改进区域生长法及其在冠状动脉造影图中的应用

An Improved Region Growing Algorithm and Its Application in Coronary Artery Angiographic

冠状动脉造影过程中,由于人体骨骼、肌肉、器官等组织对X射线吸收程度不同,得到的冠状动脉造影图像亮度不均匀,传统的区域生长算法无法准确分割不均匀亮度的图像,而且种子点的选取需要人工交互,效率低下.针对这些问题,提出了一种改进区域生长算法,该算法自动生成一组种子点,种子点生长时,使用生长区域的局部平均值作为生长准则中的参数,最后使用医学影像计算与计算机辅助介入(medical image computing and computer assisted intervention,MICCAI)准则对分割后的图像进行评价.实验表明,使用该算法对冠状动脉造影图像进行分割,能得到较好的结果,且不需要人工交互,提高了图像分割的效率和准确性.

The intensity of coronary artery angiograms is non uniform since different organizations, such as the bones, muscles, and organs, have diffcrem absorption of X ray during angiography. The classical region growing algorithm has poor effect on these non-u- niform intensity images,and it is also inefficient since the seeds must select manually. An improved region growing algorithm is pres ented by this paper, which not only produce seeds automatically,but also use a local parameter in growing criteria. Then we used Mic- cai criteria to evaluate the result of our algorithm. The efficacy of the approach is demonstrated with experiments.