A Multimodal Analytical Approach to Alzheimer's Disease Diagnosis Using Machine Learning and Convolutional Neural Networks on MRI Datasets

Abstract

Alzheimer's Disease (AD) is a progressive neurode- generative \disorder that impairs cognitive function, memory, and the ability to perform daily tasks in older adults, often accompanied by changes in behavior and personality. With no cure currently available, existing treatments are most effective in the early and middle stages of AD, emphasizing the importance of early diagnosis as the aging population grows and AD prevalence increases. This study aims to detect early signs of AD by employing machine learning and deep learning techniques to analyze MRI images from the Open Access Series of Imaging Studies (OASIS) dataset. Various models were utilized, including Random Forest and Logistic Regression from traditional machine learning, Extra Trees from ensemble learning, and Convolutional Neural Networks (CNN) from deep learning. The models were evaluated using metrics such as accuracy, precision, recall, and AUC, with CNN showing the highest accuracy and AUC, while Extra Trees excelled in precision and recall, indicating that both deep learning and ensemble learning can provide competitive performance for early AD detection.