A Voxel Based Morphometry Approach for Identifying Alzheimer From MRI Images Using an Optimized PSO Algorithm

Abstract

Alzheimer’s Disease (AD) is a commonly occurring brain disorder that affects elderly people. It is a progressive, neurodegenerative brain disorder that attacks neurotransmitters, and causes dementia. For the evaluation of normal ageing and AD, Voxel Based Morphometry (VBM) using structural brain Magnetic Resonance Imaging (MRI) has been widely used. This VBM of MRI has data that has been segmented as Gray Matter (GM), White Matter (WM), and Cerebro-Spinal Fluid (CSF) partitions. Anatomical standardization of all the images to the same stereotactic space is done. It makes use of linear affine transformation as well as non-linear warping, smoothing and at last performs statistical analysis. The work suggests the following- Particle Swarm Optimization (PSO) based AdaBoost, using Principal Component Analysis (PCA) for feature reduction and feature extraction using curvelet transform classifier optimization. It is not completely possible by the curvelet transform to characterize the high dimensional signals that contain hyper plane singularities, lines or curves. For decreasing the data set dimensions that contain several interrelated variables, PCA is an effective tool and it can also retain most of the differences. The work also presents an improvised AdaBoost algorithm that is based on optimizing the sample space search. In order to find a threshold in AdaBoost algorithm, more time is needed for comparing samples while working with data on a large scale while making use of the decision stump as a weak classifier. This work makes use of the PSO algorithm in order to change and also choose the most optimal feature in sample space for weak classifiers to reduce computation time. It has been shown via empirical outcomes that the suggested technique performs better compared to the other techniques.