COVID-19 Recommendation System of Chest X-Ray Images Using CNN Deep Learning Technique with Optimizers and Activation Functions

Sinha, N.; Karjee, P.; Agrawal, R.; Banerjee, A.; Pradhan, C.

Most people who are infected with the COVID-19 virus show respiratory illness. As the virus is newly discovered, the world is not ready with an adequate number of testing kits so as to test each individual in the world. So, a different method that tells if a person may be infected or not will prove to be a big relief and will also help in decreasing the tension and fear among people. This problem can be solved by training machines using deep learning algorithms. The proposed CNN model takes input as chest X-ray images and predicts whether a person needs to go for COVID-19 testing or not. This will decrease the fear of being COVID-19 infected and can be easily implemented even in rural areas;which will also decrease the load of the COVID-19 testing team. At present, the COVID-19 tests are being done clinically through blood tests or nose/throat swab tests which requires around 24 h to give results. The proposed recommendation system takes around 35 min to give the result of a sample and curbs down the chances of virus spread through contact while testing, unlike the presently used methods. The experimental results yielded an accuracy of 97.62% using the chest X-ray scans and requires less computational time. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.