Atlas-Based Hippocampus Image Segmentation and the Impact to Nerve Fibers Measurements in Magnetic Resonance Imaging

• Main Authors: Ta-Chih Lin, 林達智
• Other Authors: JA CHIH Fu
• Format: Others
• Language: zh-TW
• Published: 2016
• Online Access: http://ndltd.ncl.edu.tw/handle/3bvftr

碩士 === 國立雲林科技大學 === 工業工程與管理系 === 105 === With the increasing age, the patient who suffer from Alzheimer’s disease may occur cognitive dysfunction. The main dysfunction is about memories declining. According to the research, the earliest change area of brain is in hippocampus. The hippocampus can store and locate memories. Not only the short-term memories but also the long-term memories are relevant to the hippocampus. And, it is also an important area when the doctor makes diagnoses of Alzheimer’s disease. In terms of currently medical imaging, the Magnetic Resonance Imaging which can provide the hippocampus images, is the only way to capture the nervous information in the living body. Diffusion Tensor Imaging is used to quantize the nervous fiber information, and we get the average of eigenvalues in concerning area by calibrating DTI with MRI. This study use Voxel-Based Morphometry and DARTEL-based VBM of hippocampus and compare the results of DARTEL-based VBM parameters optimization and with Golden Standard. We get the average of eigenvalues by calibrating MRI from VBM , golden standard and DARTEL-based VBM, and we also calculate FA、MD、C_l、C_p、C_s . After that, we can find out which one is more accurate than others The result of this study, In normal subjects, by DARTEL based-VBM after segmentation the volume of statistical analysis performance paired sample T test, P value of 0.002 was less than 5% of significant level. The nerve fiber performance metrics aspect, FA、MD、C_l、C_p、C_s five kinds of indicators are significant levels, Therefore, DARTEL-based VBM segmentation image and VBM segmentation image has significant differences. In Alzheimer’s subjects, by DARTEL based-VBM after segmentation the volume of statistical analysis performance paired sample T test, P value of 0.000 was less than 5% of significant level. The nerve fiber performance metrics aspect, FA、MD、C_l、C_p、C_s five kinds of indicators are significant levels, Therefore, DARTEL-based VBM segmentation image and VBM segmentation image has significant differences.