COVID-19 Detection using Convolutional Neural Network (CNN)

Abstract

Coronavirus disease (COVID-19) is spreading across the whole world and has habitual expressive community spread. The sudden hike in the number of patients with COVID-19, i.e. a new respiration virus has an extreme impact on the healthcare system. There are limited kits for diagnosis, hospital beds for such kind of patients, a limited number of personal protective equipment (PPE) for healthcare temporary staff and limited ventilators. A prediction system based on deep learning can help the healthcare system to respond immediately. The X-ray images can play an important role in the early prediction of COVID-19 patients and help in the patient treatment at an earlier stage. For the disease prediction, this study presents the use of Convolutional Neural Network (CNN) that extracts the features from the images of chest X-ray. To get the edges from the images, three convolution layers with different filters are applied. Keras’s Image Data Generator class is used to generate augmented images to deal with the small size of the training dataset. Classification is performed on three classes having X-ray images from COVID-19, normal people and viral pneumonia. The final results show that the proposed CNN model can predict COVID-19 patients with high accuracy.