MicroCT图像中周围神经束轮廓的建模  

Modeling of Contour of Peripheral Fasciculi in MicroCT Images

作　　者：孙思语 钟映春[1] 郑海阳 戚剑[2] SUN Si-yu;ZHONG Ying-chun;ZHENG Hai-yang;QI Jian(School of Automation,Guangdong University of Technology,Guangzhou Guangdong 510006,China;Department of Microsurgery,First Affiliated Hospital of Sun Yat-sen University,Guangzhou Guangdong 510080,China)

机构地区：[1]广东工业大学自动化学院,广东广州510006 [2]中山大学附属第一医院显微外科,广东广州510080

出　　处：《计算机仿真》2022年第1期258-262,317,共6页Computer Simulation

基　　金：国家自然科学基金项目(61975248);广东省自然科学基金项目(2018A0303130137);广东省高性能计算重点实验室开放项目(TH1528);广州市科技计划项目(2020007040004)。

摘　　要：针对周围神经MicroCT图像中非分裂/合并阶段神经束轮廓的模型难以构建的问题,在建立神经束轮廓离散控制点数据集的基础上,提出了一种采用3阶准均匀B样条曲线拟合神经束轮廓的方法。方法以Dice系数为评价指标,探索模型中合适的控制点数目,以豪斯多夫距离作为轮廓建模误差的评价指标,度量所有神经束轮廓建模的精度。实验结果表明,采用等间距方式采样神经束轮廓的离散点数据集,再采用3阶准均匀B样条曲线构建神经束轮廓模型,不仅具有较高的建模精度,而且模型复杂度也比较低。On the basis of sampling the data set of contours, the contours of nerve bundles were modeled with 3-order quasi-uniform B-spline method respectively, which was proposed to handle the modeling problem of the contours of the fascicular groups during the non-splitting/merging phase. Then the appropriate number of control points was optimized, which was used to evaluate the model results using Dice coefficient. And the modeling error was evaluated by the Hausdorff distance. The experiments show that, the approach of this paper is not only accurate but also simple to model the contour of fascicular groups.

关 键 词：周围神经 神经束轮廓 样条曲线模型 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]