Automated coronary arterial tree extraction using parallel and topology

A method for the coronary arterial tree extraction was presented in virtue of the arterial parallel border and its topological characteristic. The method consists of two steps: the branch extraction based on the arterial curvature and parallelity; the whole coronary-artery-tree border extraction based on the arterial connectivity and topology. Compared with previous approaches, the artery branches were traced from the centerline starting point to the distal ends instead of in an arbitrary direction. The method implemented more accurate approximation to the topology of the actual arterial tree. It circumvented the problem that large branches are divided into smaller segments as a result of previously segmented small branches crossing the main branch. Using the parallelity and topology of the coronary arterial border, the edge was found partnership and the difficulties caused by the overlapping and superposition due to the 3D-to-2D projection were overcome by the method. The method can be used to detect the integrated and single-pixel edge of the coronary arterial tree. Good accuracy was observed in the experiment results.