Detection of COVID-19 from Chest X-rays Using Resnet-50

Abstract

COVID-19 is a pandemic caused by SARS-COV2 which started from December, 2019 and is being spreading since then. There are more than 12 million cases in India and more than 130 million cases worldwide according to the current statistics. One of the basic challenges we face today is the early detection of this disease and putting the patients in special care as soon as possible. It's fast spreading nature makes the situation more complex and putting together all the solutions we have today it seems like the wave is not under control. Current diagnosing methods and cures have its own drawbacks and it is clear that we need an alternate solution. The aim of the study is to evaluate performance of state-of-the-art pre-trained model ResNet-50 on COVID-Chest X-ray dataset consisting of 1000 samples. Resnet-50 achieved 96% accuracy with 0.98 sensitivity and 0.95 specificity.