Morphological Lesion Detection in CT Colonography

DESCRIPTION (provided by applicant): Colorectal cancer is the second leading cause of cancer death in industrialized nations. Current screening techniques lack sufficient sensitivity or are expensive and require patient sedation. This project aims to develop a new set of tools to assist physicians in detecting cancerous lesions using CT colonography, a new screening technique that is fast and painless for the patient. The tools are based on the concept of a computerized detection algorithm to serve as a backup reader whose primary goal is to make sure that the clinician is alerted to suspicious lesions. The detection algorithm has been designed to be fast, sensitive and specific--capable of detecting polyps that are often missed using present techniques because they are buried on folds of the colon. The tools developed under this program may also assist the clinician in subtyping lesions and will provide a powerful research tool for quantifying lesion growth over time.