Unveiling Alzheimer's Disease via MRI: Deep Learning Approaches for Accurate Detection

Abstract

This study conducted a comparative analysis of machine learning algorithms to discern the most effective approach for detecting Alzheimer's disease (AD). Timely and precise detection of AD stands crucial for efficient intervention in this prevalent neurodegenerative disorder among the elderly. Deep learning algorithms have demonstrated promising outcomes in AD diagnosis through the examination of magnetic resonance imaging (MRI) scans. The research focused specifically on evaluating the performance of diverse deep learning architectures, encompassing CNN (Convolutional Neural Network), in detecting AD using MRI images. Utilizing a substantial dataset comprising MRI scans from both AD patients and healthy individuals, the models were trained to automatically extract discriminative features from these images. Experimental results underscore the effectiveness of the proposed models, notably the active use of MobileNet and CNN, which achieved an impressive accuracy of 95.92% in identifying Alzheimer's disease. These findings highlight the superior performance of CNN and MobileNet compared to DenseNet and Inception V3 in AD detection, emphasizing their potential for accurate identification of AD compared to other algorithms. Such insights offer valuable direction for selecting the most appropriate algorithm for AD diagnosis, considering critical factors such as accuracy, computational efficiency, and resource availability. However, further exploration and validation employing larger and more diverse datasets are essential to establish the broader applicability and clinical relevance of these algorithms in real-world scenarios for AD detection.