Efficacy of automatic path tracking in virtual colonoscopy

Abstract

Virtual colonoscopy is a minimally invasive technique allowing early detection of colorectal polyps. Several visualization techniques exist to inspect this large amount of data on the presence of lesions. The most common ones are slice-by-slice viewing, multiplanar reformatting and perspective volume rendering of the inner colon wall (virtual endoscopy). A path or centerline through the colon can be very useful to perform virtual endoscopy. Frequently, this path has to be tracked manually. Unfortunately, manual path tracking is a very time-consuming task and the resulting path depends a lot on the experience of the operator. This severely limits the applicability of the path-based visualization and inspection methods. We have developed an automatic path tracker for virtual endoscopy based on the minimal cost path algorithm described in Deschamps and Cohen [1]. In this paper, we briefly summarize the minimal cost path method and its implementation into an automatic path tracker tool. We also describe the results of a multiuser study, where we measured the speed and operator dependence of the automated path tracker.