Exploring Classification Models for COVID-19 Novel Coronavirus Disease

ABSTRACT

Coronavirus disease (COVID-19) is defined as a disease caused by severe acute respiratory syndrome coronavirus (SARS-CoV-2). Coronavirus has been declared a global pandemic in March 2020 by World Health Organization (WHO). The spread of coronavirus can be limited by early detection of the disease, for which RT-PCR and imaging studies are being used. The chest x-rays taken upon the arrival of the patient in the hospital can be used as the input source for early detection of disease with machine and deep learning algorithms. Even though, this is the most regular and supreme imaging modality, chest radiography is question to notable intra-observer variability and has almost minor sensitivity for major clinical findings. With advances in deep learning, convolutional neural networks (CNNs) not only improved chest radiograph evaluation but are also capable of staging radiologist-level performance. In this paper, we are applying CNN with PyTorch to train ResNet18 model as PyTorch is a lower-level application programming interface concentrated on direct work with the use of array expressions. This model implementation will be beneficial in rural areas where RT-PCR test results are delayed due to the geographical location, but portable chest x-ray machines are already installed. Here, we have collated different deep learning-based classification models at hand for identification of novel coronavirus. The results are present in tabular form. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.