Study On Tracking Method Of White Matter Fiber Based On Diffusion Tensor Image

There are a large amount of neural fibers in the human brain. The distributing of the fibers is anfractuous and the diameter of the neural fiber is very small. The distribution of fiber can help doctors to diagnose neural diseases and play the role of navigation in the operation of certain brain. So reconstructing the nerve fibers accurately in the brain has great scientific significance and practical application. At present, one of the research focuses in magnetic resonance (MR) diffusion imaging is to reconstruct neural fibers in the brain white matter with the diffusion tensor imaging (DTI) data.Tractography is an intuitive method to display diffusion tensor images. It is a method to reconstruct nerve fibers which calculates parameters based on diffusion tensor image data and connects the nerve fibers by some rules. The process of fiber tracking is divided into calculating parameters and determining the tracking model. Calculating parameters contains of diffusion tensor and other parameters of the tensor, such as eigenvalues, eigenvectors, fractional anisotropy and so on. Because determining the tracking model will affect the quality of the tracking results directly, it is the key of fiber tracking algorithm. The study of reconstructing nerve fibers is focused on two problems. One problem is that how to identify crossing and branching fibers better so that the results will be more accurate and smooth. The other problem is that how to reconstruct fibers quickly which is to reduce the time of diagnosis by the doctor and diagnose neurological disease faster.This article researches on the tractography based on DTI. First of all, researching on the calculation of diffusion tensor gets the related parameters. Then, we analyze characteristics and problems during the process through studying FACT algorithm, Tensorline algorithm, VCT and probabilistic tracking algorithm. To solve these problems, a new more accurately and faster fiber tracking algorithm is proposed. Finally, we compare these algorithms through experiments. Tracking fibers from one seed is to test the identification capability of crossing fibers between different algorithms. Using the same algorithm but different parameters to track fibers is to validate the efficiency. And we compare ability of fiber tracking through overall tracking. The results show that this method can identify more crossing and branching fibers and improve the efficiency of fiber tracking than other methods. It can reflect the distribution of the neural fibers better in the brain white matter.