Performance Evaluation of Chest X-ray Image Based Deep Learning for COVID-19 Detection

Global society has been battling the novel Coronavirus (COVID-19) and mitigating its effects on the health and economic systems since the beginning of 2020. Multiple artificial intelligence techniques built on deep convolutional neural networks (CNN) prove to have good performances in detecting cases infected by COVID-19 using medical images. This paper uses three CNN models with pre-trained weights to perform a three-way classification on a chest X-ray dataset containing 317 images. Their performances are evaluated and compared. The results show that the classification obtained by the VGG-16 network is the most accurate in terms of precision and recall in identifying COVID-19-positive cases.