Prediction of COVID 19 Using Chest X-Ray Images through CNN optimised using Genetic Algorithm

ABSTRACT

It's been over two years since the novel Coronavirus first appeared and with the constantly evolving new variants found around the world, the havoc of n-Coronavirus seems to be unstoppable. With more than 264 million people being affected by the n-coronavirus as of 2nd December 2021 and 5.2 million deaths around the world, there is a dire need to increase the COVID-19 testing to stop the virus from spreading further. With COVID - 19 devastating the economic situation of various countries across the globe, it has become necessary to come up with a fast, efficient, and inexpensive way to test the presence of the n-Coronavirus in people. However, the methods currently being used to test COVID 19 are rather very expensive and unavailable to a large section of society. One of the most feasible solutions to this problem is through radiological detection i.e., with Chest X - ray images. Contrary to the prevalent testing methods, Chest X - ray scans are much lesser in cost and are readily available. One major problem that arises is that COVID and pneumonia have very similar X-RAY results, so having a binary classification (COVID and NOT COVID) isn't enough. In this paper, we have put forward a model based on Convolutional NN for detection of Pneumonia, COVID - 19, and Normal patients using X - ray photos of Chest. We achieved an AUC score of 90% in our results while classifying the X-Ray Images. Besides Accuracy, we have also made the ROC Curve, confusion matrix, and classification report for our model. To keep our model lightweight, we have used a Genetic Algorithm to get the best hyperparameters possible for the model. © 2022 IEEE.