Automated detection of colorectal lesions in non-cathartic CT colonography

Abstract

Cathartic bowel preparation is a major barrier in implementing clinically feasible colorectal screening programs. We developed a fully automated computer-aided detection (CAD) scheme for non-cathartic computed tomographic colonography (CTC). In a pilot evaluation with an independent testing set of 46 clinical non-cathartic CTC cases, the CAD scheme detected 100% of large ≥10 mm lesions with 3.7 false-positive (FP) detections per patient and 90% of ≥6 mm lesions with 5.0 FP detections per patient, on average. The results indicate that CAD can detect colorectal lesions with high accuracy in non-cathartic CTC. © 2012 Springer-Verlag.