https://orcid.org/0000-0001-5251-6023
Abstract
Purpose
To investigate the role of chest computed tomography (CT) scan in the prediction of admission of pregnant women with COVID-19 into intensive care unit (ICU).
Methods
This was a single-center retrospective case–control study. We included pregnant women diagnosed with COVID-19 by reverse transcriptase polymerase chain reaction between February 2020 and July 2021, requiring hospital admission due to symptoms, who also had a CT chest scan at presentation. Patients admitted to the ICU (case group) were compared with patients who did not require ICU admission (control group). The CT scans were reported by an experienced radiologist, blinded to the patient’s course and outcome, aided by an artificial intelligence software. Total CT scan score, chest CT severity score (CT-SS), total lung volume (TLV), infected lung volume (ILV), and infected-to-total lung volume ratio (ILV/TLV) were calculated. Receiver operating characteristic curves were constructed to test the sensitivity and specificity of each parameter.
Results
8/28 patients (28.6%) required ICU admission. These also had lower TLV, higher ILV, and ILV/TLV. The area under the curve (AUC) for these three parameters was 0.789, 0.775, and 0.763, respectively. TLV, ILV, and ILV/TLV had good sensitivity (62.5%, 87.5%, and 87.5%, respectively) and specificity (84.2%, 70%, and 73.7%, respectively) for predicting ICU admission at the following selected thresholds: 2255 mL, 319 mL, and 14%, respectively. The performance of CT-SS, CT scan score, and ILV/TLV in predicting ICU admission was comparable.
Conclusion
TLV, ILV, and ILV/TLV as measured by an artificial intelligence software on chest CT, may predict ICU admission in hospitalized pregnant women, symptomatic for COVID-19.
Introduction
The association between the severe acute respiratory syndrome coronavirus 2 (SARS-CoV2) infection and pregnancy is becoming clearer. In a cohort study with propensity score matching for age, body mass index (BMI), and comorbidities (diabetes, hypertension, asthma), pregnant women at ≥20 weeks of gestation were at significantly higher risk for intensive care unit (ICU) admission, endotracheal intubation, hospitalization for disease-related symptoms, and need for oxygen therapy [Citation1]. A systematic review subsequently showed an increased risk of ICU admission in infected pregnant women compared to infected nonpregnant and to noninfected pregnant women [Citation2]. The multicenter international study (PregOuTCOV study) also demonstrated that the infection significantly increases adverse obstetrical and neonatal outcomes, from 20 weeks’ gestation [Citation3].
At the beginning of the pandemic, computed chest tomography (CT) scan was a very useful tool to diagnose the new coronavirus disease (COVID-19) [Citation4,Citation5]. However, its diagnostic role is now somewhat limited due to the widespread use and availability of the gold standard reverse transcriptase polymerase chain reaction (RT-PCR). Recently, several studies have demonstrated that disease severity as well as ICU admission in men and non-pregnant women may be predicted by visual and software-based quantitative chest CT [Citation6–8]. Data about the use of these tools in pregnant women population infected by SARS-CoV2 are limited.
The aim of this study was to examine whether chest CT scan parameters, as measured by an artificial intelligence software, could predict ICU admission in symptomatic pregnant women hospitalized for SARS-CoV2 infection.
Material and methods
Study design and population
This was a retrospective case–control study conducted at a single referral center from February 2020 to July 2021. We included consecutive pregnant patients admitted to the high-risk pregnancy unit with COVID-19 who had undergone chest CT during their hospital stay. The diagnosis of the disease was confirmed by a positive RT-PCR from nasopharyngeal swabs. We excluded patients who had no CT scan and those who were admitted for reasons other than COVID-19 infection. Patients were categorized into two groups. A case group consisting of all those who were admitted to the ICU due to respiratory deterioration and the remaining patients, who formed the control group.
Clinical data were routinely collected in real time and recorded in the patient’s electronic medical records. Data were then extracted retrospectively for the study and merged into a dedicated, secured, and anonymized database. A data control was performed before any analysis and in cases where there were inaccurate or missing data, the medical records were rechecked to correct the identified issues.
The following data were collected: maternal age, smoking status, pre-pregnancy BMI, gravidity, parity, chronic arterial hypertension, diabetes mellitus type I or II, preexisting pulmonary diseases (such as asthma, tuberculosis, and previous pulmonary embolism), gestational age (GA) at the time of RT-PCR, GA at the onset of symptoms, and COVID-19 symptoms at presentation. We also recorded GA at the delivery—regardless of whether the pregnancy was induced due to COVID-19—rate of Cesarean delivery, nonreassuring fetal heart rate (FHR), birthweight, APGAR score at 5 min of life, umbilical cord pH, respiratory distress at birth, and admission to the neonatal intensive care unit (NICU). The study was approved by the institutional Ethical Committee (approval number: CE 2021/125).
