基于医学影像的血管快速提取与可视化  

Medical Image Based Fast Extraction and Visualization of Blood Vessel

作　　者：吕新荣[1] 高新波[1] 邹华[1] 

机构地区：[1]西安电子科技大学电子工程学院,西安710071

出　　处：《中国生物医学工程学报》2009年第4期527-534,共8页Chinese Journal of Biomedical Engineering

基　　金：国家自然科学基金资助项目(60771068);973项目(2006CB705700);教育部长江学者和创新团队支持计划(IRT0645)

摘　　要：在计算机辅助诊断系统中,血管的提取与可视化是血管研究的重要组成部分。本研究利用八叉树实现了血管的双阈值快速分割和边界距离场的快速计算;然后建立基于边界距离场的血管组织最大生成树,并提取感兴趣血管分支的树的主干,即该分支的中心路径,最后用基于图形处理器的三维纹理体绘制方法沿着路径显示血管虚拟内窥镜,用曲面重建方法显示该血管分支的相关信息。实验结果表明八叉树能有效的提高血管分割速度,并减少边界距离场的计算时间,体绘制算法和曲面重建方法能快速的显示血管组织的详细信息,为疾病诊断提供技术支持。Vessel extraction and visualization are two important issues of vessel research in computer-aided diagnosis system. In this paper, an octree structure is introduced to accelerate the extraction of blood vessel based on doublethreshold segmentation and the computation of distance from boundary （DFB） field. Based on results above, a maximum spanning tree （MST） is established. Then, the trunk of the MST is extracted from the vessel branch of interest, which is the centerline of this vessel branch. Finally, a 3D texture volume rendering algorithm based on graphics processing unit （GPU） is employed to realize the fast vessel endoscopy along the path. A curved planar reformation （CPR） algorithm is utilized to reveal the features of the vessel branch. The experimental results illustrate that the octree structure is able to accelerate vessel segmentation and computation of DFB field effectively. The detailed information of vessels can be shown clearly by volume rendering and CPR for disease diagnosis.

关 键 词：八叉树 边界距离 最大生成树 体绘制 曲面重建 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]