Parameterized Model Construction Of White Matter Fiber Based On Healthy Subjects

Diffusion tensor imaging is a magnetic resonance imaging technique capable of detecting the diffusion properties of water molecules in vivo. It has the characteristics of non-invasive and is widely used in brain research. The brain white matter bundle imaging based on diffusion tensor imaging is one of the hotspots of current neuroscience research. It can restore the white matter fiber tracts through tracing the diffusion movement of water molecules in the brain with tensor information. The restoration of white matter bundles can help people to understand the influence of disease on white matter pathway correctly, so as to diagnose white matter disease of brain caused by lack of fiber effectively.However, since the brain white matter morphology are different across subjects, there is no comparability between different subjects. So it is difficult to make statistical analysis of white matter bundles on group level. This dissertation studied and realized a series of parametric method of white matter fiber based on brain tensor atlas constructed by our team. The method can establish one-to-one correspondence between the fiber bundle points across different individuals. The mainly work and innovation points are as follows.First, we proposed a parametric method of white matter fiber based on tensor atlas in this dissertation. Here the tensor atlas we used was constructed by group-based tensor atlas construction method proposed by our team. It is study-specific and is able to reduce the individual diversity and make the spatial standardization of individual subject better. The parametric method of white matter fiber includes bundle clustering, prototype fiber selection, prototype fiber spatial discretization and fiber points matching. This procedure established a common coordinate system through the atlas tract parameterization, so as to establish correspondence of fiber bundle point across subjects. As for the selection of prototype and fiber points matching, we introduced more than one method, respectively. And we selected the optimal method through estimating and analyzing all these methods to implement a set of fiber bundle parametric method.Then, we applied the parametric method of white matter fiber based on tensor atlas to construct parameterized model of corticospinal tract of healthy subjects. We can obtain the individual and group level diffusion properties curve as functions of arc length through the mapping of fiber properties from atlas to individuals. On this basis, we made preliminary discussion on statistical analysis of parameterized model.According to the research of this dissertation, the parameterized model construction method of white matter fiber based on tensor atlas can not only accurately position and quantitatively analyze white matter fiber tracts, but also make it possible for quantitative characteristic value of white matter bundle to make statistical analysis on group level. So it is of great significance on the study of white matter disease.