Computer Aided Detection for CT Colonography

Clinical Center

Investigators

Linked publications & trials

Abstract

We are improving CT colonography (virtual colonoscopy) by developing computer-assisted diagnosis methods. These methods attempt to identify and characterize colonic polyps automatically, thereby increasing physician accuracy and efficiency and helping patients by finding their polyps. We made a number of advances over the past year, including advances in polyp detection, false positive reduction and classifier optimization. We improved our innovative technique to use the tenia coli (longitudinal muscle bands) as a guide for orienting the supine and prone virtual colonoscopies. This technique has broad implications for research and clinical image interpretation. We are improving methods to co-register virtual colonoscopies and optical colonoscopies of the same patient by modeling the mechanical properties of colonoscopes. We are using novel crowdsourcing methods to conduct large-scale observer performance experiments to help us understand factors that promote improved diagnosis using computer-aided polyp detection. We developed a method to electronically cleanse the colon so that patients can avoid the need for cathartics in the bowel preparation for virtual colonoscopy, potentially leading to a more patient-friendly bowel preparation.