基于LOG算法的DSA图像边缘检测

Automatic vessel boundary extraction and stenosis quantification method based on Laplacian-of-Gaussian algorithm using DSA images

在对心血管狭窄病人的治疗中,常常需要对病人心血管的狭窄情况进行准确评估,而数字减影血管造影技术(DSA)是血管疾病诊断,特别是介入治疗不可缺少的检查手段,如何准确提取DSA血管边缘对于血管狭窄率的测量具有非常重要的意义。针对DSA血管图像的特点,本文在分析传统的LOG(Laplacian-of-Gaussian algorithm)轮廓检测算法存在问题的基础上,进行了改进使其能正确地获得血管的边缘图像,同时利用改进的边界链码对血管边缘进行了分割图像的结构化,将跟踪结果用于血管狭窄率的测量。最后,基于改进的边缘检测算法,开发了血管边缘检测和狭窄率测量工具,取得了良好的检测效果。

The stenosis degree need to be accurately assessed in the treatment of vascular stenosis patients, while digital subtraction angiography（DSA） is the indispensable technology to carry out the assessment, especially in intervention treatment. How to extract the artery boundary in DSA images has good significance on stenosis quantification. In view of the features of DSA imaging, we put forward a new edge-detecting method based on the classical Laplacian-of-Gaussian algorithm. Some improvement is done on the LOG algorithm to accurately acquire the edge image of blood vessel; a boundary chain code method is also adopted to structure the segmented image. The edge-line extracted is subsequently used in stenosis percentage measuring. As a result,we realized a blood vessel boundary extraction and stenosis quantification software using the new algorithm. The new computerized stenosis measuring system will not only greatly increase the speed and accuracy of stenosis measuring,but also reduce the variability between readers.