Introduction

Coronavirus Disease 2019 (COVID-19) is currently a global pandemic, and the early screening of COVID-19 is one of the key factors for COVID-19 control and treatment. Here, we developed and validated chest CT-based imaging biomarkers for COVID-19 patient screening. We identified the vasculature-like signals from CT images and found that, compared to healthy and community acquired pneumonia (CAP) patients, the COVID-19 patients revealed significantly higher abundance of these signals. Furthermore, unsupervised feature learning leads to the discovery of clinical-relevant imaging biomarkers from the vasculature-like signals for accurate and sensitive COVID-19 screening that has been double-blindly validated in an independent hospital (sensitivity: 0.941, specificity: 0.920, AUC: 0.971). Our findings could open a new avenue to assist screening of COVID-19 patients.

Study Design

A graphic illustration of the study design. A case-control study was designed to identify chest CT-based imaging biomarkers for COVID-19 patient screening. Biomarker discovery and biomarker-based predictive model construction were conducted using the data from Hospital A (training cohort), which were validated in Hospital B (validation cohort) with the double-blind design.

Result

Chest CT-based imaging biomarkers highly predicts COVID-19. a.  Representative examples for 3D multispectral imaging biomarker visualization in COVID-19, CAP and healthy samples b. The boxplot shows differences in the vasculature-like signals among healthy, community acquired pneumonia (CAP), and COVID-19 patients in the training cohort. The p-values were obtained by the non-parametric Mann-Whitney test. c. PCA of 8 imaging biomarkers in the training cohort. 20 healthy participants (green dots), 49 CAP patients (blue dots), and 47 COVID-19 patients (red dots). The p-values were obtained from permutational multivariate analysis of variance (PERMANOVA). d. The boxplot shows differences in the vasculature-like signals among healthy, community acquired pneumonia (CAP), and COVID-19 patients in the validation cohort. The p-values were obtained by the non-parametric Mann-Whitney test. e. PCA of 8 imaging biomarkers in the validation cohort. 60 healthy participants (green dots), 90 CAP patients (blue dots), and 153 COVID-19 patients (red dots). The p-values were obtained from permutational multivariate analysis of variance (PERMANOVA). f. Screening performance of signal-based model, imaging biomarker-based model, and two COVID-19 experienced radiologist on validation cohort.