Classifying COVID-19 positive X-ray using deep learning models
Iago Richard Rodrigues, Guto Leoni Santos, Djamel Sadok, Patrícia Takako Endo
Abstract
COVID-19 is a pandemic characterized by uncertainty not only in transmission and pathogenicity, but also in disease-specific control options. Despite many governmental measures, the disease is spreading and in many countries, the public health system is close to be collapsed. Alternative techniques should be taken in order to minimize the COVID-19 negative impacts on the society. This work presents preliminary results of deep learning models to classify COVID-19 positive based on X-ray images. We provide binary classification (COVID-19 vs healhty, and COVID-19 vs pneumonia) and also multiclass (COVID-19 vs pneumonia vs healhty) regarding five metrics: accuracy, percision, sensibility, specificity and F1-score. Results show that VGG models present the best results, achiving 98.81% of precision in binary classification, and 91.68% in multiclass classification.