Comparison of COVID-19 Diagnosis by CNN Model and ResNet Using Chest X-Ray

COVID-19 pandemic posed a threat to productivity, health, and human life. During pandemic, the rise in COVID-19 cases demands high number of tests of RT-PCR, Swab tests and blood tests, but sometimes these tests do not give accurate results and so a lot of people prefer CT scans and X-rays over the above mentioned tests. In this situation, the hospitals have to house enormous numbers of patients and it is highly impossible for doctors and radiologists to look into each and every X-Ray manually and detect the virus, and sometimes there could be manual errors as well. Hence, this research is an attempt to automate COVID-19 detection by making use of X-Ray images with the help of Convolutional Neural Network. Machine learning models are used on COVID-19 X-Ray images to predict infectious cases. In this work, the aim to classify Chest X-Ray images of COVID-19 patients into normal images and COVID affected images using Keras API with Tensorflow, so that this helps in finding the affected patients on a large scale. CNN extracts spatial characteristics from images which is a very expeditious for image classification problem. Experimental results show that CNN proves to be the best in obtaining precise and accurate results of COVID-19 detection up to the accuracy of 96.8%.