结合硬斑块特征的心血管内超声图像中-外膜边缘检测

摘要

This paper presents a new approach for the detection of media-adventitia border in intravascular ultrasound ( IVUS) images, to overcome problems of hard plaque and acoustic shadow which make media-adventitia difficult to detect. The approach started with image classification using k-means algorithm, such that the region of hard plaques was delineated from the IVUS image. In a next step, a cost matrix was created by combining image gradient, hard plaque location and image intensity. A heuristic graph-searching was then applied to find the media-adventitia border from the cost matrix. Experiment results showed that the approach improved the accuracy of media-adventitia border detection by overcoming problems caused by acoustic shadow and hard plaque inhomogeneity, and the correct rate reached 95. 57% .%针对心血管内超声(IVUS)图像中-外膜(media-adventitia)边缘检测中,硬斑块和声影等造成中-外膜边缘难以准确检测的问题,提出一种结合硬斑块特征的中-外膜边缘检测方法.首先,采用k均值聚类分割IVUS图像,根据图像中不同类型斑块的成像特点检测硬斑块位置；然后,计算IVUS图像方向梯度,结合斑块位置和图像灰度,获得代价矩阵；最后,利用启发式图搜索方法,实现IVUS图像的中-外膜边缘检测.对临床图像的实验结果表明,本方法能克服图像中声影和斑块干扰等问题,使检测正确率达到95.57％,提高了IVUS图像中-外膜边缘检测的准确性.