A 3D Convolutional Neural Network Approach for Diagnosing Alzheimer's Disease using Modified Owl Search Optimization Technique

Understanding the early stages of Alzheimer's disease (AD) is proving critical for treating the disease and preventing future degeneration. Doctors would examine patients more thoroughly if they could visualise the many morphological aspects for better clinical practises. Previous research has demonstrated the utility of using deep learning to distinguish AD from Normal Control (NC) and achieve a high level of accuracy using T1 weighted MRI images. In this paper, a novel 3D Convolutional Neural Network (3D-CNN) has been proposed for classify three binary classifications using 3D T1-MRI images. For optimizing the weights of proposed 3D CNN network, A Modified Owl Search Algorithm (MOSA) has been applied for optimizing the weights of the proposed 3D CNN network. The proposed model's viability is tested on 404 ADNI subjects, and it achieves the highest classification accuracy when compared to other methods currently in use. The proposed method could assist doctors in the early detection of Alzheimer's disease.