Abstract
The manual reading of capsule endoscopy (CE) videos in small bowel disease diagnosis is time-intensive. Algorithms introduced to automate this process are premature for real clinical applications, and multi-diagnosis using these methods has not been sufficiently validated. Therefore, we developed a practical binary classification model, which selectively identifies clinically meaningful images including inflamed mucosa, atypical vascularity or bleeding, and tested it with unseen cases. Four hundred thousand CE images were randomly selected from 84 cases in which 240,000 images were used to train the algorithm to categorize images binarily. The remaining images were utilized for validation and internal testing. The algorithm was externally tested with 256,591 unseen images. The diagnostic accuracy of the trained model applied to the validation set was 98.067%. In contrast, the accuracy of the model when applied to a dataset provided by an independent hospital that did not participate during training was 85.470%. The area under the curve (AUC) was 0.922. Our model showed excellent internal test results, and the misreadings were slightly increased when the model was tested in unseen external cases while the classified ‘insignificant’ images contain ambiguous substances. Once this limitation is solved, the proposed CNN-based binary classification will be a promising candidate for developing clinically-ready computer-aided reading methods.
Highlights
The manual reading of capsule endoscopy (CE) videos in small bowel disease diagnosis is timeintensive
The wireless capsule endoscopy (WCE) is a widely used non-invasive and patient-friendly endoscopic exploration of the entire small bowel with complete video-facilitating detection and monitoring of lesions that was introduced by Iddan et al in 20002
Previous studies have reported the application of the convolutional neural network (CNN), which is the main deep learning algorithm for image analysis, in reading WCE images
Summary
The manual reading of capsule endoscopy (CE) videos in small bowel disease diagnosis is timeintensive. Our model showed excellent internal test results, and the misreadings were slightly increased when the model was tested in unseen external cases while the classified ‘insignificant’ images contain ambiguous substances Once this limitation is solved, the proposed CNN-based binary classification will be a promising candidate for developing clinically-ready computer-aided reading methods. After two decades since its introduction, the way of reading WCE has not changed, which is very time intensive and prone to reader error It is performed manually by expert gastroenterologists who check the entire recorded video, which is approximately 10 to 13 h long with an average reading time of approximately 30 to 40 min[4,5]. The utilization of CNN-based reading algorithms remains on the research stage despite their advancements, which includes their potential detection of various small bowel lesions with high accuracy in unseen images[18,19]. The complete application of this technology has not yet been fully realized despite various efforts such as increasing the number of images for AI learning
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have