ALZHEIMER’S DISEASE PROGRESSION ANALYSIS AND CLASSIFICATION USING T1-WEIGHTED MRI

Abstract

Alzheimer’s disease (AD) is a considered one of the common elderly diseases. It is a type of dementia that causes changes in behavior in addition to memory loss because of the death of brain cells. There are three stages for Alzheimer disease named: Alzheimer’s Disease patient (AD), Mild cognitive impairment (MCI) and Early stage. In this work, we proposed a promising method to classify the different categories of Alzheimer and the healthy control (HC) subjects using multiple T1-weighted MRI scans of the whole brain volume directly to extract several features by subtracting the longitudinal data of different visits and compute the associated changes in the brain. These features are then fed to the Support Vector Machine (SVM) classifier. The main advantage of this method is that it doesn’t involve lots preprocessing steps including the segmentation that was done to extract the hippocampus or amygdala or any other region of interest, which is considered as an expensive and complicated process. The second part of this thesis is employing a bio-statistical anaylsis to compute the cross-sectional correlation/regression between different clinical assessments such as MMSE,… and … and four Volume of Interest (VOI) named hippocampus, amygdala, lateral ventricles and total brain volume formed of WM and GM. It was observed that MMSE is the most significant assessment, having a high correlation with the four VOI. The graphical representation of the volumetric changes in the different VOI was studied longitudinally along with the shrinkage rate of hippocampus, amygdala and overall brain volume as well as the enlargement rate of lateral ventricles through the progression stages of the disease compared to the normal subjects.