基于血管平行性和拓扑性的冠脉树分割

Automated Coronary Arterial Tree Extraction using Parallel and Topology

本研究提出了一种基于血管平行性和拓扑性的冠脉树分割方法。利用冠脉的曲率和平行性,获取冠脉段的边缘;利用冠脉的连通性和拓扑性,获取整个冠脉树的边缘。本方法的优点是,冠脉树的跟踪是从中心初始点到冠脉的末梢点,跟踪冠脉树的拓扑结构与真实冠脉的拓扑结构接近,解决了大冠脉段分支被分成几段的缺点,使冠脉段达到了很好的连续性;利用血管边缘的平行性和连通性,能找出准确冠脉的配对边缘,很好的解决了由三维图像投影为二维图像冠脉之间的部分重叠和遮挡问题。上述方法能获取完整、单连通和单像素的冠脉树边缘。实验结果表明,本研究提出的方法有很好的准确性。

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 singlepixel edge of the coronary arterial tree. Good accuracy was observed in the experiment results.