Medial axis registration of supine and prone CT colonography data

Computed Tomographic Colonography (CTC) is a minimally invasive method that allows the examination of the colon wall from the source CT sections or in a virtual environment. The primary goal of CTC is the detection of colonic polyps. There are typically two data sets recorded in supine and prone positions. An important step in polyp detection is the anatomical registration of these datasets. We developed and experimentally validated a method to register complementary supine and prone CTC datasets anatomically, using linear stretching/shrinking operations on the media axis colonic path based on the relative path geometries. To compare improvement in spatial registration of supine and prone datasets, a radiologist determined 5 unique reference points (RP) in the registered and unregistered datasets from each of 5 patients by viewing supine and prone data simultaneously. Initial results suggest that our algorithm is capable of registering supine and prone CTC data anatomically within an approximate range of /spl plusmn/13.2 mm, which corresponds to 0.8% error with respect to the average colon length.