Towards Convolutional Neural Networks for Diabetic-Retinopathy Detection
Resumo
This study explores the potential of two lightweight convolutional neural networks, EfficientNet-B3 and DenseNet-169, for the diagnosis of Diabetic Retinopathy (DR), a microvascular complication of chronic hyperglycemia and a leading cause of preventable blindness among working-age adults. Threatening vision in nearly a million people who live with diabetes, timely and accurate detection is crucial. We analyze related theoretical concepts and metric evaluations, demonstrating the feasibility of compact CNN backbones for real-world deployment. All results are derived from the public EyePACS dataset, with the quadratic-weighted kappa metric used for evaluation.
Palavras-chave:
Convolutional Neural Network, Deep Learning, Diabetic Retinopathy Detection, EfficientNet-B3, DenseNet-168
Referências
Akhtar, S., Aftab, S., Kousar, S., and Saeed, A. Q. (2023). Rsg-net: A deep neural network for diabetic retinopathy severity grading. Computers in Biology and Medicine, 161:107081.
American Academy of Ophthalmology (2020). Diabetic retinopathy preferred practice pattern. [link]. Accessed 4 May 2025.
Asif, M., Ur Rehman, F., Rashid, Z., Hussain, A., Mirza, A., and Qureshi, W. S. (2025). An insight on the timely diagnosis of diabetic retinopathy using traditional and ai-driven approaches. IEEE Access, 13:116869–116886. DOI: 10.1109/ACCESS.2025.3583647
Esteva, A., Kuprel, B., Novoa, R. A., and et al. (2017). Dermatologist-level classification of skin cancer with deep neural networks. Nature, 542:115–118. DOI: 10.1038/nature21056
Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press.
Gulshan, V., Peng, L., Coram, M., and et al. (2016). Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA, 316(22):2402–2410. DOI: 10.1001/jama.2016.17216
Haque, M., Winston, K., and et al. (2022). Densenet-169 ensemble for robust diabetic retinopathy detection. In IEEE International Conference on Healthcare Informatics.
Huang, G., Liu, Z., van der Maaten, L., and Weinberger, K. Q. (2017). Densely connected convolutional networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). DOI: 10.1109/CVPR.2017.243
International Diabetes Federation (2023). Idf diabetes atlas, 10th edition. [link]. Accessed 4 May 2025.
Kaggle (2015). Diabetic retinopathy detection dataset. [link]. Accessed 3 May 2025.
Krause, J., Gulshan, V., Rahimy, E., Widner, K., and et al. (2018). Grader variability and the importance of reference standards for evaluating machine-learning models for diabetic retinopathy. Ophthalmology, 125(8):1264–1272. DOI: 10.1016/j.ophtha.2018.01.034
Krizhevsky, A., Sutskever, I., and Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems, pages 1097–1105.
Li, J., Wang, Y., and Chen, R. (2024). On-device diabetic retinopathy screening with a mobile efficientnet model. IEEE Journal of Biomedical and Health Informatics, 28(4):1553–1562.
Lin, Q., Zhang, X., and Chen, Y. (2020). Diabetic retinopathy grading with efficientnet and enhanced pre-processing. Biomedical Signal Processing and Control, 62:102123.
Lin, T.-Y., Goyal, P., Girshick, R., He, K., and Dollár, P. (2017). Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 2980–2988. DOI: 10.1109/ICCV.2017.324
Mehboob, M., Raza, M. A., and Shahzad, H. (2023). Diabetic retinopathy detection using densenet-169 with transfer learning. In International Conference on Intelligent Systems, pages 343–350.
Nair, R., Phene, S., and Krause, J. (2023). Deep learning for diabetic retinopathy screening: A systematic review and meta-analysis. Ophthalmology Retina, 7(2):97–109.
Rajpurkar, P., Irvin, J., Zhu, K., and et al. (2017). Chexnet: Radiologist-level pneumonia detection on chest x-rays with deep learning. arXiv preprint arXiv:1711.05225. DOI: 10.48550/arXiv.1711.05225
Siahkali, M. and Bagherian, L. (2023). A lightweight multi-stage framework for smartphone-based diabetic retinopathy screening. Frontiers in Medicine, 10:1184756.
Silva, R., Pereira, P., and Mendonça, L. (2021). Efficientnet-b3 for highly accurate diabetic retinopathy grading. In Medical Image Computing and Computer-Assisted Intervention (MICCAI).
Tan, M. and Le, Q. (2019). Efficientnet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36th International Conference on Machine Learning (ICML). [link]
Ting, D. S. W., Pasquale, L. R., Peng, L., and et al. (2019a). Artificial intelligence and deep learning in ophthalmology. Nature Medicine, 25:14–24. DOI: 10.1136/bjophthalmol-2018-313173
Ting, D. S. W., Pasquale, L. R., Peng, L., and et al. (2019b). Artificial intelligence for diabetic retinopathy screening: A meta-analysis. British Journal of Ophthalmology, 103(10):167–175.
