Prediction models for recurrence in patients with small bowel bleeding

Obscure gastrointestinal bleeding(OGIB)has traditionally been defined as gastrointestinal bleeding whose source remains unidentified after bidirectional endoscopy.OGIB can present as overt bleeding or occult bleeding,and small bowel lesions are the most common causes.The small bowel can be evaluated using capsule endoscopy,device-assisted enteroscopy,computed tomography enterography,or magnetic resonance enterography.Once the cause of smallbowel bleeding is identified and targeted therapeutic intervention is completed,the patient can be managed with routine visits.However,diagnostic tests may produce negative results,and some patients with small bowel bleeding,regardless of diagnostic findings,may experience rebleeding.Predicting those at risk of rebleeding can help clinicians form individualized surveillance plans.Several studies have identified different factors associated with rebleeding,and a limited number of studies have attempted to create prediction models for recurrence.This article describes prediction models developed so far for identifying patients with OGIB who are at greater risk of rebleeding.These models may aid clinicians in forming tailored patient management and surveillance.

Video capsule endoscopy vs double-balloon enteroscopy in the diagnosis of small bowel bleeding:A systematic review and metaanalysis

AIM To compare the diagnostic accuracy of video capsule endoscopy(VCE) and double-balloon enteroscopy(DBE) in cases of obscure gastrointestinal bleeding(OGIB) of vascular origin.METHODS MEDLINE(via PubMed), LILACS(via BVS) and Cochrane/CENTRAL virtual databases were searched for studies dated before 2017. We identified prospective and retrospective studies, including observational, cohort, single-blinded and multicenter studies, comparing VCE and DBE for the diagnosis of OGIB, and data of all the vascular sources of bleeding were collected. All patients were subjected to the same gold standard method. Relevant data were then extracted from each included study using a standardized extraction form. We calculated study variables(sensitivity, specificity, prevalence, positive and negative predictive values and accuracy) and performed a meta-analysis using Meta-Disc software.RESULTS In the per-patient analysis, 17 studies(1477 lesions) were included. We identified3150 exams(1722 VCE and 1428 DBE) in 2043 patients and identified 2248 sources of bleeding, 1467 of which were from vascular lesions. Of these lesions, 864(58.5%) were diagnosed by VCE, and 613(41.5%) were diagnosed by DBE. The pretest probability for bleeding of vascular origin was 54.34%. The sensitivity of DBE was 84%(95%CI: 0.82-0.86; heterogeneity: 78.00%), and the specificity was92%(95%CI: 0.89-0.94; heterogeneity: 92.0%). For DBE, the positive likelihood ratio was 11.29(95%CI: 4.83-26.40; heterogeneity: 91.6%), and the negative likelihood ratio was 0.20(95%CI: 0.15-0.27; heterogeneity: 67.3%). Performing DBE after CE increased the diagnostic yield of vascular lesion by 7%, from 83% to90%.CONCLUSION The diagnostic accuracy of detecting small bowel bleeding from a vascular source is increased with the use of an isolated video capsule endoscope compared with isolated DBE. However, concomitant use increases the detection rate of the bleeding source.

Outpatient management of obscure gastrointestinal bleeding:A new perspective in high-risk patients

Mid-gastrointestinal bleeding accounts for approximately 5%-10%of all gastrointestinal bleeding cases,and vascular lesions represent the most frequent cause.The rebleeding rate for these lesions is quite high(about 42%).We hereby recommend that scheduled outpatient management of these patients could reduce the risk of rebleeding episodes.

Improving CAD Hemorrhage Detection in Capsule Endoscopy

This study explores an automated framework to assist the recognition of hemorrhage traces and bleeding lesions in video streams of small bowel capsule endoscopy (SBCE). The proposed methodology aims to achieve fast image control (<10 minutes), save valuable time of the physicians, and enable high performance diagnosis. A specialized elimination algorithm excludes all identical consecutive frames by utilizing the difference of gray levels in pixel luminance. An image filtering algorithm is proposed based on an experimentally calculated bleeding index and blood-color chart, which inspects all remaining frames of the footage and identifies pixels that reflect active or potential hemorrhage in color. The bleeding index and blood-color chart are estimated of the chromatic thresholds in RGB and HSV color spaces, and have been extracted after experimenting with more than 3200 training images, derived from 99 videos of a pool of 138 patients. The dataset has been provided by a team of expert gastroenterologist surgeons, who have also evaluated the results. The proposed algorithms are tested on a set of more than 1000 selected frame samples from the entire 39 testing videos, to a prevalence of 50% pathologic frames (balanced dataset). The frame elimination of identical and consecutive frames achieved a reduction of 36% of total frames. The best statistical performance for diagnosis of positive pathological frames from a video stream is achieved by utilizing masks in the HSV color model, with sensitivity up to 99%, precision 94.41% to a prevalence of 50%, accuracy up to 96.1%, FNR 1%, FPR 6.8%. The estimated blood-color chart will be clinically validated and used in feature extraction schemes supporting machine learning ML algorithms to improve the localization potential.

Detection of Angioectasias and Haemorrhages Incorporated into a Multi-Class Classification Tool for the GI Tract Anomalies by Using Binary CNNs

The proposed deep learning algorithm will be integrated as a binary classifier under the umbrella of a multi-class classification tool to facilitate the automated detection of non-healthy deformities, anatomical landmarks, pathological findings, other anomalies and normal cases, by examining medical endoscopic images of GI tract. Each binary classifier is trained to detect one specific non-healthy condition. The algorithm analyzed in the present work expands the ability of detection of this tool by classifying GI tract image snapshots into two classes, depicting haemorrhage and non-haemorrhage state. The proposed algorithm is the result of the collaboration between interdisciplinary specialists on AI and Data Analysis, Computer Vision, Gastroenterologists of four University Gastroenterology Departments of Greek Medical Schools. The data used are 195 videos (177 from non-healthy cases and 18 from healthy cases) videos captured from the PillCam(R) Medronics device, originated from 195 patients, all diagnosed with different forms of angioectasia, haemorrhages and other diseases from different sites of the gastrointestinal (GI), mainly including difficult cases of diagnosis. Our AI algorithm is based on convolutional neural network (CNN) trained on annotated images at image level, using a semantic tag indicating whether the image contains angioectasia and haemorrhage traces or not. At least 22 CNN architectures were created and evaluated some of which pre-trained applying transfer learning on ImageNet data. All the CNN variations were introduced, trained to a prevalence dataset of 50%, and evaluated of unseen data. On test data, the best results were obtained from our CNN architectures which do not utilize backbone of transfer learning. Across a balanced dataset from no-healthy images and healthy images from 39 videos from different patients, identified correct diagnosis with sensitivity 90%, specificity 92%, precision 91.8%, FPR 8%, FNR 10%. Besides, we compared the performance of our best CNN algorithm versus our same goal algorithm based on HSV colorimetric lesions features extracted of pixel-level annotations, both algorithms trained and tested on the same data. It is evaluated that the CNN trained on image level annotated images, is 9% less sensitive, achieves 2.6% less precision, 1.2% less FPR, and 7% less FNR, than that based on HSV filters, extracted from on pixel-level annotated training data.