Simbios Talk by Padma Sundaram, Stanford University, March 21, 2006

Title: Finding folds on the colon surface

Radiologists read colon CT images in order to detect polyps, which are considered to be precursors to colon cancer. Computer-aided detection (CAD) of polyps could significantly help reduce reading times for the radiologist. The most common source of false positive (FP) results in CT colonography are the haustral folds. An algorithm capable of detecting colonic folds would therefore be very useful, since it could be used to reduce FPs in CAD of polyps. It could also be used to register prone and supine scans of the same patient.

We present and evaluate an algorithm for extracting haustral folds from colon (CT) images. The input to the algorithm can be an unstructured point cloud or a triangulated mesh representation of the colon surface. The output of our algorithm is a list of likely fold patches on the colon surface.