Deep Learning-Based Recognizing COVID-19 and other Common Infectious Diseases of the Lung by Chest CT Scan Images

Min Fu; Shuang-Lian Yi; Yuanfeng Zeng; Feng Ye; Yuxuan Li; Xuan Dong; Yan-Dan Ren; Linkai Luo; Jin-Shui Pan; Qi Zhang

Because of the relatively high positive rate of CT imaging in the early stage and the 299 characteristic lesions of COVID-19 such as ground-glass opacity (3, 4, 12) , CT imaging 300 has potential in the diagnosis of COVID-19 that cannot be ignored. There is large 301 number of potential patients in need. More than that, each examination of CT imaging 302 will generate a large number of images and significant inter-observer-variation exists 303 in the interpretation of CT images. Thus, it is necessary to develop new auxiliary 304 measures for the interpretation of CT images. In this study, we report an artificial 305 intelligence framework based on deep learning for identifying COVID-19, which has 306 balanced sensitivity and specificity. More than that, the area under the ROC curve was 307 high as 99.0% evaluated by test dataset. Another advantage of our study is that five 308 diseases or conditions were enrolled, which cover the most common infectious 309 diseases of the lung. The limitation is that our AI framework will need further 310 evaluation by more wide clinical application. 311 312 All rights reserved. No reuse allowed without permission. the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.