Automatic Detection of Covid-19 Infection using Chest X-Ray Images through Transfer Learning

Abstract

The World Health Organization has called the new corona virus (COVID-19) a pandemic. It has infected more than 1 million people and killed more than 50,000. A COVID-19 infection can lead to pneumonia, which can be found with a chest X-ray and should be treated properly. In this paper, we show how chest X-rays can be used to automatically find COVID-19 infections. For this study, 194 X-ray images of people with coronavirus and 194 X-ray images of healthy people were put together to make two sets of data. We use the idea of transfer learning for this task because there aren't many images of COVID-19 patients that are available to the public. We use different architectures of convolutional neural networks (CNNs) trained on ImageNet and change them to work as feature extractors for the X-ray images. The CNNs are then combined with consolidated machine learning methods like k-Nearest Neighbor, Bayes, Random Forest, multilayer perceptron (MLP), and support vector machine (SVM). The results show that for one of the datasets, the best extractor-classifier pair is the Mobile Net architecture with the SVM classifier using a linear kernel, which gets an accuracy and F1-score of 98.5%. For the other set of data, the best combination is DenseNet201 and MLP, with a 95.6% accuracy and F1-score. So, the proposed method works well for spotting COVID-19 in X-ray images.