Analysis of Alzheimer Disease using Optimization Techniques

Abstract

Alzheimer Disease (AD) is a neurodegenerative and most common form of disorder. Most of the researchers are concentrating on AD pathology for the past few decades since it is a major public health disease in the world wide. AD causes neuronal damage in the Brain Internal Regions (BIR) specifically Corpus Callosum (CC), White Matter (WM), Grey Matter (GM), Hippocampus (HC) and ventricle. The neuronal loss makes the changes in anatomical structure of the brain which affects the BIR. Identification of structural difference between the normal controls (NC) and AD of the brain is a challenging process due to its complex networks of nerves. Hence, there is a requirement of segmentation process to extract the BIR. The proposed work focused to segment the CC and ventricle regions using multi-level thresholding methods namely Ant colony Optimization (ACO) and Artificial Bee Colony (ABC) optimization techniques. These methods have been analyzed quantitatively and qualitatively using various measures with ground truth (GT) images. Artificial Bee Colony optimization techniques have been achieved the better accuracy of 93 % than ACO. Deep Learning (DL) classifier has been used to classify normal controls (NC) and AD and acquired the better accuracy of 78% in ABC than ACO. Finally the proposed pipeline provides the clinical evidence for analyzing AD.