SARS-CoV-2 Detection Using Chest X-Ray Images with Deep Learning Methods

Abstract

With the coronavirus invading the world from China since December 2019, it has led every country into crisis. The World Health Organization (WHO) declared the coronavirus a pandemic on March 11, 2020. The search for solutions began all over the world. One of the solutions is applying artificial neural networks for classification methods on chest X-ray images. Chest X-ray (CXR) scan images can be considered a confirmatory approach as they are quick to obtain and easily accessible. When these images are used, transfer learning from deep learning methods is the most preferred method to detect infected patients. Three different datasets, varying in different sample sizes, were used for training our models and further detailed analysis. The outputs of the results are measured by looking at the F1 score and accuracy. With the comparative performance analysis, it was seen that the InceptionV3 and Xception models had the highest overall accuracy and F1 scores than the other models for our datasets.