COVID-19 classification in X-ray chest images using a new convolutional neural network: CNN-COVID
Pedro Moisés de Sousa, Pedro Cunha Carneiro, Mariane Modesto Oliveira, Gabrielle Macedo Pereira, Carlos Alberto da Costa, Luís Vinícius de Moura, Christian Mattjie, Ana Maria Marques da Silva, Ana Cláudia Patrocínio
Abstract
COVID-19 causes lung inflammation and lesions, and chest X-ray and computed tomography images are remarkably suitable for differentiating the new disease from patients with other lung diseases. In this paper, we propose a computer model to classify X-ray images of patients diagnosed with COVID-19. Chest X-ray exams were chosen over computed tomography scans because they are low cost, results are quickly obtained, and X-ray equipment is readily available. A new CNN network, called CNN-COVID, has been developed to classify X-ray patient’s images. Images from two different datasets were used. The images of Dataset I is originated from the COVID-19 image data collection and the ChestXray14 repository, and the images of Dataset II belong to the BIMCV COVID-19+ repository. To assess the accuracy of the network, 10 training and testing sessions were performed in both datasets. A confusion matrix was generated to evaluate the model’s performance and calculate the following metrics: accuracy (ACC), sensitivity (SE), and specificity (SP). In addition, Receiver Operating Characteristic (ROC) curves and Areas Under the Curve (AUCs) were also considered. After running 10 tests, the average accuracy for Dataset I and Dataset II was 0.9787 and 0.9839, respectively. Since the weights of the best test results were applied in the validation, it was obtained the accuracy of 0.9722 for Dataset I and 0.9884 for Dataset II. The results showed that the CNN-COVID is a promising tool to help physicians classify chest images with pneumonia, considering pneumonia caused by COVID-19 and pneumonia due to other causes.