一种基于管状特征的冠脉中心线自动提取算法

Automatic Centerline Extraction Based on Tubular Feature for Coronary Arteries

针对冠脉分割不准确以及血管曲率变化比较大导致的中心线提取不准确问题,提出了一种基于管状结构特征的冠脉中心线自动提取算法.首先对冠脉进行基于骨架线节点结构特征的冠脉血管中心线粗提取;然后利用冠脉的管状结构特征,提出三种血管结构判断准则,以检测血管断开情况;最后利用血管的局部结构特征,校正血管中心线,以得到完整、连续的血管中心线.对冠脉CTA体数据进行实验,实验结果表明该算法能够很好地解决冠脉分割不准确以及血管曲率变化比较大情况下造成的中心线提取不正确问题,具有很强的鲁棒性.

To solve the problem of improper centerline extraction due to the improper segmentation of coronary arteries and obviously changed vessel curvature,an algorithm based on tubular structure was proposed for the automatic centerline extraction of coronary arteries.The coronary artery centerline was extracted roughly on the basis of the configuration of the skeleton nodes,and then the criteria were proposed for the three kinds of blood vessel structures according to the tubular structure of coronary so as to detect the actualities of the improper segmentation of coronary arteries.As a result,the disconnected centerline was regulated and joined on the basis of local vascular structure,thus obtaining the complete and connective centerline of coronary arteries.The experimental results of coronary CTA volume data indicated that the proposed algorithm is able to solve well the problem of improper centerline extraction due to improper segmentation of coronary arteries and obviously changed vessel curvature,with high robustness provided.