Our institutional review board approved this retrospective study and written informed consent from all patients was waived.
Patients and CT image data acquisition
Fifty-three patients with confirmed COVID-19 from seven hospitals in Sichuan province in China were enrolled in our study. They underwent chest CT examination and reexamination from 16 January 2020 to 25 February 2020. Patient selection was consecutive, and the exclusion criteria were COVID-19 patients without abnormal manifestations on CT. Finally, three patients were excluded because of no abnormal manifestations on CT, and fifty patients were included. The clinical history, laboratory, epidemic characteristics, and chest CT images were collected. The date of both the initial chest CT examination and the first positive RT-PCR test was recorded for each participant. All patients underwent CT scanning and laboratory tests when the initial mouth swab test was performed. All patients were followed up with chest CT during the study period without intravenous contrast agents. All patients were imaged with 0.6 mm to 1.25 mm thick slices in commercial multi-detector CT scanners (SOMATOM Definition AS and STRATON MX, Siemens Healthineers, Erlangen, Germany; BrightSpeed scanner, GE Medical Systems, Milwaukee, Wis; Philips Ingenuity Core128, Philips Medical Systems, Best, Netherlands; and UCT 760 scanner, United Imaging, Shanghai, China). The mean CTDIvol was 6.6 ± 2.3 mGy (range: 4.1-10.3 mGy).
All patients were at least two positive results by real-time RT-PCR assay for COVID-19 at laboratory testing of respiratory secretions obtained from the nasopharyngeal swab, oropharyngeal swab, endotracheal aspirate, or bronchoalveolar lavage. According to the guideline of COVID-19 (Trial Version 6), the patients were typed into two groups. The moderate group was defined as: the patient had a fever, respiratory symptoms, and abnormal imaging findings of pneumonia. The severe and critical group was defined as meeting any of the following: Severe respiratory distress (respiratory rate > 30 breaths/min); Oxygen saturation (SpO2) £ 93% at rest; Partial arterial oxygen pressure (PaO2)/ Fraction of inspired oxygen (FiO2) ≤ 300 mmHg (1 mmHg = 0.133kPa); Respiratory failure and requirement for mechanical ventilation; Shock; and Combined with multi-organ failure and requirement for intensive care unit (ICU).
The patients with diagnosed COVID-19 pneumonia were isolated and hospitalized for treatment. Discharge standards[3, 13] are as follows: 1. The temperature returned to normal for more than 3 days; 2. Respiratory symptoms were significantly improved; 3. Pulmonary imaging showed obvious signs of absorption in acute exudation inflammation of the lungs; and 4. Respiratory nucleic acid was negative for two consecutive times (at least one-day sampling time interval). The patient who meets the four conditions simultaneously can be released from isolation.
Chest CT evaluation
The initial and follow-up chest CT images were assessed as the following ten characteristics according to the Fleischner Society Glossary[18, 19] and peer-reviewed literature on viral pneumonia[10, 14], such as ground-glass opacity (GGO), crazy-paving pattern, consolidation, pleural thickening or adhesion, fibrosis, discrete nodules, cavitation, lymph node enlargement, pleural effusion, and bronchiectasis. A semi-quantitative scoring system was used to quantitatively estimate the extent of pulmonary involvement. The area of abnormal pulmonary involvement was scored for each of the five lung lobes as follow: zero scores (no abnormal involvement), one score (1%-25% abnormal involvement), two scores (26%-49% abnormal involvement), three scores (50%-75% abnormal involvement), and four scores (76%-100% abnormal involvement). Finally, the total CT severity score was a sum of 5 lobe scores ranging from 0 to 20. The interval time between initial CT and the first follow-up CT scan was defined as Interval-1, and the interval between initial CT and the second follow-up CT was defined as Interval-2, and so on.
Image analysis was performed by two radiologists with more than 8 years of experience (N.L. and XH.H.) by using a DICOM Viewer software (Medixant. RadiAnt DICOM Viewer [Software]. URL: https://www.radiantviewer.com). Images were reviewed independently, and final scores were reached by consensus.
All statistical analyses were conducted using SPSS software (version 22.0, U.S.A.). Quantitative variables were expressed as mean±standard deviation (minimum-maximum) and the categorical variables were expressed as the percentage of the total. Demographic variables (e.g., age and gender), clinical characteristics (i.e., the hospitalized period and mean number of scans), and the occurrence rate of CT characteristics were compared by independent sample t test, chi-square test, or Fisher’s exact test. The Shapiro-Wilk test was used for the normal distribution. The CT severity scores were compared by the Mann-Whitney U test. SPSS curve estimation module was used to quantitatively evaluate the total CT severity score of pulmonary as a function of CT follow-up time. The statistical significance level was set at p = 0.05 with two-tailed.