中文摘要：

为了解决基于扩散张量成像（DTI）的跟踪过程难以刻画脑白质内复杂纤维结构的问题，提出一种基于离散球面反卷积的确定性跟踪算法．该算法采用离散纤维方向密度函数建立球面反卷积模型，解除对连续球面函数模型的依赖，获得高角度分辨率识别；借助球面高斯函数拟合以补偿离散误差，并在此基础上实现流线型白质纤维跟踪．合成数据、仿真数据和实际临床数据表明：该模型能显著提高体素内小角度交叉纤维的分辨率，并有效抑制噪声．相比于基于其他模型的重构算法，该算法能够更准确地反映活体脑神经组织的真实连接情况．

英文摘要：

In order to resolve the problem that the tracking process based on diffision tensor imaging （DTI） has difficulty in describing the complex fiber structure of the brain white matter, a deterministic tracking algorithm based on discrete spherical deconvolution was proposed. The algorithm uses discrete fiber orientation density function to build the spherical deconvolution model, which aims at relieving the dependence on the continuous spherical function model and getting high angular resolution identification. A spherical Gaussian function was used to make up for the discretization error, then the streamline tracking algorithm was implemented on the basis of the aboving model. Experimental results concluded from the synthetic data, platform data and real clinical data demonstrate that the proposed model evidently improves the resolution of small angle crossing fibers within voxel, meanwhile the noise is effectively restrained. Compared with the reconstruction algorithms based on other models, the proposed algorithm can reflect the true connection of brain neural tissue in vivo more accurately.