Summary: | 碩士 === 長庚大學 === 電機工程學系 === 99 === Abstract
The aim of this thesis is to assess the feasibility of an image based diagnose of Alzheimer’s disease (AD) at an early stage, and furthermore to probe the underlying mechanism of the disease for an improved disease control or possible intervention. The image based biomarker in this study was focused on Magnetic Resonance Image (MRI). Diffusion imaging is an advance of MRI, which could provide functional information and is able to assess the microenvironmental changes in tissues. The hypothesis in the thesis is that functional alteration would precede morphological changes as the disease progresses. In this regard, current study uses diffusion imaging to investigate the disease. Because mild cognitive impairment (MCI) was often considered as a transit state between cognitively normal subjects and patients with Alzheimer’s disease, 40 MCI patients were included. Additional MR Images were obtained from 54 normal subjects and 40 patients with AD.
The algorithm to reconstruct diffusion image was implemented in a user-friendly matlab program in this thesis, including the diffusion tensor and the derived indices. A pure water phantom was developed and images were acquired for validation study. Additionally, an algorithm was developed to normalize the images from individual subjects into a standard coordinate of anatomical automatic labeling (AAL). The program further divides the cortex into 15 areas. Diffusion indices of interest were calculated. Statistical analysis was performed, which showed 94% diagnostic efficiency when comparing normal subjects to patients with AD in temporal lobe.
Because diffusion tensor imaging is mostly a white matter technique, the interpretation of changes in gray matter could be difficult. To explore the neurophysiological changes in cortex, diffusion kurtosis imaging was also implemented here. The extent of water diffusion as deviated from Gaussian distribution was calculated. It could provide the information of tissue complexity in cortex. The finding was showed 75% diagnostic efficiency when comparing normal subjects to patients with MCI in cingulate gyrus.
|