Abstract: This paper analyzes Diffusion Tensor Imaging(DTI) from the perspective of Riemannian manifold. The brain white matter fiber between two voxels is transformed into geodesic between two points in Riemannian manifold, and geodesic’s numerical solution is based on Level-Set method, whose result is the brain white matter tractography. Phantom of brain white matter fiber is used to compare the present algorithm with traditional algorithm. Experimental result shows that the algorithm improves accuracy and robustness.

Key words: Diffusion Tensor Imaging(DTI), Riemannian manifold, brain white matter fiber, geodesic, Level-Set method

摘要： 从Riemannian流形的角度分析扩散张量成像，将脑白质中任意两点间的纤维束生成问题转化为计算Riemannian流形中两点间测地线的问题，通过Level-Set方法计算测地线，并将其作为脑白质中两点间的纤维束。利用模拟脑白质纤维束对该算法和传统算法进行比较，实验结果表明，该算法在准确性、鲁棒性等方面有较大改进。

关键词: 扩散张量成像, Riemannian流形, 脑白质纤维束, 测地线, Level-Set方法

CLC Number: 

• TP391