Automated Diabetic Retinopathy Detection Using Capsule Networks with Explainability

This paper presented an automated diabetic retinopathy detection system using a Capsule Network (CapsNet) integrated with Explainable AI techniques for retinal image analysis. The system effectively classifies DR severity levels and provides visual explanations of predictions, reducing dependency on manual diagnosis. By leveraging the spatial feature learning capability of CapsNet and the interpretability of Grad-CAM, the proposed model achieved high accuracy and reliability in DR detection tasks. The results demonstrate that the model outperforms traditional CNN and ANN-based approaches in both performance and transparency, making it highly suitable for clinical decision support systems. In the future, the system can be enhanced by incorporating larger and more diverse retinal datasets to improve generalization. The integration of advanced explainable AI techniques such as SHAP or LRP can further improve interpretability. Additionally, hybrid models combining CapsNet with other deep learning architectures can be explored to achieve even higher accuracy and robustness.  

[1]               [1] K. Zhang, H. Li, and Y. Liu, "Explainable Deep Learning for Microaneurysm Detection in Diabetic Retinopathy," IEEE Access, vol. 11, pp. 7894-7905, 2023. https://doi.org/10.1109/ACCESS.2023.3245501 [2]               [2] R. Patel, S. K. Singh, and P. Tiwari, "Microaneurysm Detection Using Hybrid CNN-SVM Model with Explainable AI," IEEE Journal of Biomedical and Health Informatics, vol. 27, no. 3, pp. 1332-1343, 2023. https://doi.org/10.1109/JBHI.2023.3284692 [3]               [3] T. Nguyen, M. Rodriguez, and C. Lee, "Deep Learning-Based Microaneurysm Detection in Retinal Images with Integrated Explainability," IEEE Access, vol. 12, pp. 11029-11040, 2023. https://doi.org/10.1109/ACCESS.2023.3300075 [4]               [4] A. Rahman, D. Singh, and H. Nakamura, "Attention-Based Transformer Models for Explainable Microaneurysm Detection in Retinal Images," IEEE Transactions on Artificial Intelligence, 2024. https://doi.org/10.1109/TAI.2024.3356709 [5]               [5] K. Al-Mansoori, X. Chen, and N. Ahuja, "Multi-Scale Feature Fusion with Explainable AI for Microaneurysm Detection," IEEE Transactions on Neural Networks and Learning Systems, vol. 35, no. 2, pp. 2451-2462, 2023. https://doi.org/10.1109/TNNLS.2023.3300541 [6]               [6] M. Chen, S. Patel, and R. Kumar, "Explainable AI Techniques for Robust Microaneurysm Detection Using Hybrid CNN Approaches," IEEE Transactions on Medical Imaging, vol. 43, no. 5, pp. 3225-3236, 2023. https://doi.org/10.1109/TMI.2023.3316751 [7]               [7] W. Zhang, Y. Sun, and R. Gupta, "Microaneurysm Detection in Retinal Images Using Explainable AI and Transfer Learning," IEEE Transactions on Image Processing, vol. 33, no. 8, pp. 5562-5573, 2024. https://doi.org/10.1109/TIP.2024.3356901 [8]               [8] H. Wang, F. Luo, and L. Zhang, "Explainable AI-Driven Ensemble Learning for Microaneurysm Detection in Diabetic Retinopathy," IEEE Transactions on Computational Imaging, vol. 10, pp. 159-171, 2023. https://doi.org/10.1109/TCI.2023.3349805 [9]               [9] N. Ahmed, M. Hasan, and A. Chatterjee, "A Comprehensive Review on Explainable AI for Microaneurysm Detection," IEEE Reviews in Biomedical Engineering, 2024. https://doi.org/10.1109/RBME.2024.3367809 [10]             [10] Y. Chen, X. Li, and H. Zhou, "Explainable AI-Based Microaneurysm Detection Using Deep Neural Networks," IEEE Transactions on Biomedical Engineering, vol. 71, no. 1, pp. 111-122, 2023. https://doi.org/10.1109/TBME.2023.3358701