Automatic detection and diagnosis of retinal pathologies using a Transformer-based architecture

• Thalisson J. C. Silva UFMA
• Saulo E. R. Fernandes UFMA
• João D. S. de Almeida UFMA
• Darlan B. P. Quintanilha UFMA
• Geraldo Braz Junior UFMA

DOI: https://doi.org/10.5753/sbcas.2024.2134

Abstract

Globally, more than 2.2 billion people are visually impaired, with around one billion of these cases being preventable. Early detection of eye diseases is crucial to prevent the progression of irreversible conditions such as blindness. Therefore, this study presents a new method for detecting multiple eye pathologies in fundus images, using a neural network architecture based on transformers, called Query2Label. The experiments were carried out on the RFMiD public dataset, revealing promising results, with an average accuracy of 99.8% in the “D. Risk” category. Compared to the state of the art, the method effectively detected the “ODP” class. It surpassed accuracy in other specific categories, such as “CSR”, “LS”, highlighting its feasibility and effectiveness in classifying ophthalmic pathologies.

References

Araújo, J., de Paiva, A., de Almeida, J., Neto, O. P. S., de Sousa, J., Silva, A., and Júnior, G. B. (2017). Diagnóstico de glaucoma em imagens de fundo de olho utilizando os Índices de diversidade de shannon e mcintosh. In Anais do XVII Workshop de Informática Médica, Porto Alegre, RS, Brasil. SBC.

Ceschini, L., Policarpo, L., Rodrigues, V., Righi, R., and Ramos, G. (2022). Otimizando o diagnóstico automatizado de glaucoma a partir de imagens de fundo de olho. In Anais da XXII Escola Regional de Alto Desempenho da Região Sul, pages 9–12, Porto Alegre, RS, Brasil. SBC.

Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., and Fei-Fei, L. (2009). Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee.

Dominik Müller, I. S.-R. and Kramer, F. (2021). Multi-disease detection in retinal imaging based on ensembling heterogeneous deep learning models.

Farnell, D. J., Hatfield, F. N., Knox, P., Reakes, M., Spencer, S., Parry, D., and Harding, S. P. (2008). Enhancement of blood vessels in digital fundus photographs via the application of multiscale line operators. Journal of the Franklin institute, 345(7):748–765.

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, pages 4700–4708.

IBGE (2010). Censo demográfico.

Kaur, P. and Gosain, A. (2018). Issues and challenges of class imbalance problem in classification. International Journal of Information Technology, pages 1–7.

Liu, S., Zhang, L., Yang, X., Su, H., and Zhu, J. (2021). Query2label: A simple transformer way to multi-label classification. arXiv preprint arXiv:2107.10834.

Mittal, K. and Rajam, V. (2020). Computerized retinal image analysis-a survey. Multimedia Tools and Applications, 79(31):22389–22421.

Oh, Y.-t. and Park, H. (2022). End-to-end two-branch classifier for retinal imaging analysis. In 2022 International Conference on Electronics, Information, and Communication (ICEIC), pages 1–3. IEEE.

Organization, W. H. et al. (2019). World report on vision.

Pachade, S., Porwal, P., Thulkar, D., Kokare, M., Deshmukh, G., Sahasrabuddhe, V., Giancardo, L., Quellec, G., and Mériaudeau, F. (2021). Retinal fundus multi-disease image dataset (rfmid): a dataset for multi-disease detection research. Data, 6(2):14.

Ridnik, T., Lawen, H., Noy, A., Ben Baruch, E., Sharir, G., and Friedman, I. (2021). Tresnet: High performance gpu-dedicated architecture. In proceedings of the IEEE/CVF winter conference on applications of computer vision, pages 1400–1409.

Rodriguez, M., AlMarzouqi, H., and Liatsis, P. (2022). Multi-label retinal disease classification using transformers. arXiv preprint arXiv:2207.02335.

Silva, M., Pessoa, A., de Almeida, J., Júnior, G. B., and de Paiva, A. (2018). Diagnóstico do glaucoma em imagens de retinografia usando variantes de padrões locais binários. In Anais do XVIII Simpósio Brasileiro de Computação Aplicada à Saúde, Porto Alegre, RS, Brasil. SBC.

Umbelino, C. C. and Ávila, M. P. (2023). As condições de saúde ocular no brasil. São Paulo: Conselho Brasileiro de Oftalmologia.

Zhang, M.-L., Li, Y.-K., Liu, X.-Y., and Geng, X. (2018). Binary relevance for multi-label learning: an overview. Frontiers of Computer Science, 12(2):191–202.