CNNs, Vision Transformers, and Ensembles for Glaucoma Classification: A Comparative Analysis
Resumo
O glaucoma é uma das principais causas de cegueira irreversível no mundo, caracterizado pelo dano progressivo ao nervo óptico. O diagnóstico precoce é essencial para minimizar os impactos da doença. Neste trabalho, arquiteturas neurais são exploradas para classificar imagens de fundo de olho da base SMDG, comparando CNNs pré-treinadas e transformers, além de estratégias de ensemble por votação. ResNet-50 e VGG-19 obtiveram desempenho consistente, mas menor sensibilidade. Entre os transformers, MaxViT alcançou o maior recall, favorecendo a detecção de positivos, enquanto ViT apresentou equilíbrio entre acurácia e sensibilidade. A votação por média superou modelos individuais, atingindo 89,45% de acurácia.
Referências
Ashtari-Majlan, M., Dehshibi, M. M., and Masip, D. (2023). Deep learning and computer vision for glaucoma detection: A review. arXiv preprint arXiv:2307.16528.
Cardoso, A. P. (2024). Glaucoma é a principal causa de cegueira irreversível no mundo; veja como se prevenir. [link]. Acesso em: 11 ago. 2026.
Chan, Y. M., Ng, E. Y. K., Jahmunah, V., Koh, J. E. W., Lih, O. S., Leon, L. Y. W., and Acharya, U. R. (2019). Automated detection of glaucoma using optical coherence tomography angiogram images. Computers in Biology and Medicine, 115:103483.
Chen, X., Xu, Y., Yan, S., Wong, D. W. K., Wong, T. Y., and Liu, J. (2015). Automatic feature learning for glaucoma detection based on deep learning. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18, pages 669–677. Springer.
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., and Houlsby, N. (2021). An image is worth 16x16 words: Transformers for image recognition at scale.
Filho, M. (2023). Precisão, recall e f1 score em machine learning. [link]. Acesso em: 04 ago. 2023.
Hossin, M. and Sulaiman, M. N. (2015). A review on evaluation metrics for data classification evaluations. International journal of data mining & knowledge management process, 5(2):1.
Hwang, E. E., Chen, D., Han, Y., Jia, L., and Shan, J. (2023). Multi-dataset comparison of vision transformers and convolutional neural networks for detecting glaucomatous optic neuropathy from fundus photographs. Bioengineering, 10(11).
Kiefer, R. (2023). Smdg, a standardized fundus glaucoma dataset. [link]. DOI: 10.34740/KAGGLE/DS/2329670. Publicado por Kaggle.
Kiefer, R., Abid, M., Steen, J., Ardali, M. R., and Amjadian, E. (2023). A catalog of public glaucoma datasets for machine learning applications: A detailed description and analysis of public glaucoma datasets available to machine learning engineers tackling glaucoma-related problems using retinal fundus images and oct images. In Proceedings of the 2023 7th International Conference on Information System and Data Mining, ICISDM ’23, page 24–31, New York, NY, USA. Association for Computing Machinery.
Kundu, N. (2023). Exploring resnet50: An in-depth look at the model architecture and code implementation.
Oh, S., Park, Y., Cho, K. J., and Kim, S. J. (2021). Explainable machine learning model for glaucoma diagnosis and its interpretation. Diagnostics, 11(3).
Prevent Senior (2023). Estima-se que 64 milhões de pessoas tenham glaucoma no mundo. será que você é uma delas? [link]. Publicado originalmente em preventsenior.com.br. Acesso em: 11 ago. 2026.
Rakotomamonjy, A. (2004). Optimizing area under roc curve with svms. In Hernández-Orallo, J., Ferri, C., Lachiche, N., and Flach, P. A., editors, Proceedings of the 1st International Workshop on ROC Analysis in Artificial Intelligence (ROCAI-2004), pages 71–80, Valencia, Spain.
Rodrigues, G. B., Abe, R. Y., Zangalli, C., Sodre, S. L., Donini, F. A., Costa, D. C., Leite, A., Felix, J. P., Torigoe, M., Diniz-Filho, A., et al. (2016). Neovascular glaucoma: a review. International journal of retina and vitreous, 2:1–10.
Saxena, A., Vyas, A., Parashar, L., and Singh, U. (2020). A glaucoma detection using convolutional neural network. In 2020 international conference on electronics and sustainable communication systems (ICESC), pages 815–820. IEEE.
Serte, S. and Serener, A. (2019). A generalized deep learning model for glaucoma detection. In 2019 3rd International symposium on multidisciplinary studies and innovative technologies (ISMSIT), pages 1–5. IEEE.