Abbreviated institutional protocol for management of pregnant patients with COVID-19 (during the data collection phase) [Citation9]
Admission criteria for pregnant patients with COVID-19 include high-grade fever, persistent cough, dyspnea, reduced level of consciousness, severe fatigue, signs of pneumonia, severe gastrointestinal disturbances, decreased fetal movements, preterm labor, or other serious obstetric indications [Citation10–14]. Chest CT scan or X-ray are done at admission, especially when physical exam raises suspicion of pulmonary infection. Oxygen (O2) therapy is started when oxygen saturation (SaO2) falls below 96% and titrated to response using noninvasive ventilatory modalities. We begin with a simple nasal cannula (1–6 L/min) followed by progressively increasing oxygen concentrations via a Venturi-type face mask. Our pragmatic target for SaO2 is 93%–94%, but the patient’s breathing efforts must also be considered when adjusting O2 therapy, in order to minimize the risk of maternal fatigue. If saturation can be maintained at these target levels and no other complications occur, then pregnancy can safely continue, provided that maternal cardiorespiratory status and obstetric investigations remain reassuring.
The patient’s cardiorespiratory status may deteriorate, leading to progressive respiratory failure, acute respiratory distress syndrome, septic shock, or multiorgan failure. At this point, the patient is no longer suitable for management in a high-risk maternity unit or an intermediate COVID-19 unit. This represents a failure of basic support and the level of care provided should be increased, using more advanced methods of oxygen delivery, such as nasal high flow O2, continuous positive airway pressure, or in selected cases bilevel positive airway pressure, sometimes in combination with inotropic support, where optimization cardiovascular function is required [Citation15]. This level of support is usually provided in the ICU.
Technical conditions
Patients were scanned in the supine position. The abdomen was covered with a lead shield. The scan was performed at the end of inspiration. The following CT scan machines were used: Siemens Definition AS/AS+ 128 slices, Siemens Definition AS/AS+ 64 slices, Siemens Somaton Drive 256 slices, and Siemens Somatom Sensation 64 slices. The parameters of low-dose acquisition were 80 kV, 30 mA, 0.6 mm slice thickness with multiplanar reconstructions in mediastinal, lung parenchyma, and MIP algorithm at 1, 3, and 4 mm, respectively. On the Somatom Drive, care doses for kV and mA were set automatically. In addition, when the Pulmonary Embolic Protocol was scheduled, a pre-monitoring on the common pulmonary trunk and the injection of 40 mL of lobitridol 350 mgL/mL (Xenetix®) followed by 60 mL of physiological solution with a flow rate of 5 mL/min if possible were done.
Interpretation of the CT scan
The analysis of the CT scan was done by a senior (MMC) and an in-training radiologists (FDL) who were blind to the patients’ characteristics and outcome. They measured the chest CT severity score (CT-SS), the CT scan score, total lung volume, and infected lung volume. In the CT-SS, the lungs are divided into 20 regions. The score for each region is 0, 1, or 2 if it is not involved by the infection, involved at <50%, or involved at >50%, respectively [Citation16]. The CT scan score consists in scoring the degree of involvement of each of the five lobes (0: no involvement, 1: <5% involvement, 2: 5%–25% involvement, 3: 26%–49% involvement, 4: 50%–75% involvement, and 5: >75% involvement) and then calculating the sum of the scores for each lobe [Citation17].
The total lung volume and infected lung volume were automatically calculated with the Syngo.via application, called, pneumonia analysis. The infected lung volume was identified by an automatic contouring of the opacities based on a density threshold value (Hounsfield unit, UH). A manual correction was performed when necessary. The infected to total lung volume ratio (%) was calculated via the following formulae: (infected lung volume × 100)/total lung volume.
Statistical analysis
Data were analyzed using the SPSS 26 statistical software (IBM SPSS statistics). Continuous variables were represented as mean ± 1 SD and categorical variables were represented as number (frequency). The Shapiro–Wilk test was used to examine the normal distribution of continuous variables and either Student’s test or Mann–Whitney U test, to compare the means of these variables. Fisher’s exact test or Pearson’s Chi-square were used to compare categorical variables. Finally, receiver operating characteristic (ROC) curves were constructed to test the sensitivity and specificity of each CT scan parameter, as well as those of the CT scan score and the CT-SS, in predicting ICU admission in symptomatic pregnant women, hospitalized patients with COVID-19. We set the threshold in order to obtain the highest sensitivity and specificity for each parameter. Statistical significance was assumed when the p ≤ .05.
