Ensemble Explainable Artificial Intelligence Model for COVID-19 Detection Using Chest X-Ray Images

After the virus was initially detected in Wuhan, China, in December 2019, it was announced in the midst of 2020 that an outbreak of COVID-19 had occurred, marking the onset of a global pandemic. Experts have put forth the hypothesis that this particular virus has the capability to infect both the upper respiratory system, including the sinuses, nose, and throat, as well as the lower respiratory system, encompassing the trachea and lungs. The most prevalent modes of transmission are through airborne droplets and direct contact with individuals carrying the COVID-19 virus. Currently, there are countless individuals across the globe who are grappling with this illness. Despite widespread preventive measures, the number of individuals affected by this disease and the death toll associated with it continue to rise at a disconcerting pace. There is a pressing need for ComputerAided Diagnosis (CAD) methods that can accurately identify this disease, yet many existing models are not utilized due to their lack of explainability. In this particular investigation, we present a methodology for swiftly and precisely identifying COVID-19 in Chest X-Ray (CXR) images, as this is the most commonly used and easily accessible medical diagnostic tool. To accomplish this, we employ DenseNet 121, a Deep Learning Model, on our dataset and achieve an impressive accuracy rate of 98.88%. Furthermore, we integrate Explainable Artificial Intelligence (XAI) models, specifically GRAD CAM ++ and SHAP, as an ensemble approach to enhance the interpretability of our classification model. This integration allows for a clearer visualization of the decision-making process and supports medical professionals in understanding how the algorithm arrives at its decisions.