CNN for Detection of COVID-19 Using Chest X-Ray Images

Abstract

AbstractCoronavirus spread globally in the late 2019, causing the whole world to face an existential health crisis. According to the recent report, animals may also get infected by the virus, so something needs to be done to eliminate this threat named corona. What if we are able to detect the virus at an early stage so that the time it gets to the critical condition, we would be equipped with certain measures. The first thing that gets affected in the body of the infected person are the lungs. So to check out the lung infection, we already have certain traditional techniques, but the automated detection of lung infection using CT Images gives an edge over the traditional healthcare system. Several challenges are faced in the segmentation of infected regions, including high variation in infection characteristics and low-intensity contrast between infections and normal tissues. In this work, we have taken the PA view of the chest X-ray images, which were found unhealthy at the time of screening. After cleaning up all the images or after we are done with data cleaning, we have applied. This work only focuses on the possible methods of classifying COVID-19 infected points, not claiming any medical accuracy. Deep learning to various models evaluates their performances. We have compared CNN, VGG19, Inception V3, Inception-ResNet, ResNet 152, XCEPTION, and we saw that ResNet 152 gave astonishing results.KeywordsCoronavirusCT imagesCNNVGG19Inception V3Inception-ResNetResNet 152XCEPTION