基于3D医学图像的血管三维分割  

3D Segmentation for Blood Vessels Based on 3D Medical Images

作　　者：董建园[1] 郝重阳[1] 

机构地区：[1]西北工业大学电子信息学院,西安710072

出　　处：《中国生物医学工程学报》2008年第6期848-853,共6页Chinese Journal of Biomedical Engineering

基　　金：国家博士点基金(20040699015)

摘　　要：血管三维分割在血管疾病(如狭窄或畸形)诊断、手术规划和手术引导等许多实际应用中发挥重要作用。但三维分割的实时性仍是一个难题。本研究提出一种基于水平集的快速三维血管分割方法,该方法用内、外邻域曲面来描述被分割目标的边界,并定义水平集函数为简单的整数符号距离函数。通过扫描内外邻域曲面上的点,使之在速度场的作用下向目标边界移动。该方法的不同之处在于利用简单的模型极大地减小了计算量,分割速度快,大容量的MS-CTA[2563体素]图像可在20s内处理完毕。同时,展示了一些三维血管的分割实例。特别需要指出的是,对于人体的大血管,可在不经血管造影的情况下,直接从CT等三维图像中分割出来。3D segmentation for blood vessels plays an important role in solving several piratical application such as diagnosis of vessel diseases （stenosis or malformations） , surgical planning and surgical guidance. But the real-time 3D segmentation is still a problem to be solved. In this article, a fast 3D segmentation method based on the level sets for blood vessels was proposed. Object boundaries were represented as interior and exterior neighboring surface and the associated level set function was defined as a simple integer sign distance function on the same grid. By scanning points in inner and outer surface driven by the speed field, the surface was move inward or outward to the boundary of the object. This approach differed from previous methods in that the model was very simple, which greatly enhanced the computational speed. Volumes of MS-CTA images [256^3 voxol] could he segmented within 20s. Some 3D blood segmentation experiments was made on medical images as well, especially big artery vessels could be directly segmented from 3D CT images without angiography.

关 键 词：三维分割 血管分割 水平集 医学图像处理 

分 类 号：R318[医药卫生—生物医学工程]