COVID-19 Diagnosis from X-Ray Images Using Deep Learning

Efficient screening of COVID-19 patients is very critical to the allocation and management of hospital resources. This paper presents a model Covid-DenseNet that can rapidly diagnose COVID-19 from chest X-ray (CXR) images. By combining transfer learning and the addition of attention mechanism, we extracted the feature information from CXR images and carried out internal weight allocation of the feature information. We also design a multiscale fusion architecture to fuse multiple scale features after weight allocation to obtain more feature information, so as to improve the model’s ability to correctly identify chest radiographs. Experiments were conducted on the COVID-QU-Ex dataset, and the accuracy of Covid-DenseNet was as high as 96.89%.