Computer-aided detection of polyps in CT colonography:: effect of feature-guided polyp segmentation

We evaluated the effect of our novel feature-guided polyp segmentation technique on the reduction of false-positive (FP) colorectal polyp findings generated by our computer-aided diagnosis (CAD) scheme for the detection of colonic polyps in CT colonography (CTC). A total of 72 patients, including 14 patients with 21 colorectal polyp findings, were used in the evaluation. At a 100% (95%) by-patient (by-polyp) sensitivity level, the FP rate was 1.9 FPs (1.9 FPs) per patient for the round-robin analysis. The results indicate that our CAD scheme has the potential of making CTC a viable option for screening of colonic polyps in large patient populations.