Uma Arquitetura de Rede Neural Convolucional Simplificada para Reconhecimento da COVID-19 em Imagens de Raios-X

• Fabio J. Barbosa IF Sudeste MG
• Alessandra M. Coelho IF Sudeste MG
• Matheus F. O. Baffa USP

DOI: https://doi.org/10.5753/ercas.2021.17428

Abstract

COVID-19 is an emerging problem that has already caused great damage to society. An accurate diagnosis helps in providing more appropriate treatment for the patient. In the Computer Vision literature, many works implement robust deep learning architectures to create COVID classification models, this work aims to simplify a Convolutional Neural Network in order to reduce the computational cost, without losing performance in detecting COVID-19 in images of X-rays. With three layers of convolution, the work developed obtained an accuracy of 99.36% accuracy.

References

Garcia, L. P. and Duarte, E. (2020). Intervenções não farmacológicas para o enfrentamento à epidemia da COVID-19 no Brasil. Epidemiologia e Serviços de Saúde, 29.

Lima, C. M. A. d. O. (2020). Information about the new coronavirus disease (COVID-19). Radiologia Brasileira, 53:V – VI.

Martins, J. V. G., Gregório, M. P., Baffa, M. d. F. O., and Coelho, A. M. (2021). Classicação da covid-19 em radiograas do tórax utilizando redes neurais profundas e padrões binários locais. Journal of Health Informatics, 12:387–392.

Meirelles, G. d. S. P. (2020). COVID-19: a brief update for radiologists. Radiologia Brasileira, 53:320–328.

Patel, P. (2021). Chest X-ray (Covid-19 & Pneumonia). Disponível em: https://www.kaggle.com/prashant268/chest-xray-covid19-pneumonia. Acesso em: 01 jan. 2021.

Rodrigues, L., Rodrigues, L., da Silva, D., and Mari, J. F. (2020). Evaluating convolutional neural networks for covid-19 classication in chest x-ray images. In Anais do XVI Workshop de Visão Computacional, pages 52–57. SBC.