Colonic Polyp Classification using Projection Image and Convolutional Neural Network

Abstract

Now adays, Computer-aided detection (CAD) systems are used to assist radiologists to detect colonic polyps. In this work, we aimed to develop convolutional neural network based classification system for automated detection of polyps. 2D projection images of polyps were used as the input of convolutional neural network. Our classification system performs at 91.89% sensitivity for polyps with 0 false positives per dataset.