Convolutional Neural Network-based Model to Detect COVID-19Using Chest X-ray Images

ABSTRACT

COVID-19, a new coronavirus disease, is caused by Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) and was declared a pandemic by the World Health Organization (WHO). Since its outbreak in December 2019, the total number of affected cases reported worldwide has exceeded 215 million, and reported death cases are 4.49 million as of August 2021. The early detection of the disease is necessary to treat the infected people and control the spreading of the disease. In this report, a convolutional neural network-based methodology is introduced to detect and distinguish the infection caused by SARS-CoV-2 from common viral Pneumonia through Chest X-ray images. The method consists of four convolutional layers with dense connections along with ReLU activation functions for image classification. The images are collected from the internet to create the entire dataset and are classified as COVID Pneumonia, non-COVID Pneumonia, and normal. The method produces a training accuracy of 97.61% and validation accuracy of 96.98%.