A Teleophthalmology Screening Platform for Diabetic Retinopathy with Lesion-Based Evidence and Ordinal Grading

• Laura Bernardes IFFar
• Artur Heckler IFFar
• Alejandro Pereira UFPel
• Marcelo Dias IFFar
• Marilton Aguiar IFFar
• Daniel Welfer UFSM
• Carlos Santos IFFar

DOI: https://doi.org/10.5753/sbcas.2026.20164

Resumo

Diabetic retinopathy screening in public health settings demands scalable workflows, traceability, and clinically meaningful visual evidence. This paper presents an open teleophthalmology platform that couples multi-label lesion segmentation with ordinal DR grading (0–4) and exposes the inference trail through an interactive interface. Using the public IDRiD dataset, we evaluated segmentation performance across U-Net++, Attention U-Net, and Swin-Unet, and grading performance with EfficientNetV2-S and ConvNeXt-Tiny, including QWK, Macro-F1, Weighted-F1, AUC (OvR), per-class reports, and confusion matrices. Beyond model metrics, the contribution is a healthcare-oriented system design that operationalizes visual evidence, class-wise threshold calibration, and experiment governance, enabling remote triage scenarios and supporting future external and clinical validation in partner ophthalmology clinics.

Referências

Abràmoff, M. D., Lavin, P. T., Birch, M., Shah, N., and Folk, J. C. (2018). Pivotal trial of an autonomous ai-based diagnostic system for detection of diabetic retinopathy in primary care offices. npj Digital Medicine.

Arenas-Cavalli, J. T. et al. (2022). Clinical validation of an artificial intelligence-based diabetic retinopathy screening tool for a national health system. Eye.

Cao, H., Wang, Y., Chen, J., Jiang, D., Zhang, X., Tian, Q., and Wang, M. (2021). Swin-unet: Unet-like pure transformer for medical image segmentation. arXiv preprint arXiv:2105.05537.

Chokshi, T., Cruz, M. J., Ross, J., and Yiu, G. (2024). Advances in teleophthalmology and artificial intelligence for diabetic retinopathy screening: a narrative review. Annals of Eye Science, 9.

Farahat, Z. et al. (2024). Diabetic retinopathy screening through artificial intelligence algorithms: a systematic review. Survey of Ophthalmology, 69(5):707–721.

Gulshan, V., Peng, L., Coram, M., et al. (2016). Development and validation of a deep learning algorithm for detection of diabetic retinopathy in retinal fundus photographs. JAMA.

Liu, Z., Mao, H., Wu, C.-Y., Feichtenhofer, C., Darrell, T., and Xie, S. (2022). A convnet for the 2020s. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR).

Nakayama, L. F., Ribeiro, L. Z., Silva, P. S., et al. (2023). Artificial intelligence for telemedicine diabetic retinopathy screening: a review. Annals of Medicine, 55(2):2258149.

Oktay, O., Schlemper, J., Le Folgoc, L., Lee, M., Heinrich, M., Misawa, K., Mori, K., McDonagh, S., Hammerla, N., Kainz, B., Glocker, B., and Rueckert, D. (2018). Attention U-Net: Learning where to look for the pancreas. arXiv preprint arXiv:1804.03999.

Porwal, P., Pachade, S., Kamble, R., Kokare, M., Deshmukh, G., Sahasrabuddhe, V., and Meriaudeau, F. (2020). Idrid: Diabetic retinopathy – segmentation and grading challenge. Medical Image Analysis, 59:101561.

Tan, M. and Le, Q. V. (2021). Efficientnetv2: Smaller models and faster training. In Proceedings of the 38th International Conference on Machine Learning (ICML).

Ting, D. S. W., Cheung, C. Y.-L., Lim, G., et al. (2017). Development and validation of a deep learning system for diabetic retinopathy and related eye diseases using retinal images from multiethnic populations with diabetes. JAMA.

Zhou, Z., Siddiquee, M. M. R., Tajbakhsh, N., and Liang, J. (2018). UNet++: A nested U-Net architecture for medical image segmentation. arXiv preprint arXiv:1807.10165.