Towards a Classification Model using CNN and Wavelets applied to COVID-19 CT images

• Autor(a) principal: Sousa, Pedro Moises de
• Data de Publicação: 2022
• Outros Autores: Carneiro, Pedro Cunha, Pereira, Gabrielle Macedo, Oliveira, Mariane Modesto, Costa Junior, Carlos Aberto da, Moura, Luis Vinicius de, Mattjie, Christian, Silva, Ana Maria da, Macedo, Túlio Augusto Alves, Patrocinio, Ana Claudia
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Research, Society and Development
• Texto Completo: https://rsdjournal.org/index.php/rsd/article/view/27919
• topic: CT images; Convolutional Neural Networks; COVID-19; Wavelets; WCN-COVID.; Redes Neuronales Convolucionales; COVID-19; Wavelets; Imágenes de TC; WCN-COVID.; Redes Neurais Convolucionais; COVID-19; Wavelets; Imagens de TC; WCNN-COVID.

In late 2019, a new type of coronavirus emerged in China and was named SARS-CoV-2. It first impacted the country where it emerged and then spread around the world. SARS-CoV-2 is the cause of COVID-19 disease that leaves characteristic impressions on chest CT images of infected patients. In this article, we propose a classification model, based on CNN and wavelet transform, to classify images of COVID-19 patients. It was named WCNN-COVID. The model was applied and tested in open and private TC image repositories. A total of 25534 images of 200 patients were processed. The confusion matrix was generated by calculating Accuracy (ACC), Sensitivity (Sen) and Specificity (Sp). The Receiver Operating Characteristic (ROC) curve and Area Under the Curve (AUCs) were also plotted and used for evaluation. Metric results were ACC = 0.9950, Sen = 99.16% and Sp = 99.89%.