Covid-19 Diagnosis from Chest X-Ray Images using Transfer Learning and Data Augmentation

Abstract

The World Health Organisation (WHO) provided an update on the COVID-19 coronavirus illness on December 31, 2019. COVID-19 is a coronavirus disease that spread around the world in 2019. An viral sickness that affected people of all ages spread to more than 100 nations, causing a worldwide health disaster. This illness is extremely infectious because it may spread from person to person by respiratory droplets. The second wave caused numerous liver issues, pneumonia, respiratory failure, cardiovascular disorders, etc. and almost killed billions of people. This may be both symptomatic and asymptomatic in certain people, increasing their communicability as a result. A recent development that is beneficial in practically all study fields is machine learning. It is quite possible to handle this situation by using these methods to diagnose corona. There are several ways to test for the corona virus, but they are all time-consuming, costly, heavily reliant on test kits, more likely to result in false negatives, and subject to human error. The state-of-the-art of the covid diagnosis utilising chest X-ray pictures is presented in this article, which may serve as guidance for both doctors and technicians. Various machine learning (ML) and specifically Deep learning (DL) model are trained, which also classifies the images into normal, pneumonia, and covid images using a small dataset. The impact of Transfer learning and data augmentation is also studied in these schemes and using the best model, a 95% overall accuracy, 90% precision, and 90% F1 score is attained.