Early and Multi-Stage Classification of Alzheimer’s Disease Using Deep Learning: A Convolutional Neural Network Approach on MRI Data

Creative Commons License

Atanda O. G., Adebiyi M. O., AWOTUNDE J. B., Abiodun M. K., Olawoye P. O., Ajetunmobi A. J., ...Daha Fazla

NIPES - Journal of Science and Technology Research, cilt.7, sa.1 Special Issue, ss.1174-1182, 2025 (Scopus)

Özet

Alzheimer's disease (AD) is the most prevalent form of dementia, significantly impacting global healthcare. While existing computer-aided systems detect AD, most fail to differentiate its progression stages. This research proposes a novel deep learning-based system leveraging three-dimensional convolutional neural networks (3D CNNs) for automated detection and classification of Alzheimer's disease using MRI scans. The study utilizes a Kaggle dataset, applying rigorous preprocessing techniques, including image resizing, brightness adjustment, and data augmentation, to enhance model performance. To address class imbalance in the dataset, the Synthetic Minority Over-sampling Technique (SMOTE) was employed, ensuring balanced representation across different AD stages and improving the model’s generalizability. The proposed system achieves a remarkable accuracy of 94%, effectively distinguishing between non-demented, very mild, mild, and moderate AD stages. Performance evaluation through Receiver Operating Characteristic (ROC) analysis further validates its robustness. These results highlight the efficacy of 3D CNNs in medical imaging and their potential integration into clinical diagnosis workflows. Future work should explore the incorporation of multimodal data sources, such as genetic markers and cognitive assessments, to enhance predictive accuracy.