Abstract: December 2019 marked the emergence of a novel coronavirus in China that rapidly spread worldwide. Early detection of COVID-19 is crucial to limit transmission. The standard diagnostic method, reverse transcription-polymerase chain reaction (RT-PCR), is accurate but costly and time-consuming. Recent studies indicate that deep learning models can reliably detect COVID-19 from chest X-ray (CXR) images, offering a faster and more cost-effective alternative. In this paper, we propose COVID-MRSNet, a deep Convolutional Neural Network-based ResNet architecture for classifying COVID-19, viral pneumonia, and normal cases using CXR images. The model was trained and evaluated on a public dataset of 2,905 images, including 219 COVID- 19, 1,345 viral pneumonia, and 1,341 normal cases. After 50 epochs, COVID-MRSNet achieved an average accuracy of 98.34%, recall of 98.67%, precision of 98.55%, and F1-score of 98.61%. The results demonstrate superior multi-class classification performance compared to existing CNN-based methods.

Keywords: Artificial intelligence, deep learning, Convolutional neural models, Chest radiography, COVID-19 detection, Pulmonary disease classification, Medical image analysis

Pages: 66-88