摘要： CT血管造影(CTA)序列图像中冠状动脉(冠脉)的自动跟踪分割能帮助医生进行直观和快速地诊断,具有重要的临床意义。为此,提出一种针对CTA序列图像特点的目标跟踪算法,通过特征匹配和改进粒子滤波进行冠脉目标跟踪,最终生成高精度的冠脉三维模型。其中改进粒子滤波算法改变了粒子重采样规则,为追踪到运动状态突变目标提供基础。实验结果表明,该算法能够解决冠脉追踪中分叉繁多、形态变化以及运动状态变化剧烈等难点问题,具有较高的追踪准确度以及良好的鲁棒性,能够为冠心病的临床诊断起到有效的辅助作用。

关键词: 粒子滤波, 目标跟踪, 特征匹配, CT血管造影图像, 冠状动脉

Abstract: Automatic tracking and segmentation of coronary arteries in Computed Tomographic Angiography(CTA) sequence images have clinical significance and especially can help cardiologists make intuitive and rapid diagnosis.According to the characteristics of CTA images,this paper proposes a tracking algorithm based on feature matching and improved particle filtering to track the target of coronary artery.It realizes the automatic tracking of coronary object,and finally generates a high-precision 3D model of coronary artery.The proposed improved particle filtering algorithm changes particle resampling rules and provides the basis for tracking the suddenly changed motion targets.Experimental results show that the proposed algorithm can solve many difficult problems such as bifurcation,morphological change and severe change of motion state in coronary tracking.It has higher accuracy and better robustness,which can be used for the clinical diagnosis of coronary heart disease.

Key words: particle filtering, target tracking, feature matching, Computed Tomographic Angiography(CTA) image, coronary artery

中图分类号: 

• TP391