ID: 3492451 VALIDATION OF A BINARY CLASSIFICATION MODEL USING A DEEP CONVOLUTIONAL NEURAL NETWORK FOR WIRELESS CAPSULE ENDOSCOPY

Abstract

Manual reading of an entire capsule endoscopy (CE) video is a time-consuming process in the diagnosis of small bowel diseases. Despite introduction of reading models with deep convolutional neural networks (CNN), it is still insufficient for clinical use since most models can only detect a single abnormality or failed to validate with unseen images. Therefore, we developed a binary classification model and tested whether it could suggest images correctly from unseen cases.