Study Of Brain Tissue Segmentation And Anisotropic Conductivity Model Based On DTI

Diffusion tensor imaging (DTI) is of great significance and clinical importance in the study of cognitive function and neural activity of brain. DTI gets self-diffusion tensor images by measuring the diffusing characteristics of water molecules in brain tissues based on diffusion weighted imaing(DWI). Compared to T2-weighted MRI, DTI could get the imformation of the asymmetric construction distribution of brain tissues. This thesis concentrates on the application of DTI in the study of EEG with the following achievements:According to the diffusion property of water molecues in brain tissues, this thesis puts forward a new 3-D invariant-parameter K-means clustering method of getting the segmentation of WM, CSF and GM. Then the new method was applied to the patient's head DTI data, and satisfactory result was obtained. An accurate segmentation of DTI tensor space needs the combination of the efficient algorithm and the priori knowledge about the diffusion characteristics of different brain tissues.Until now, DTI is the only non-invasive technique for inferring the structure of the white matter tracts. In order to accurately segment corpus callosum out of brain diffusion tensor images, the section of white matter was firstly segmented out using K-means clustering algorithm, and then corpus callosum was segmented out of white matter using graph cuts that was expanded to tensor space. The segmentation result of the patient brain DTI data set shows that tensor-expanded graph cuts is capable of accurate segmenting of corpus callosum. In the process of graph-cut segmentation, the selection of the kernel similar function is very important.In order to track white matter fiber pathways, this thesis put forward a new method based on moving least square method. First, DWI data was converted to DTI data according to the b-value and the table of magnetic gradients, and DTI data was filtered by Gaussian filter. Then, cubic spline function was used as weighted function to fit the diffusion tensors, and under the conditions of provided bending angle and step distance the tracking process began from the ROI. The result shows that this method is suitable for the tracking of WM fiber pathways macroscopically.DTI is also of great significance in the calculation of anisotropic conductivities of brain tissues. The conductivity of brain tissue is an important parameter in EEG study. To get the conductivity, this thesis put forward a new method from the view of electrochemistry based on DTI using Stokes-Einstein and Nernst-Einstein equations. The proposed new method was tested on DTI data of the human subject, and the result was compared with the experiential conductivities of different brain tissues (white matter, grey matter, CSF). The diffusion time is the most important parameter for the model veracity. At last, this model was used to analyze the influence of WM's anisotropy conductivity in EEG forward problem.