Detection of new coronavirus disease from chest x-ray images using pre-trained convolutional neural networks

Abstract

Purpose: The COVID-19 viruses have affected seriously and caused death especially for older people and patients with chronic diseases. Rapid and accurate early diagnosis has a key role to reduce the mortality and to decrease the economic cost of this pandemic. For this purpose, diagnostic kits, diagnostic aids, and diagnosis using medical imaging methods have been investigated. Although the chest imaging using Computed Tomography (CT) has been accepted as a golden standard among them, there is big challenge to reach this equipment in general. Hence, the diagnosis using more accessible devices like X-rays is very crucial. In this study, it is aimed to show the effectiveness of deep learning models based on parameter transferring in the diagnosis of COVID-19 among normal subjects and patients with Viral Pneumonia. Theory and Methods: Kaggle's chest X-ray images called the “COVID-19 radiography database” were used in this study. Three different ResNet models (ResNet-50, ResNet-101, and ResNet-152) were investigated (a) to discriminate patients with COVID-19 from normal subjects, (b) to discriminate patients with COVID-19 from patients with Viral Pneumonia, and (c) to discriminate patients with COVID-19, patients with Viral Pneumonia, and normal subjects. Results: ResNet-50 model gave the highest performances among these three models. We achieved the accuracy of 99.3% to discriminate COVID-19 and Normal, the accuracy of 99.3% to discriminate COVID-19 and Viral Pneumonia, and the accuracy of 97.3% to discriminate COVID-19, Normal, and Viral Pneumonia. Conclusion: The pre-trained ResNet-50 model has a big potential to detect the patients with COVID-19 quickly and accurately using chest X-Ray images only. Since X-ray devices are relatively more accessible devices in health organizations, the proposed model has a big potential, which may help defeating this pandemic.