Application of Deep Learning Techniques for Diabetic Retinopathy Classification

• Arthur Carneiro Kaizer PUC Minas
• Alexei Manso Correa Machado PUC Minas / UFMG

DOI: https://doi.org/10.5753/sbcas_estendido.2025.6954

Abstract

Diabetic retinopathy (DR) is a degenerative condition of the retina caused by diabetes. Early diagnosis is crucial to prevent disease progression and preserve vision. However, traditional diagnostic methods require a thorough analysis of retinal images by experts, making the process time-consuming and subjective. This study proposes a deep learning model for the automated classification of DR, based on the analysis of different convolutional neural networks and image processing techniques. The results indicate that the combination of preprocessing using CLAHE and the ResNet50 with the ADAMW optimizer achieved the best performance, with an accuracy of 0.83, ROC-AUC of 0.87, and a Kappa coefficient of 0.85.

References

Huang, Y., Lin, L., Cheng, P., Lyu, J., Tam, R., and Tang, X. (2023). Identifying the key components in resnet-50 for diabetic retinopathy grading from fundus images: a systematic investigation. Diagnostics, 13(10):1664.

Kalyani, G., Janakiramaiah, B., Karuna, A., and Prasad, L. N. (2023). Diabetic retinopathy detection and classification using capsule networks. Complex & Intelligent Systems, 9(3):2651–2664.

Karthik, Maggie, S. D. (2019). Aptos 2019 blindness detection.

Lee, C.-H. and Ke, Y.-H. (2021). Fundus images classification for diabetic retinopathy using deep learning. In Proceedings of the 13th International Conference on Computer Modeling and Simulation, ICCMS ’21, page 264–270, New York, NY, USA. Association for Computing Machinery.

Momeni Pour, A., Seyedarabi, H., Abbasi Jahromi, S. H., and Javadzadeh, A. (2020). Automatic detection and monitoring of diabetic retinopathy using efficient convolutional neural networks and contrast limited adaptive histogram equalization. IEEE Access, 8:136668–136673.

Patra, P. and Singh, T. (2022). Diabetic retinopathy detection using an improved resnet50-inceptionv3 structure. In 2022 13th International Conference on Computing Communication and Networking Technologies (ICCCNT), pages 1–6. IEEE.

Porwal, P., Pachade, S., Kamble, R., Kokare, M., Deshmukh, G., Sahasrabuddhe, V., and Meriaudeau, F. (2018). Indian diabetic retinopathy image dataset (idrid).

Saeed, F., Hussain, M., and Aboalsamh, H. A. (2021). Automatic diabetic retinopathy diagnosis using adaptive fine-tuned convolutional neural network. IEEE Access, 9:41344–41359.

Sudarmadji, P. W., Deviani Pakan, P., and Yefrenes Dillak, R. (2020). Diabetic retinopathy stages classification using improved deep learning. In 2020 International Conference on Informatics, Multimedia, Cyber and Information System (ICIMCIS), pages 104–109.