基于双源CT图像的冠状动脉分割

Segmentation of coronary artery from dual-source CT images

背景:在计算机辅助下,从双源CT图像中把三维冠状动脉分割出来能为其定量评价提供基础。但冠状动脉的三维形态复杂多变,且其管径细小,因而实现冠状动脉的高精度分割是一项有挑战性的课题。目的:解决冠状动脉难以实现高精度分割的问题。方法:采用三步数据处理策略实现冠状动脉分割。先采用阈值方法对三维双源CT图像进行预分割;然后,采用交互式的策略分割出与主动脉相连的左、右冠状动脉始端;最后,根据冠状动脉始端的位置,利用形态学方法和三维断层图像相邻层间的关系分割出三维冠状动脉。结果与结论:提出的基于形态学与断层图像层间关系的分割方法能较精确地从双源CT图像中分割出左、右冠状动脉,说明该方法适用于三维冠状动脉的分割。

BACKGROUND： Segmenting the three-dimensional coronary artery from dual-source CT images aided by computer can provide the basis for quantitative evaluation. However, the complicated three-dimensional shape and narrowness of coronary artery make it a challenging task to accurately segment the coronary artery. OBJECTIVE： To solve accurate coronary artery segmentation problem. METHODS： A three-step data processing strategy was used to achieve coronary artery segmentation. The three-dimensional dual-source CT images were first pre-segmented with threshold method. Then, the origins of left and right coronary artery which were connected to aorta were segmented using an interaction method. Finally, according to the location of coronary artery’s origin, three-dimensional coronary artery was segmented using a method based on morphology and neighboring connection relationship between slice images. RESULTS AND CONCLUSION： The proposed approach based on morphology and connection relatioship between slice images could accurately segment the left and right coronary artery from dual-source CT images. It suggests that the proposed approach is adequate for segmenting three-dimensional coronary artery.