A Ensembled Model to Detect COVID-19 and Pneumonia Using Chest X-Ray Images

The novel coronavirus (COVID-19) has caused global outbreaks, leading to an epidemic with severe respiratory implications and high mortality rates. Chest X-rays are crucial tools for detecting COVID-19 infection at any stage, potentially reducing the death rate. This research focuses on proposing an optimized deep learning approach to automatically classify and diagnose COVID-19 and pneumonia using X-ray images. A globally available dataset from Kaggle containing a large number of chest X-ray images was utilized, consisting of normal, COVID-19, and viral pneumonia categories. Data augmentation techniques were applied to increase data size and prevent overfitting. The DL approach involves three stages: image enhancement, data augmentation, and transfer learning algorithms. By applying transfer learning with ensemble models, such as MobileNet and EfficientNet, on augmented enhanced images with frozen weights, the study achieved a high classification accuracy of 98%. Future work aims to further improve model efficiency, which is encouraging.