committee predefined clinically relevant lesions as major lesions and irrelevant lesions as minor lesions. The number of TPs for major and minor lesions and the reading times were compared between the modalities. The predefined non-inferiority margin for the TP ratio of the Omni mode compared with the control was 0.9. Results: The estimated TP ratios and 95 % confidence intervals for total, major, and minor lesions were 0.87 (0.80 – 0.95), 0.93 (0.83 – 1.04), and 0.83 (0.74 – 0.94), respectively. Although non-inferiority was not demonstrated, the rate of detection of major lesions was not significantly different between the modalities. The reading time was significantly lower when using the Omni mode than when using the control. Conclusions: The Omni mode may be only appropriate for the assessment of major lesions.

Recommendations to quantify villous atrophy in video capsule endoscopy images of celiac disease patients.

AIMTo quantify the presence of villous atrophy in endoscopic images for improved automation.METHODSThere are two main categories of quantitative descriptors helpful to detect villous atrophy: (1) Statistical and (2) Syntactic. Statistical descriptors measure the small intestinal substrate in endoscope-acquired images based on mathematical methods. Texture is the most commonly used statistical descriptor to quantify villous atrophy. Syntactic descriptors comprise a syntax, or set of rules, for analyzing and parsing the substrate into a set of objects with boundaries. The syntax is designed to identify and distinguish three-dimensional structures based on their shape.RESULTSThe variance texture statistical descriptor is useful to describe the average variability in image gray level representing villous atrophy, but does not determine the range in variability and the spatial relationships between regions. Improved textural descriptors will incorporate these factors, so that areas with variability gradients and regions that are orientation dependent can be distinguished. The protrusion syntactic descriptor is useful to detect three-dimensional architectural components, but is limited to identifying objects of a certain shape. Improvement in this descriptor will require incorporating flexibility to the prototypical template, so that protrusions of any shape can be detected, measured, and distinguished.CONCLUSIONImproved quantitative descriptors of villous atrophy are being developed, which will be useful in detecting subtle, varying patterns of villous atrophy in the small intestinal mucosa of suspected and known celiac disease patients.

Pondering the potential of quantitative analysis of video capsule endoscopy images in the management of children with celiac disease

Celiac disease is common throughout the world, affecting approximately 1% of patients of all age groups. In this review, the role of video capsule endoscopy in characterizing the small intestinal lumen of pediatric patients with celiac disease is discussed in detail. Quantitative aspects of video capsule endoscopy which may be helpful in diagnosing pediatric patients are highlighted.

A re-review of caspule endoscopies of patients referred for deep enteroscopy changes their management

BackgroundPatients are commonly referred to tertiary centers for deep enteroscopy because of abnormal findings on video capsule endoscopy (VCE). The aim of this study was to determine how often clinical management changes when VCEs are reviewed by an enteroscopist prior to scheduling a procedure. MethodsA retrospective review was performed of patients referred for deep enteroscopy because of abnormal capsule endoscopy. All VCE images were reviewed prospectively by the tertiary center's enteroscopist. Patients were then scheduled for deep enteroscopy or other management based on the capsule review. The rate of disagreement in the capsule findings, changes in management, and the diagnostic and therapeutic yield of enteroscopy were calculated. ResultsVideo capsule endoscopy was available in 45 patients who were referred for deep enteroscopy. The mean age was 61 years (51% were females). Indications included obscure GI bleeding (37 patients), abnormal imaging (3 patients), abdominal pain (2 patients), Peutz-Jegher syndrome (2 patients), and weight loss (1 patient). Referring physician findings included polyps or masses (13 patients), angioectasia (13 patients), ulcers (9 patients), active bleeding (9 patients), nonspecific findings (8 patients), and normal (2 patients). A capsule review led to disagreement of the findings of 13 (29%) patients and led to a change in the management of 9 (20%) patients. The most common reason for a change in management was overcalled lesions. Thirty-seven patients underwent enteroscopy with a diagnostic yield of 48.8% and therapeutic intervention in 24.4%. ConclusionA review of referral VCE studies led to a change in management in a large percentage of patients, particularly when the indication was polyp, mass, or ulcer. Patients referred for deep enteroscopy should have their capsule re-read by an enteroscopist prior to scheduling the procedure.

Methods to quantitate videocapsule endoscopy images in celiac disease.

In this work, bioengineering methods that can be used to quantitatively analyze videocapsule endoscopy images that have been acquired from celiac patients versus controls are described. For videocapsule endoscopic analysis, each patient swallows a capsule which contains an imaging device and light source. In celiac and control patients, images are acquired and analyzed at the level of the small intestine. The data used for videocapsule analysis consisted of high resolution images of dimension 576 × 576 pixels, acquired twice per second. The goal of the quantitative analysis is to detect abnormality in celiac patient images as compared with controls. Several types of abnormality can exist at the level of the small intestine in celiac patients. In untreated patients, and often even after treatment with a gluten-free diet, there can be villous atrophy, as well as presence of fissures and a mottled appearance. To detect and discern these abnormalities, several methods of statistical and structural feature extraction and selection are described. It was found that there is a significantly greater variation in image texture and average brightness level in celiac patients as compared with controls (p < 0.05). Celiac patients have a longer dominant period as compared with controls, averaging 6.4 ± 2.6 seconds versus 4.7 ± 1.6 seconds in controls (p = 0.001). This suggests that overall motility is slower in the celiac patients. Furthermore, the mean number of villous protrusions per image was found to be 402.2 ± 15.0 in celiac patients versus 420.8 ± 24.0 in control patients (p < 0.001). The average protrusion width was 14.66 ± 1.04 pixels in celiacs versus 13.91 ± 1.47 pixels in controls (p = 0.01). The mean protrusion height was 3.10 ± 0.26 grayscale levels for celiacs versus 2.70 ± 0.43 grayscale levels for controls (p < 0.001). Thus celiac patients tended to have fewer protrusions, and these were more varied in shape, tending to be blunted, as compared with controls, which more often had fine, uniform protrusions. A variety of computerized methods are now available to quantitate videocapsule images for comparison of celiac versus control patients. Since these methods are based on computer algorithms, they can be automated and there is no variation in the results due to observer bias. These methods readily lend themselves to automation, so that it may be possible to map the entire small intestine for presence of abnormality in real-time. It is also possible to develop an automated, quantitative clinical score which can be displayed with real-time update during the procedure. This would be useful to determine progress in celiac patients on a gluten-free diet, and to better understand the properties of the healing process in these patients.