Model-guided and inertia-driven directional minimal path for coronary centerline extraction

Extracting centerlines of coronary arteries is a challenging but important task in clinical applications of cardiac computed tomography angiography (CTA). In this paper, we propose an inertia-driven path tracking method based on the directional minimal path (DMP) to automatically extract the coronary centerlines. In DMP, the path tracing is divided into a number of segments. The proposed method adopts the previous segment in the DMP to form the inertia-driven force after the first segment, and thereby increase the robustness of the DMP. To trace the coronary path deeply and thus to obtain a long centerline, we construct a coronary model combining multiple subjects and extract the longest coronary centerline from a branch with a new end-point determination criterion. We evaluated the performance of this method using the Rotterdam Coronary Artery Algorithm Evaluation Framework. The average inside of our centerline extraction results was 0.38mm and the overlap was 83.7%.