Diagnosis of Alzheimer disease in MR brain images using optimization techniques

Abstract

Nature-inspired algorithms play a vital role in various applications, namely image processing, engineering, industrialized designs and business. Generally, these algorithms are inspired by the nature which is helpful in segmenting the brain internal regions, namely cerebrospinal fluid, grey matter HC, white matter, ventricle and so on. Segmentation of hippocampus (HC) is a very hectic process due to its anatomical structure of the brain. This work has been recommended for different optimization techniques such as lion optimization algorithm (LOA), genetic algorithm, BAT algorithm, particle swarm optimization and artificial bee colony optimization to segment HC region from the brain subregions. The comparison of these optimization methods has been evaluated, and it showed better performance in LOA due to its individualities of escaping from local optima. From the obtained results, it is witnessed that the LOA has ability to segment HC region with high accuracy of 95%. The LOA method showed the best classification accuracy compared to all other methods. Finally, the mini-mental state examination score validation has been attempted to reach the clinical targets as HC region is a major hallmarks for diagnosing AD. The overall process of the proposed work demonstrates the abnormalities in the brain natural history which provides the reliable and accurate indication to the clinician about AD progression.