Simonyan, K. and Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.
Steiner, A., Kolesnikov, A., Zhai, X., Wightman, R., Uszkoreit, J., and Beyer, L. (2022). How to train your vit? data, augmentation, and regularization in vision transformers.
Tan, M. and Le, Q. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 6105–6114. PMLR.
Tham, Y.-C., Li, X., Wong, T. Y., Quigley, H. A., Aung, T., and Cheng, C.-Y. (2014). Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology, 121(11):2081–2090.
Tohye, T. G., Qin, Z., Al-Antari, M. A., Ukwuoma, C. C., Lonseko, Z. M., and Gu, Y. H. (2024). Ca-vit: Contour-guided and augmented vision transformers to enhance glaucoma classification using fundus images. Bioengineering, 11(9):887.
U, R., Fujita, H., Bhandary, S., Gudigar, A., Tan, J. H., and Acharya, U. (2018). Deep convolution neural network for accurate diagnosis of glaucoma using digital fundus images. Information Sciences, 441.
Velpula, V. K., Sharma, D., Sharma, L. D., Roy, A., Bhuyan, M. K., Alfarhood, S., and Safran, M. (2024). Glaucoma detection with explainable ai using convolutional neural networks based feature extraction and machine learning classifiers. IET Image Processing, 18(13):3827–3853.
Velpula, V. K. and Sharma, L. D. (2023). Multi-stage glaucoma classification using pre-trained convolutional neural networks and voting-based classifier fusion. Frontiers in Physiology, 14.
Wang, Z., Xue, C. C., Li, Y., Wu, Y., Pan, Z., Li, F., Cheung, C. Y., Ohno-Matsui, K., Friedman, D. S., Garway-Heath, D., et al. (2026). Global glaucoma prevalence: burden and projection to 2060. American journal of ophthalmology, 283:324–335.
Wu, J., Yu, S., Chen, W., Ma, K., Fu, R., Liu, H., Di, X., and Zheng, Y. (2020). Leveraging undiagnosed data for glaucoma classification with teacher-student learning.
Yurdakul, M., Uyar, K., and Tasdemir, S. (2025). Maxglavit: A novel lightweight vision transformer-based approach for early diagnosis of glaucoma stages from fundus images. arXiv preprint arXiv:2502.17154.
Zhen, Y., Wang, L., Liu, H., Zhang, J., and Pu, J. (2018). Performance assessment of the deep learning technologies in grading glaucoma severity. arXiv preprint arXiv:1810.13376.
Cardoso, A. P. (2024). Glaucoma é a principal causa de cegueira irreversível no mundo; veja como se prevenir. [link]. Acesso em: 11 ago. 2026.
Chan, Y. M., Ng, E. Y. K., Jahmunah, V., Koh, J. E. W., Lih, O. S., Leon, L. Y. W., and Acharya, U. R. (2019). Automated detection of glaucoma using optical coherence tomography angiogram images. Computers in Biology and Medicine, 115:103483.
Chen, X., Xu, Y., Yan, S., Wong, D. W. K., Wong, T. Y., and Liu, J. (2015). Automatic feature learning for glaucoma detection based on deep learning. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part III 18, pages 669–677. Springer.
Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., and Houlsby, N. (2021). An image is worth 16x16 words: Transformers for image recognition at scale.
Filho, M. (2023). Precisão, recall e f1 score em machine learning. [link]. Acesso em: 04 ago. 2023.
Hossin, M. and Sulaiman, M. N. (2015). A review on evaluation metrics for data classification evaluations. International journal of data mining & knowledge management process, 5(2):1.
Hwang, E. E., Chen, D., Han, Y., Jia, L., and Shan, J. (2023). Multi-dataset comparison of vision transformers and convolutional neural networks for detecting glaucomatous optic neuropathy from fundus photographs. Bioengineering, 10(11).
Kiefer, R. (2023). Smdg, a standardized fundus glaucoma dataset. [link]. DOI: 10.34740/KAGGLE/DS/2329670. Publicado por Kaggle.
Kiefer, R., Abid, M., Steen, J., Ardali, M. R., and Amjadian, E. (2023). A catalog of public glaucoma datasets for machine learning applications: A detailed description and analysis of public glaucoma datasets available to machine learning engineers tackling glaucoma-related problems using retinal fundus images and oct images. In Proceedings of the 2023 7th International Conference on Information System and Data Mining, ICISDM ’23, page 24–31, New York, NY, USA. Association for Computing Machinery.
Kundu, N. (2023). Exploring resnet50: An in-depth look at the model architecture and code implementation.
