中文摘要：

不同功能神经束组的正确分类是神经三维可视化的关键，神经束组交叉穿插、混合重组，导致基于单张图像进行聚类分析的传统算法并不适用。文章以臂丛神经为例分析了神经解剖结构的特殊性，在已解决神经束组轮廓对应问题的基础上，建立起神经束组的三维拓扑结构，引入先验知识，提出了基于三维结构的神经束组分类算法，并在分类过程中进行多次纠正。文章最后给出的实验结果证明，相对传统算法，该文所提出的方法能确保神经束组的可靠分类。

英文摘要：

The correct classification of the nerve fascicles with different functions plays a vital role in the 3D visualization of nerves. Traditional algorithms based on clustering analysis of a single image are not suitable due to the branching, blending and merging of the nerve fascicles. The paper first uses the brachial plexus as an example to analyze the specificity of the topegraphic anatomy of the nerves. Then based on the work that has already solved the problem of nerve contours correspendence, the paper builds up a 3D topological structure of the nerves. Finally after introducing some priori knowledge, a classification algorithm has been proposed for the nerve fascicles based on the 3D structures of the nerves. And in the procedure of classification, many corrections have been done. The experiment resuits show that in the classification of the nerve fascicles, comparing to other traditional algorithms, the method proposed is more efficient and can ensure a satisfactory classification result.