Early chest computed tomography to diagnose COVID-19 from suspected patients: A multicenter retrospective study

Congliang Miao; Mengdi Jin; Li Miao; Xinying Yang; Peng Huang; Huanwen Xiong; Peijie Huang; Qi Zhao; Jiang Du; Jiang Hong

The copyright holder for this preprint (which was not peer-reviewed) is . https://doi.org/10.1101/2020.03.24.20042432 doi: medRxiv preprint 8 As shown on Table 5 , the GGO with bilateral pulmonary distribution and GGO with peripleural distribution both presented highest as 57% on sensitivity, followed by GGO combined with bilateral pulmonary distribution and peripleural distribution as at 48%. The remaining combinations were around 25%. Contrarily, the significance of combination presented more obvious on specificity. When combining two items, the specificity of GGO with crazy-paving pattern was up to 96%, followed by GGO with air bronchogram (95%). When three features were combined, the specificity was over 95%, and GGO with crazy-paving pattern and bilateral pulmonary distribution was the highest at 99%. Specificity of diagnosis was almost 99% for four items.