A Complete System for Polyps Flagging in Virtual Colonoscopy

Computer tomographic colonography, combined with computer-aided detection, is a promising emerging technique for colonic polyp analysis. We present a complete pipeline for polyp flagging based on a simple segmentation technique that enhances polyps, a multi-scale candidate polyp delineation, and new texture and geometric features that consider both the information in the candidate polyp location and its immediate surrounding area. The proposed algorithm is tested with ground truth data, including flat and small polyps which are hard to detect even with optical colonoscopy, obtaining very promising results. For polyps greater than 6mm in size we achieve 100% sensitivity with just 0.8 false positives per study, and for polyps greater than 3mm in size we achieve 100% sensitivity with 2.2 false positives per study.