Research On The Construction Of Real Head Model For EEG And MEG

This thesis mainly deals with various aspects for constructing real head and brain Boundary Element Method (BEM) calculation model by Magnetic Resonance images (MRI), the research is dedicated to electroencephalogram (EEG) and magnetoencephalogram (MEG) research. The main aspects of the thesis are as below: (1) Image De-noising: Developed image filter software based on Donoho抯 wavelet threshold method; made some improvement on the method in choosing threshold by combining soft-hard threshold, and made the threshold different with different scales and different directions in the same scale. (2) Developed Image Segmentation Algorithm: Due to the Gradient Vector Flow (GVF) active contour model method having weak convergence at corner, some improvement was made in finding brain contours; Multi-resolution analysis method was adopted to further optimize the algorithm. An axial brain image was segmented into gray matter and white matter by using histogram based Fuzzy c means algorithm. (3) Acquisition of Head and Brain Volumetric Data and Development Real Head and Brain Reconstruction and Visualization Algorithms: Firstly, by using threshold technique and morphological operations such as dilate and erosion, head data were obtained. Secondly, using obtained brain contour to construct an initial template, designed and developed a naval algorithm to make the initial template dynamic change with image slice to get the brain data; Thirdly, designed a framework for searching, sampling and storing brain contour data. Uniformed contours stacked together and constructed real Boundary Element Method (BEM) calculation models h respectively by minimum distance method and 3D-Delaunary based triangulation growth method. Finally developed brain and head visualization algorithm based on surface rendering for the purpose of integration of structural and functional information. (4) The Numerical Calculation for EEG Forward Problem.