基于概率跟踪的冠状动脉造影图像的血管树提取

采用传统的基于跟踪的方法分割冠状动脉X射线造影图像的血管时,易受血管节点和曲率的影响.对此,本文提出了基于多特征测度的概率跟踪模型和中心线优化算子,该方法能够准确地分割出血管树的骨架并进行中心线优化.实验中,该方法不仅避免了传统方法的弊端,而且具有较低的计算耗时、较高的鲁棒性、全自动性:一次选定跟踪的起点,可以自动提取出80%以上冠状动脉树.

Extracting the Coronary Artery Tree in Angiographic Projections Based on Probability Tracking

Using the tracking-based methods segmenting the coronary artery tree in X-Ray Angiogram(XRA),the traditional methods were subject to the blood vessel's node and the local high curvature.To overcome the disadvantages,we presented a Multi-Features Measure(MFM) based Probability Tracking Model(PTM) to extract the blood vessel tree,and used a centerline operator to optimize the vessel skeleton-line.In the experiments,the proposed method on one hand avoids the problems of the traditional one;on the other hand the low time-consuming,the high robustness and the full automation are attained.Given an initial point in the interest vessel object,over 80% of the visible vessel branches in XRA image can be automatically delineated.