Segmentation of the Vascular Tree in CT Data Using Implicit Active Contours

Abstract. We propose an algorithm for the segmentation of blood vessels in the kind of CT-data typical for diagnostics in a clinical environment. Due to poor quality and variance in the properties of the data sets a two level approach using implicit active contours is chosen for the task. A fast pre-segmentation using the fast marching method followed by propagation of a sparse field level set allows a robust segmentation of the vascular tree. Evaluation of the results and observed problems are described.

• Type: Conference
• Year: 2006
• Where: BILDMED
• Authors: Karsten Rink, Arne-Michael Törsel, Klaus D. Tönnies
• Keywords: BILDMED 2006, BILDMED 2007, Fast Marching Method, Implicit Active Contours, Sparse Field Level