Assisting scalable diagnosis automatically via CT images in the combat against COVID-19.

The pandemic of Coronavirus Disease 2019 (COVID-19) is causing enormous loss of life globally. Prompt case identification is critical. The reference method is the real-time reverse transcription PCR (RT-PCR) assay, whose limitations may curb its prompt large-scale application. COVID-19 manifests with chest computed tomography (CT) abnormalities, some even before the onset of symptoms. We tested the hypothesis that the application of deep learning (DL) to 3D CT images could help identify COVID-19 infections. Using data from 920 COVID-19 and 1,073 non-COVID-19 pneumonia patients, we developed a modified DenseNet-264 model, COVIDNet, to classify CT images to either class. When tested on an independent set of 233 COVID-19 and 289 non-COVID-19 pneumonia patients, COVIDNet achieved an accuracy rate of 94.3% and an area under the curve of 0.98. As of March 23, 2020, the COVIDNet system had been used 11,966 times with a sensitivity of 91.12% and a specificity of 88.50% in six hospitals with PCR confirmation. Application of DL to CT images may improve both efficiency and capacity of case detection and long-term surveillance.

Clinical differences in chest CT characteristics between the progression and remission stages of patients with COVID-19 pneumonia.

INTRODUCTION: Computed tomography (CT) can be effective for the early screening and diagnosis of COVID-19. This study aimed to investigate the distinctive CT characteristics of two stages of the disease (progression and remission). METHODS: We included all COVID-19 patients admitted to Wenzhou Central Hospital from January to February, 2020. Patients underwent multiple chest CT scans at intervals of 3-10 days. CT features were recorded, such as the lesion lobe, distribution characteristics (subpleural, scattered or diffused), shape of the lesion, maximum size of the lesion, lesion morphology (ground-glass opacity, GGO) and consolidation features. When consolidation was positive, the boundary was identified to determine its clarity. RESULTS: The ratios of some representative features differed between the remission stage and the progression phase, such as round-shape lesion (8.0% vs 34.4%), GGO (65.0% vs 87.5%), consolidation (62.0% vs 31.3%), large cable sign (59.0% vs 9.4%) and crazy-paving sign (20.0% vs 50.0%). Using these features, we pooled all the CT data (n = 132) and established a logistic regression model to predict the current development stage. The variables consolidation, boundary feature, large cable sign and crazy-paving sign were the most significant factors, based on a variable named "prediction of progression or remission" (PPR) that we constructed. The ROC curve showed that PPR had an AUC of 0.882 (cutoff value = 0.66, sensitivity = 0.75, specificity = 0.875). CONCLUSION: CT characteristics, in particular, round shape, GGO, consolidation, large cable sign, and crazy-paving sign, may increase the recognition of the intrapulmonary development of COVID-19.