Keywords
segmentation
deep neural networks
How to Cite
Abstract
Diabetic retinopathy is one of the main causes of vision loss and is diagnosed by investigating the presence of retinal lesions, such as hard exudates, soft exudates, microaneurysms and hemorrhages. Identification and treatment of these lesions in the early stages can prevent vision loss. The diagnosis is complex and requires specialized professionals and infrastructure to meet the growing demand. In this context, this work proposes a method based on deep neural networks to segment lesions associated with Diabetic Retinopathy. The proposed method demonstrated to be effective in identifying lesions, obtaining accuracies of 99.91%, 99.96%, 99.51% and 99.98% in the test stage, for hard exudates, soft exudates, hemorrhages and microaneurysms, respectively.
References
BUSLAEV, A. et al. Albumentations: fast and flexible image augmentations. In: COMPUTER VISION AND PATTERN RECOGNITION, 2018, Cornel. 2018. Dis-ponível em: https://arxiv.org/abs/1809.06839. Acesso em: 22 nov. 2024.
INTERNATIONAL DIABETES FEDERATION. Facts & figures, 2024. Disponível em: https://idf.org/about-diabetes/diabetes-facts-figures/. Acesso em: 14 out. 2024.
LI, T. et al. Diagnostic assessment of deep learning algorithms for diabetic reti-nopathy screening. Information Sciences: an international journal, v. 501, n. C, p. 511-522, 2019. Disponível em: https://dl.acm.org/doi/10.1016/j.ins.2019.06.011. Acesso em: 14 out. 2024.
OLIVEIRA, M, V. et al. PPM-UNET: uma rede neural convolucional para a seg-mentação de rins em imagens de TC. In: SIMPÓSIO BRASILEIRO DE COMPU-TAÇÃO APLICADA À SAÚDE, 22., 2022, Teresina. Anais eletrônicos[...]. Teresi-na: Sociedade Brasileira de Computação, 2022. Disponível em: https://sol.sbc.org.br/index.php/sbcas/article/view/21638. Acesso em: 09 ago. 2024.
RONNEBERGER, O.; FISCHER, P.; BROX, T. U-Net: Convolutional networks for biomedical image segmentation. In: MEDICAL IMAGE COMPUTING AND COM-PUTER-ASSISTED INTERVENTION (MICCAI 2015), 2015, Cornell. 2015. Dispo-nível em: https://arxiv.org/abs/1505.04597v1. Acesso em: 09 ago. 2024.
SANTOS, C. et al. A new approach for detecting fundus lesions using image pro-cessing and deep neural network architecture based on YOLO Model. Sensors, v. 17, n. 22, p. 1-37, 2022. Disponível em: https://www.mdpi.com/1424-8220/22/17/6441. Acesso em: 09 ago. 2024.
SILVA, W. A. Segmentação do disco óptico em imagens retinoscópicas utili-zando a Rede Neural Convolucional U-Net. 2019. Monografia (Trabalho de Conclusão de Curso) - Universidade Federal do Piauí, Piauí, 2019. Disponível em: https://bit.ly/46JHzqf. Acesso em: 09 ago. 2024.
This work is licensed under a Creative Commons Attribution 4.0 International License.
Copyright (c) 2024 Boletim Técnico-Científico