Using Deep Learning Algorithms in Chest X-ray Image COVID-19 Diagnosis

We are experiencing heavy COVID-19 outbroke globally since January 2020. In Taiwan, because its low infection rate (< 0.01%), there was not enough evidence for diagnosis through medical imaging. At present, chest X-ray is widely used in lung infection diagnoses. This study uses deep learning methods to assist doctors in classifying COVID-19 disease from chest X-ray images. After pre-processing, the images were put into the VGG16 model to automatically classify into three categories to assist the radiologist in the treatment of the disease. The results show that the classification accuracy was 78%. Detail analyses disclosed that this accuracy can be improved by rectifying the unbalanced images problem. In addition, choosing proper image pre-processing algorithms has a high tendency to generate better results.