基于多尺度Hessian矩阵和Gabor滤波的造影图像冠脉中心线提取  

Algorithm for centerline extraction of coronary arterial tree in coronary angiographic projections based on Hessian matrix and Gabor filter

作　　者：李颖超[1] 刘越[1] 王涌天[1] 

机构地区：[1]北京理工大学信息科学技术学院光电工程系,北京100081

出　　处：《中国医学影像技术》2007年第1期133-136,共4页Chinese Journal of Medical Imaging Technology

基　　金：国家重点基础研究发展规划项目资助(973项目;2003CB716105)。

摘　　要：目的提取冠脉造影图像中血管的中心线。方法提出一种基于Hessian矩阵和Gabor滤波的血管中心线提取算法,该算法首先利用Hessian矩阵将血管提取,然后根据Gabor滤波器的方向响应特性利用跟踪策略对细化后的血管图像进行校正,得到血管中心线,并提出了精度评价方法。结果将该算法用于随机抽取的图像,均取得高精度的血管中心线以及较理想的血管宽度。结论该算法对于造影图像冠脉中心线提取精度较高,并能同时求得血管宽度。Objective To extract centerline of coronary arterial tree （CAT） in X-ray projections （coronary angiography）. Methods An automatic algorithm for centerline extraction of CAT was proposed. Hessian matrix based segmentation algorithm was applied for the extraction of CAT, followed by thinning the extracted CAT to obtain the skeleton of CAT. By applying Gabor filter based centerline correction algorithm to the whole CAT obtained by tracking the skeleton the centerline as well as the diameter of CAT was obtained. In addition, criterions of precision evaluation of extracted centerline were proposed. Results This experiment showed the proposed algorithm was effective to obtain the accurate centerline of CAT as well as the diameter of CAT. Conclusion The algorithm proposed can be employed for extracting accurate centerline of CAT and the diameter of CAT.

关 键 词：HESSIAN矩阵 GABOR滤波 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]