改进型骨骼细化算法提取冠状动脉中心线  

Extracting coronary artery centerline based on improved skeleton thinning algorithm

作　　者：张子恒 祝磊 马骏 徐平[1] 刘亦安 薛凌云[1] ZHANG Ziheng;ZHU Lei;MA Jun;XU Ping;LIU Yi'an;XUE Lingyun(School of Automation,Hangzhou Dianzi University,Hangzhou 310018,China)

机构地区：[1]杭州电子科技大学自动化学院,浙江杭州310018

出　　处：《中国医学影像技术》2020年第9期1364-1369,共6页Chinese Journal of Medical Imaging Technology

基　　金：国家自然科学基金-浙江两化融合联合基金重点项目(U1609218)。

摘　　要：目的运用全自动骨骼细化算法从CT图像中精确提取冠状动脉的中心线。方法分割CT图像中的冠状动脉区域,经三维重建得到完整的冠状动脉三维数据;利用骨骼细化算法提取该冠状动脉的中心线,引入Dijkstra最短路径算法提升提取精度。结果相比未移除分支的骨骼细化算法,重叠率提升2%,平均距离减少38.2%,平均运行时间0.48 s。结论改进型骨骼细化算法可有效提取冠状动脉中心线。Objective To propose an automatic skeleton thinning algorithmfor accurate extracting the center line of coronary artery from CT images.Methods The region of coronary artery on CT images were segmented,and complete 3D data of coronary artery were obtained after 3D reconstruction.Then the center line of coronary artery was extracted using skeleton thinning algorithm.Dijkstra algorithm was introduced according to the characteristics of coronary artery to improve the accuracy of extraction.Results Compared with the skeleton thinning algorithm which didn’t remove mini branches,the rate of overlap was increased by 2%,the average distance was reduced by 38.2%,and the average running time was 0.48 s.Conclusion Using improved skeleton thinning algorithm could effectively extract the center line of coronary arteries.

关 键 词：冠状血管 中心线提取 骨骼细化算法 DIJKSTRA算法 

分 类 号：TP3-05[自动化与计算机技术—计算机科学与技术] R54[医药卫生—心血管疾病]