Oh, S., Park, Y., Cho, K. J., and Kim, S. J. (2021). Explainable machine learning model for glaucoma diagnosis and its interpretation. Diagnostics, 11(3).
Prevent Senior (2023). Estima-se que 64 milhões de pessoas tenham glaucoma no mundo. será que você é uma delas? [link]. Publicado originalmente em preventsenior.com.br. Acesso em: 11 ago. 2026.
Rakotomamonjy, A. (2004). Optimizing area under roc curve with svms. In Hernández-Orallo, J., Ferri, C., Lachiche, N., and Flach, P. A., editors, Proceedings of the 1st International Workshop on ROC Analysis in Artificial Intelligence (ROCAI-2004), pages 71–80, Valencia, Spain.
Rodrigues, G. B., Abe, R. Y., Zangalli, C., Sodre, S. L., Donini, F. A., Costa, D. C., Leite, A., Felix, J. P., Torigoe, M., Diniz-Filho, A., et al. (2016). Neovascular glaucoma: a review. International journal of retina and vitreous, 2:1–10.
Saxena, A., Vyas, A., Parashar, L., and Singh, U. (2020). A glaucoma detection using convolutional neural network. In 2020 international conference on electronics and sustainable communication systems (ICESC), pages 815–820. IEEE.
Serte, S. and Serener, A. (2019). A generalized deep learning model for glaucoma detection. In 2019 3rd International symposium on multidisciplinary studies and innovative technologies (ISMSIT), pages 1–5. IEEE.
Simonyan, K. and Zisserman, A. (2014). Very deep convolutional networks for large-scale image recognition. arXiv preprint arXiv:1409.1556.
Steiner, A., Kolesnikov, A., Zhai, X., Wightman, R., Uszkoreit, J., and Beyer, L. (2022). How to train your vit? data, augmentation, and regularization in vision transformers.
Tan, M. and Le, Q. (2019). EfficientNet: Rethinking model scaling for convolutional neural networks. In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 6105–6114. PMLR.
Tham, Y.-C., Li, X., Wong, T. Y., Quigley, H. A., Aung, T., and Cheng, C.-Y. (2014). Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology, 121(11):2081–2090.
Tohye, T. G., Qin, Z., Al-Antari, M. A., Ukwuoma, C. C., Lonseko, Z. M., and Gu, Y. H. (2024). Ca-vit: Contour-guided and augmented vision transformers to enhance glaucoma classification using fundus images. Bioengineering, 11(9):887.
U, R., Fujita, H., Bhandary, S., Gudigar, A., Tan, J. H., and Acharya, U. (2018). Deep convolution neural network for accurate diagnosis of glaucoma using digital fundus images. Information Sciences, 441.
Velpula, V. K., Sharma, D., Sharma, L. D., Roy, A., Bhuyan, M. K., Alfarhood, S., and Safran, M. (2024). Glaucoma detection with explainable ai using convolutional neural networks based feature extraction and machine learning classifiers. IET Image Processing, 18(13):3827–3853.
Velpula, V. K. and Sharma, L. D. (2023). Multi-stage glaucoma classification using pre-trained convolutional neural networks and voting-based classifier fusion. Frontiers in Physiology, 14.
Wang, Z., Xue, C. C., Li, Y., Wu, Y., Pan, Z., Li, F., Cheung, C. Y., Ohno-Matsui, K., Friedman, D. S., Garway-Heath, D., et al. (2026). Global glaucoma prevalence: burden and projection to 2060. American journal of ophthalmology, 283:324–335.
Wu, J., Yu, S., Chen, W., Ma, K., Fu, R., Liu, H., Di, X., and Zheng, Y. (2020). Leveraging undiagnosed data for glaucoma classification with teacher-student learning.
Yurdakul, M., Uyar, K., and Tasdemir, S. (2025). Maxglavit: A novel lightweight vision transformer-based approach for early diagnosis of glaucoma stages from fundus images. arXiv preprint arXiv:2502.17154.
Zhen, Y., Wang, L., Liu, H., Zhang, J., and Pu, J. (2018). Performance assessment of the deep learning technologies in grading glaucoma severity. arXiv preprint arXiv:1810.13376.
Publicado
19/10/2026
Como Citar
CELESTINO, Arthur; SILVA, Guilherme; PINTO, Luan; GETRUDES, Jadson Castro; SILVA, Pedro.
CNNs, Vision Transformers, and Ensembles for Glaucoma Classification: A Comparative Analysis. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 23. , 2026, Cuiabá/MT.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 933-944.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2026.30652.
