Krill Herd optimized feature selection with structure optimized Neural Network for classification of Alzheimer’s Disease

Abstract

Due to the increase in the aging population, Alzheimer’s Disease (AD) occurrence is also increasing, resulting in a global public health crisis. Magnetic Resonance (MR) brain images are extensively used for identifying Alzheimer’s disease. For automating bioimage processing, image processing techniques and high computing resources are required. Cloud computing is fast becoming a requirement in the medical field for the storage and processing of medical images. Various techniques are available for automatic detection of Alzheimer’s disease using Magnetic Resonance brain images. Feature selection refers to the method of getting a score for every prospective feature based on which the best features are selected. Feature selection assists in the reduction of dimensionality, increase of accuracy as well as the removal of irrelevant data in an effective manner. Krill Herd (KH) has been the recipient of increasing interest from researchers on its several advantages. Chi-Square is utilized as a feature selection method for training as well as testing data of the Artificial Neural Network (ANN) for predicting accuracy. A Krill Herd Optimized Feature Selection with Structure Optimized Neural Network has been proposed for the classification of the Magnetic Resonance Images.