To evaluate how computer aided diagnosis can support radiologists in the detection of polyps in CT colonography and to illustrate imaging techniques employed in the software we present.

Computed tomographic (CT) colonography has been recognized as a candidate for population screening programs, provided that computer aided diagnosis (CAD)
schemes could reduce the interpretation time. We are developping a CAD scheme which examines the volumetric properties of the CT scan. Among these properties, we present a preliminar (in progress) study on
second-order statistics. CT colonography was performed using helican scanning with patients in supine and prone positions, after bowel cleasing and air insufflation. CT data sets were retrospectively collected from heterogeneous sources. The colon internal surface was extracted from volumetric data sets [1]. The analysis of second order statistics was carried out on clusters of points of the segmented surface. These clusters were generated from spatial density rules. This study, based on co-occurrence matrix features, could provide a useful tool to improve the discriminant power of the CAD scheme.
[1] A. Bert, to be submtted