Materials and methods

Study participants
This retrospective study was approved by the Institutional Ethics Committee of Chongqing Three Gorges Central Hospital and the Second Affiliated Hospital of Chongqing Medical University, and the requirement for informed consent was waived.
Between January 21 and February 4, 2020, totally 72 patients of confirmed COVID-19 were enrolled, i.e., 66 patients from Chongqing Three Gorges Central Hospital and 6 patients from the Second Affiliated Hospital of Chongqing Medical University. All patients with COVID-19 were diagnosed according to the World Health Organization (WHO) interim guidance [9] and were eventually confirmed by a positive result for RT-PCR analysis of the respiratory secretions samples. Confirmed patients were eligible if they experienced symptoms within 7 days, underwent an initial chest CT examination, and they had not received any antiviral treatment. The exclusion criteria were as follows: (1) patients who recently experienced clinically defined pulmonary infection attributable to other pathogens, (2) patients with severe artifacts on CT images, and (3) patients whose age was less than 18 years. According to the exclusion criteria, 7 patients who had pulmonary infections caused by other pathogens and 3 patients whose CT images had severe artifacts were excluded. Finally, 62 confirmed COVID-19 patients (34 men, 28 women; age range 20–91 years old) who had not received antiviral treatment were retrospectively reviewed. The workflow of COVID-19 patients selection is shown in Fig. 1. Most cases had a history of SARS-CoV-2 exposure, in the absence of medical staff or family clusters of SARS-CoV-2 infection. The main symptom of patients was fever, and secondary symptoms included cough, expectoration, dizziness, nasal congestion, and myalgia. According to the changes in this disease over time [10, 11], in our study, we defined patients with an onset of symptoms within 4 days as the early-stage group (18 men, 16 women; age range 20–72 years old) and patients with an onset of symptoms within 5–7 days as the progressive-stage group (16 men, 12 women; age range 22–91 years old). All patients underwent CT scans and laboratory examinations for C-reactive protein (CRP), white blood cell count, neutrophil count, lymphocyte count, platelet count, hemoglobin level, and procalcitonin level on admission.
Fig. 1 The workflow of coronavirus disease 2019 selection

CT data acquisition
Designated independent operating rooms were used to perform the CT examinations of all confirmed patients and after each examination; the operating room was sterilized with air disinfection method. The initial chest CT data of 56 patients from Chongqing Three Gorges Central Hospital were obtained by using a 16-slice spiral CT scanner (Emotion 16 VC20B, Siemens Healthcare GmbH) with the following scanning protocol: a tube voltage of 120 kV, smart mA tube current modulation, a slice thickness of 1.5 mm, and a detector width of 1.5 mm. The initial chest CT data of the other 6 patients from the Second Affiliated Hospital of Chongqing Medical University were obtained by using a 128-slice spiral CT scanner (Ingenuity CT, Philips Healthcare) with a similar scanning protocol: a tube voltage of 120 kV, smart mA tube current modulation, a matrix size of 512 × 512, a slice thickness of 1 mm, and a detector width of 0.625 mm. All patients were scanned in the supine position and at the suspended end-inspiratory volume, and the entire lung was scanned from the apex to the bottom [12]. To reduce discrepancies in the image acquisition parameters attributable to the different CT scanners, we resampled all voxels at 1.0 × 1.0 × 1.0 mm3 before image analysis. The multiplanar reconstruction (MPR) technique was performed to better observe images at the postprocessing station when necessary.

CT data analysis
All CT findings were described according to the Fleischner Society recommendations and defined as follows: GGO (appears as hazy increased opacity in the lung, with the preservation of bronchial and vascular margins), consolidation (appears as a homogeneous increase in pulmonary parenchymal attenuation that obscures the margins of vessels and airway walls), a crazy-paving pattern (appears as thickened interlobular septa and intralobular lines superimposed on a background of GGO), and a halo sign (appears as GGO surrounding a nodule or mass) [13, 14]. Attention was paid to CT images with GGO, consolidation, GGO with consolidation, a reticular pattern, nodules, rounded opacities, cavitation, a crazy-paving pattern, an air bronchogram, a halo sign, bronchial wall thickening, a subpleural curvilinear line, and pulmonary fibrosis, and the presence, distribution, and size of these CT characteristics were assessed in all subjects. In addition, the presence of thoracic lymphadenopathy (lymph node size of greater than or equal to 10 mm in the short-axis dimension) and pleural effusion or thickening were also recorded. The distribution of each finding was classified as follows: central (predominantly in the inner two-thirds of the lung), peripheral (predominantly in the outer third of the lung), and diffuse (indications in multiple lung segments) [15]. Calculation of the CT score, a semiquantitative evaluation method, was applied to quantify the extent of COVID-19 according to previous studies [15, 16]. First, the scope of the lesions in each lobe was estimated and a score of 0 (none), 1 (affecting less than 5% of the lobe), 2 (affecting 5–25% of the lobe), 3 (affecting 26–49% of the lobe), 4 (affecting 50–75% of the lobe), or 5 (affecting more than 75% of the lobe) was assigned. Second, the CT score was obtained by adding up the scores of the five lobes. For each patient, the CT score was in the range of 0 to 25. All CT scans were independently reviewed and assessed by two chest radiologists (Z.F. and L.L.C., who had 18 years of experience in thoracic radiology, respectively). Disagreements in reader interpretation were settled by consensus agreement.

Statistical analysis
All of the data were analyzed with statistical software (SPSS statistical package, version 19.0; IBM Corporation). Continuous variables were expressed as means ± standard deviations or median (interquartile range (IQR)), and categorical variables were presented as numbers (percentages). Kolmogorov-Smirnov tests were applied to check the normality of variables. Mann-Whitney U tests were used for non-normally distributed data. Two-sample t tests were used for normally distributed data. Chi-square tests or Fisher’s exact tests were applied to compare categorical variables. Spearman’s correlation was performed to analyze the relationship between clinical variables (age, fever, CRP, white blood cell count, neutrophil count, lymphocyte count, platelet count, hemoglobin level, and procalcitonin level) and CT variables (maximum diameter of the lesion and CT score). For all test results, a p value of less than 0.05 was considered statistically significant.