Implementation and Analysis of Predictive Models for Alzheimer's and Amyotrophic Lateral Sclerosis Diseases Stage Detection using SMOTE

This work presented a hardware–software co-design framework for six-class MRI classification of ALS, Alzheimer’s Stage 1, Alzheimer’s Stage 2, Alzheimer’s Stage 3, Normal, and Others. EfficientNetB0 was used during the software stage for feature extraction, while SMOTE was applied only to the training data to address class imbalance. A lightweight Custom CNN classifier was subsequently trained and its learned parameters were prepared for fixed-point hardware implementation. The trained classifier achieved an overall test accuracy of 96.79%, with a macro-average F1-score of 96.33%. The CNN accelerator was implemented using Verilog HDL and synthesized for the PYNQ-Z2-compatible Zynq-7020 device. The post-synthesis results showed 0.58%  LUT,  0.03%  flip-flop,  0.71%  BRAM,  and 30.45% DSP utilization, demonstrating the low hardware resource requirement of the proposed lightweight architecture. The proposed framework therefore demonstrates the feasibility of combining deep-learning-based MRI classification with FPGA acceleration. By separating software model development from hardware inference, the architecture provides a practical foundation for resource-efficient embedded AI applications. Future work will focus on improving the proposed system through larger and more diverse MRI datasets, additional  independent  validation,  and  further optimization of the lightweight Custom CNN architecture. The FPGA implementation can also be extended through post-implementation timing analysis, measured inference latency, power and energy evaluation, and further fixed-point optimization. Additional hardware parallelism and optimized memory/data-transfer architectures can be investigated to improve throughput. Finally, integration with more comprehensive clinical datasets and explainable AI techniques can be explored to improve the reliability and practical applicability of the proposed FPGA-based medical classification system.

[1]      E. Kina, "TLEABLCNN: Brain and Alzheimer's Disease Detection Using Attention-Based Explainable Deep Learning and SMOTE Using Imbalanced Brain MRI," IEEE Access, 2025, doi: 10.1109/ACCESS.2025.3539550. [2]      S. M. Mahim, M. S. Ali, M. O. Hasan, A. A. N. Nafi, A. Sadat, S. A.  Hasan, B. Shareef, M. M. Ahsan, M. K. Islam, M. S. Miah, and M. B. Niu, "Unlocking the Potential of XAI for Improved Alzheimer's Disease Detection and Classification Using a ViT-GRU Model," IEEE Access, vol. 12, pp. 8390–8412, 2024, doi: 10.1109/ACCESS.2024.3351809. [3]      S. E. Sorour, A. A. A. El-Mageed, K. M. Albarrak, A. K. Alnaim, A. A. Wafa, and E. El-Shafeiy, "Classification of Alzheimer's Disease Using MRI Data Based on Deep Learning Techniques," Journal of King Saud University – Computer and Information Sciences, vol. 36, no. 2, Art. no. 101940, Feb. 2024, doi: 10.1016/j.jksuci.2024.101940. [4]      P. Goyal, R. Rani, and K. Singh, "A Multilayered Framework for Diagnosis and Classification of Alzheimer's Disease Using Transfer Learned AlexNet and LSTM," Neural Computing and Applications, vol. 36, no. 7, pp. 3777–3801, Mar. 2024, doi: 10.1007/s00521-023-09301-6. [5]      M. Likhita, K. M. Kumar, N. S. Sasank, and M. Abhinaya, "AD-ResNet50: An Ensemble Deep Transfer Learning and SMOTE Model for Classification of Alzheimer's Disease," in Proc. International Conference on Innovative Computing and Communications, Singapore: Springer, 2023, pp. 699–713. [6]      Alzheimer's Association, "2024 Alzheimer's Disease Facts and Figures," Alzheimer's & Dementia, vol. 20, no. 5, pp. 3708–3821, 2024. [7]      O. Russakovsky et al., "ImageNet Large Scale Visual Recognition Challenge," International Journal of Computer Vision, vol. 115, no. 3, pp. 211–252, 2015. [8]      M. Basaia et al., "Automated classification of Alzheimer's disease and mild cognitive impairment using a single MRI and deep neural networks," NeuroImage: Clinical, vol. 21, pp. 101645, 2019. [9]      S. Payan and G. Montana, "Predicting Alzheimer's disease: A neuroimaging study with 3D convolutional neural networks," Pattern Recognition, vol. 63, pp. 355–365, 2017. [10]    A. Krizhevsky, I. Sutskever and G. E. Hinton, "ImageNet Classification with Deep Convolutional Neural Networks," Advances in Neural Information Processing Systems (NeurIPS), 2012. [11]    K. He, X. Zhang, S. Ren and J. Sun, "Deep Residual Learning for Image Recognition," Proceedings of IEEE CVPR, pp. 770–778, 2016. [12]    M. Tan and Q. Le, "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks," Proceedings of ICML, pp. 6105–6114, 2019. [13]    N. V. Chawla, K. W. Bowyer, L. O. Hall and W. P. Kegelmeyer, "SMOTE: Synthetic Minority Over-sampling Technique," Journal of Artificial Intelligence Research, vol. 16, pp. 321–357, 2002. [14]    N. Japkowicz and S. Stephen, "The Class Imbalance Problem: A Systematic Study," Intelligent Data Analysis, vol. 6, no. 5, pp. 429–449, 2002.