Detection of respiratory viral infection Through X-Ray Using Convolutional Neural Network

Abstract

Respiratory infections, caused by various viruses and bacteria, pose a persistent and significant global health challenge. The emergence of the COVID-19 pandemic in December 2019 underscored the vulnerabilities inherent in healthcare systems worldwide. Nonetheless, the fight against COVID-19 has unveiled several challenges, including a scarcity of COVID-19 testing kits and the time-consuming nature of the current gold standard diagnostic technique, Reverse Transcription Polymerase Chain Reaction (RT-PCR). The limitations in testing capacity and the need for more efficient diagnostic methods have spurred innovative approaches, such as using deep learning techniques, to address these issues.This study explores the application of deep learning techniques, specifically a convolutional neural network (CNN), for rapid COVID-19 detection using chest X-ray images. A custom deep CNN model is designed and evaluated for ternary classification of COVID-19, viral pneumonia, and normal cases.

Experimental results demonstrate an average classification accuracy of 91.59% and an F1-score of 92.00%, indicating strong discriminative capability. These findings suggest that the proposed model holds promise as an efficient adjunctive tool for COVID-19 screening, with potential implications for pandemic response, patient triage, and resource allocation. Limitations of the current study, including dataset constraints and the need for prospective clinical validation, are also discussed, along with directions for future work involving larger and more diverse imaging datasets.