An Automated Coronavirus Disease Detection Approach Using a Deep Neural Network from X-ray Images

Abstract

The novel coronavirus has recently become a pandemic disease since December 2019. It has a detrimental effect on our everyday lives, the human species, and many innocent lives. Moreover, there is no accurate automated system for the prominent diagnosis of coronavirus at the earlier infection phase. Therefore, this research has proposed a computer-aided deep learning framework named COVID-CXDNet to detect coronavirus from the X-ray images in an easy way. The proposed approach is based on the modification of YOLO (You only look once) with ResNet (Residual neural network) trained by a vast X-ray dataset of 3788 multiclass image frames such as Normal, COVID-19, and Pneumonia. The model has achieved the highest 98.00% accuracy, 99.63% sensitivity, and 100% specificity with a loss of 0.055 for multiclass diseases. It has also obtained 99.58% accuracy, 99.04% sensitivity, and 100% specificity with a loss of 0.005 for binaryclass diseases, which