Transfer Learning Based Method for Automatic COVID-19 Cases Detection in Chest X-Ray Images

COVID-19 cases are increasing at an exponential rate, imposing challenges for healthcare organizations all around the world. According to recent research, the chest Xrays on COVID-19 individuals possess significant anomalies in the radiograph. The use of sophisticated artificial intelligence (AI) technology in conjunction with chest radiography could be beneficial in detecting this disease accurately. Therefore, this research study has implemented a deep transfer learning (TL) approach by utilizing a pre-trained network (ResNet50) for achieving a quicker and more efficient COVID-19 detection from chest X-ray images. The dataset used for this experiment comprises 2905 chest X-ray images in total (COVID-19, Viral Pneumonia, and Normal). Experimental results reveal that the implemented deep transfer learning model produced a substantial performance in terms of classification with 96% of overall performance accuracy. The model also provides high sensitivity (recall) and Specificity of 100% in detecting the COVID-19 cases in contrast to the current state-of-the-art techniques. Therefore, the proposed approach can be beneficial for quicker and more efficient early detection of COVID-19 patients reducing the strain on healthcare organizations.