Abstract

Computed tomography colonoscopy (CTC) is an emerging alternative to conventional colonoscopy for colorectal cancer screening. A series of computer assisted diagnosis (CAD) techniques have been developed for use in CTC. Although high levels of accuracy for polyp detection have been reported, the problem of excessive false positive detections still warrants attention. We present a CAD-CTC technique that has been developed specifically to reduce the number of false positive detections without compromising polyp detection accuracy. The technique incorporates a novel intermediate stage that restructures initial polyp candidates so that they conform more closely to the shape of actual polyps. The restructuring process causes false positives to expand to include more false positive characteristics, whereas, actual polyps retain their original polyp-like characteristics. An evaluation of the documented technique demonstrated that it can be successfully applied to the majority of polyp candidates, and that its use can reduce the number of false positive detections by up to 57.8%.