Abstract
Abstract Alzheimer disease (AD) destroys brain cells and the patient’s memory is lost as a result of a progressive and incurable neurodegenerative disease. Magnetic Resonance Imaging (MRI) scan brain images can be analyzed with Artificial Intelligence (AI) technology to diagnose this disease and predict its progression. Early diagnosis and personalized treatment of AD based on MRI images are crucial for improving patient outcomes. To overcome these challenges, a novel Tetra Alzheimer disease Classification via Dilated coordinate Attention based RegNet (TACADR) technique has been proposed. The proposed method utilizes Leaky ShuffleNet for extracting structural and spatial features to improve the classification accuracy. Dilated coordinate attention based RegNet is utilized to classify Tetra classes of AD using MRI images. The proposed approach is superior in terms on accuracy, recall, specificity, precision, and F measure, according to experimental data on AD. Experimental results on ADNI-MRI dataset confirms that the TACDAR approach is superior compared to other datasets. The proposed technique improves the accuracy range of 13.5%, 11.8%, 10.8%, 8.64%, 0.45%, 3.88%, 2.78%, and 2.77% better than Custom CNN, TL-CCNN, EM-DL, PSO-CNN, BiLSTM-ANN, STCNN, ML-3DCNN, and LW-CNN respectively.