Recognition Of Alzheimers Disease Over MRI Scan Images Using CNN and Transfer Learning

Abstract

Alzheimer’s Disease (AD) is a progressive neurodegenerative condition and the most prevalent cause of dementia worldwide, posing significant clinical and societal challenges. Early and accurate detection is essential for effective intervention but remains difficult due to the limitations of conventional diagnostic techniques, such as manual MRI analysis, which is time-consuming and subject to variability. This study aims to develop an efficient, accurate, and interpretable deep learning model to classify the stages of Alzheimer’s Disease using MRI scan images. Leveraging the VGG16 architecture with transfer learning, the model is fine-tuned to recognize four AD stages—Non-Demented, Very Mild, Mild, and Moderate—using a publicly available Kaggle dataset containing 11,500 T1-weighted MRI scans. Preprocessing steps including skull stripping, Z-score normalization, and data augmentation were applied to enhance model generalizability. The Adaptive Synthetic Sampling (ADASYN) method was also used to address class imbalance. The proposed model achieved a classification accuracy of 91.01%, precision of 0.92, recall of 0.91, F1-score of 0.91, and ROC-AUC of 0.94, outperforming baseline methods such as SVM and 3D CNNs. Grad-CAM was utilized to visualize key brain regions influencing predictions, improving clinical interpretability. This work demonstrates a scalable and computationally efficient solution for multi-stage AD diagnosis from MRI, bridging the gap between high-performance AI and practical clinical application. Future extensions may include multimodal data integration and temporal disease progression modelling for enhanced diagnosis