Improving Alzheimer’s Disease Detection with Transfer Learning

Abstract

Accurate and prompt diagnosis of Alzheimer's disease (AD) remains a challenge, with only a small percentage of patients receiving timely confirmation. Manual interpretation of MRI scans, the primary diagnostic tool, is time-consuming, subjective, and prone to error, particularly in differentiating between disease stages. This study aimed to develop a computer-aided diagnosis system (CAD) for AD classification using deep learning models. MobileNetV1 and Xception architectures were employed to classify AD into four stages: mild, normal, moderate, and severe. Transfer learning and layer freezing techniques were applied for feature extraction and classification. Model performance was evaluated using precision, recall score, and accuracy metrics. The Xception model achieved a higher accuracy (79%) compared to MobileNetV1 (73%) in classifying AD stages. Compared to MobileNetV1, this study shows that Xceptionbased CAD systems have the potential to diagnose AD more accurately, providing a promising path for future research and clinical application.