ID: 3519580 A MULTICENTRE RANDOMIZED CONTROLLED TRIAL TO VERIFY THE REDUCIBILITY OF ADENOMA MISS RATE OF COLONOSCOPY ASSISTED WITH ARTIFICIAL INTELLIGENCE BASED SOFTWARE

Background and aims: We have developed the computer-aided detection (CADe) system using original deep learning algorithm based on convolutional neural network for assisting endoscopists to detect colorectal lesions during a colonoscopy. It was trained with 65,421 colonoscopic images (62,726 images with lesions and 2,695 without lesions), which were collected from 4,147 colonoscopy cases including 26,729 lesions. The aim of this study was to...