Enhanced Computer Assisted Detection of Polyps in CT Colonography

Enhanced Computer Assisted Detection of Polyps in CT Colonography Robert Sadleir Under the supervision of Prof. Paul F. Whelan at Dublin City University Submitted in partial fulfilment of the requirements for the degree of Doctor of Philosophy Dublin City University, 2006 This thesis presents a novel technique for automatically detecting colorectal polyps in CT colonography (CTC). The objective of the documented computer assisted diagnosis (CAD) technique is to deal with the issue of false positive detections without adversely affecting polyp detection sensitivity. The thesis begins with an overview of CTC and a review of the associated research areas, with particular attention given to CAD-CTC. This review identifies excessive false positive detections as a common problem associated with current CAD-CTC techniques. Addressing this problem constitutes the major contribution of this thesis. The documented CADCTC technique is trained with, and evaluated using a series of clinical CTC data sets. These data sets contain polyps with a range of different sizes and morphologies. The results presented in this thesis indicate the validity of the developed CAD-CTC technique and demonstrate its effectiveness in accurately detecting colorectal polyps while significantly reducing the number of false positive detections.