Yosinski, J., Clune, J., Bengio, Y., and Lipson, H. (2014). How transferable are features in deep neural networks? In Advances in Neural Information Processing Systems, pages 3320–3328. DOI: 10.48550/arXiv.1411.1792
American Academy of Ophthalmology (2020). Diabetic retinopathy preferred practice pattern. [link]. Accessed 4 May 2025.
Asif, M., Ur Rehman, F., Rashid, Z., Hussain, A., Mirza, A., and Qureshi, W. S. (2025). An insight on the timely diagnosis of diabetic retinopathy using traditional and ai-driven approaches. IEEE Access, 13:116869–116886. DOI: 10.1109/ACCESS.2025.3583647
Esteva, A., Kuprel, B., Novoa, R. A., and et al. (2017). Dermatologist-level classification of skin cancer with deep neural networks. Nature, 542:115–118. DOI: 10.1038/nature21056
Goodfellow, I., Bengio, Y., and Courville, A. (2016). Deep Learning. MIT Press.
Gulshan, V., Peng, L., Coram, M., and et al. (2016). Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA, 316(22):2402–2410. DOI: 10.1001/jama.2016.17216
Haque, M., Winston, K., and et al. (2022). Densenet-169 ensemble for robust diabetic retinopathy detection. In IEEE International Conference on Healthcare Informatics.
Huang, G., Liu, Z., van der Maaten, L., and Weinberger, K. Q. (2017). Densely connected convolutional networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR). DOI: 10.1109/CVPR.2017.243
International Diabetes Federation (2023). Idf diabetes atlas, 10th edition. [link]. Accessed 4 May 2025.
Kaggle (2015). Diabetic retinopathy detection dataset. [link]. Accessed 3 May 2025.
Krause, J., Gulshan, V., Rahimy, E., Widner, K., and et al. (2018). Grader variability and the importance of reference standards for evaluating machine-learning models for diabetic retinopathy. Ophthalmology, 125(8):1264–1272. DOI: 10.1016/j.ophtha.2018.01.034
Krizhevsky, A., Sutskever, I., and Hinton, G. E. (2012). Imagenet classification with deep convolutional neural networks. In Advances in Neural Information Processing Systems, pages 1097–1105.
Li, J., Wang, Y., and Chen, R. (2024). On-device diabetic retinopathy screening with a mobile efficientnet model. IEEE Journal of Biomedical and Health Informatics, 28(4):1553–1562.
Lin, Q., Zhang, X., and Chen, Y. (2020). Diabetic retinopathy grading with efficientnet and enhanced pre-processing. Biomedical Signal Processing and Control, 62:102123.
Lin, T.-Y., Goyal, P., Girshick, R., He, K., and Dollár, P. (2017). Focal loss for dense object detection. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pages 2980–2988. DOI: 10.1109/ICCV.2017.324
Mehboob, M., Raza, M. A., and Shahzad, H. (2023). Diabetic retinopathy detection using densenet-169 with transfer learning. In International Conference on Intelligent Systems, pages 343–350.
Nair, R., Phene, S., and Krause, J. (2023). Deep learning for diabetic retinopathy screening: A systematic review and meta-analysis. Ophthalmology Retina, 7(2):97–109.
Rajpurkar, P., Irvin, J., Zhu, K., and et al. (2017). Chexnet: Radiologist-level pneumonia detection on chest x-rays with deep learning. arXiv preprint arXiv:1711.05225. DOI: 10.48550/arXiv.1711.05225
Siahkali, M. and Bagherian, L. (2023). A lightweight multi-stage framework for smartphone-based diabetic retinopathy screening. Frontiers in Medicine, 10:1184756.
Silva, R., Pereira, P., and Mendonça, L. (2021). Efficientnet-b3 for highly accurate diabetic retinopathy grading. In Medical Image Computing and Computer-Assisted Intervention (MICCAI).
Tan, M. and Le, Q. (2019). Efficientnet: Rethinking model scaling for convolutional neural networks. In Proceedings of the 36th International Conference on Machine Learning (ICML). [link]
Ting, D. S. W., Pasquale, L. R., Peng, L., and et al. (2019a). Artificial intelligence and deep learning in ophthalmology. Nature Medicine, 25:14–24. DOI: 10.1136/bjophthalmol-2018-313173
Ting, D. S. W., Pasquale, L. R., Peng, L., and et al. (2019b). Artificial intelligence for diabetic retinopathy screening: A meta-analysis. British Journal of Ophthalmology, 103(10):167–175.
Yosinski, J., Clune, J., Bengio, Y., and Lipson, H. (2014). How transferable are features in deep neural networks? In Advances in Neural Information Processing Systems, pages 3320–3328. DOI: 10.48550/arXiv.1411.1792
Publicado
10/09/2025
Como Citar
OLIVEIRA, Vitor A.; SANTOS, Helida; LUCCA, Giancarlo; OLIVEIRA, Lizandro; YAMIN, Adenauer; REISER, Renata.
Towards Convolutional Neural Networks for Diabetic-Retinopathy Detection. In: WORKSHOP-ESCOLA DE INFORMÁTICA TEÓRICA (WEIT), 8. , 2025, Ponta Grossa/PR.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 111-120.
