Refined Alzheimer's Disease Classification: Leveraging MRI-Based Deep Learning Techniques

Abstract

Alzheimer's Disease poses a significant global challenge, affecting numerous individuals and currently lacking effective treatment options. This study proposes a potent methodology utilizing Deep Learning techniques for early detection of Alzheimer's Disease. The dataset comprises brain magnetic resonance imaging scans used for model development and validation. The approach includes preprocessing stages, training a ResNet101 model without oversampling, and then applying Mixup Augmentation to handle class imbalance. Results demonstrate substantial improvements in classification accuracy, rising from 83.5% to 88.7%, alongside enhancements in precision, recall, and F1 score metrics across classes. This underscores the efficacy of the proposed approach in advancing Alzheimer's Disease classification, with potential implications for early detection strategies.