Results
Baseline characteristics of the study population
During the study period, 307 pregnant patients tested positive for SARS-CoV2 RT-PCR. Around 194 (63.19%) had COVID-19 symptoms, but only 28 (13.43%) were hospitalized due to these symptoms and had a chest CT scan. None of them was vaccinated against SARS-CoV2 or took SARS-CoV2 treatment (such as remdesivir or steroids) before hospital admission. Eight of these 28 patients (28.57%) required ICU admission (case group) and the remaining 20 formed the control group (Supplemental Figure 1). Maternal age, pre-pregnancy BMI, gravidity, and parity were similar in both groups. There were no patients with pre-gestational diabetes mellitus, chronic hypertension, or smoking habit in the two groups. Only one patient in the case group had asthma.
Patients who were admitted to the ICU were more likely to have fever (100% versus 40%, p = .004) and dyspnea (87.5% versus 45%, p = .04) in comparison to the control group. They also had significantly lower lymphocyte counts (0.59 × 109/L ± 0.11 × 109/L versus 1.46 × 109/L ± 0.75 × 109/L, p < .001). No statistically significant differences were reported for the other baseline characteristics ().
Table 1. Baseline characteristics, symptoms and laboratory tests of the study population.
Pregnancy and neonatal outcomes
Four patients (50%) in the case group were delivered due to symptomatic COVID-19 in comparison to only one (5%) in the control group (p = .005). Cesarean delivery was significantly higher in patients admitted to the ICU (62.5% versus 15%, p = .022). In addition, three neonates (37.5%) in the case group had respiratory distress at birth in comparison to only one (5%) in the control group (p = .026). GA at delivery, nonreassuring FHR, birthweight, APGAR score at 5 min, cord pH, and NICU admission were not statistical different between the two groups ().
Table 2. Comparison of the pregnancy and neonatal outcomes between the two groups.
CT scan parameters
CT scan was performed at the similar lead times following the onset of symptoms in the case and control groups (5.75 days ± 1.75 days versus 8.45 days ± 7.16 days, p = .381). There was no difference of lesions’ characteristics, such as ground-glass opacities, crazy-paving pattern, consolidations, or pleural effusion between the two groups. However, the number of infected lungs’ lobes was higher in patients who were admitted to ICU (5 versus 3.55 ± 1.76, p = .013).
Regarding lung volumes measured by the artificial intelligence software, total lung volume was significantly lower in the case group (2275.51 mL ± 277.58 mL versus 2841.32 mL ± 684.37 mL, p = .005) infected lung volume was significantly higher (516.59 mL ± 296.56 mL versus 302.80 mL ± 493.91 mL, p = .025), and infected to total lung volume ratio was significantly higher (22.82% ± 14.35% versus 11.32% ± 14.64%, p = .034) in comparison to the control group (). The CT scan score and the CT-SS were higher as well in patients admitted to the ICU (11.38 ± 4.24 versus 6.55 ± 5.06, p = .025, and 19.25 ± 4.71 versus 10.60 ± 9.08, p = .017, respectively).
Table 3. Comparison of the parameters of the computed tomography scan parameters between the two groups.
Sensitivity and specificity of the CT scan parameters
Supplemental Figure 2 illustrates the ROC curves for the performance of the three CT scan parameters in the prediction of ICU admission. A total lung volume less than 2255 mL predicts ICU admission with a sensitivity of 62.5% and a specificity of 84.2% (area under the curve [AUC] = 0.789, 95% confidence interval [95%CI]: 0.615–0.964, p = .019). An infected lung volume greater than 319 mL predicts ICU admission with a sensitivity of 87.5% and a specificity of 70% (AUC = 0.775, 95%CI: 0.600–0.950, p = .025). An infected to total lung volume ratio of 14% or more is predictive for ICU admission with a sensitivity of 87.5% and a specificity of 73.7% (AUC = 0.763, 95%CI: 0.582–0.944, p = .034).
In addition, ROC curves of the CT scan score showed that its AUC was 0.778 (95%CI: 0.585–0.971, p = .024), and that of CT-SS was 0.822 (95%CI: 0.667–0.977, p = .009). Both scores had good specificity and sensitivity to predict ICU admission in pregnant women. Pairwise comparison of the ROC curves of CT-SS, CT scan score, and ILV/TLV for the prediction of ICU admission showed no significant difference amongst the three AUCs (Supplemental Figures 3–5).
Discussion
This study demonstrates that measurements of total lung volume, infected lung volume, and infected-to-total lung volume ratio on CT chest scan by an artificial intelligence software have the potential to predict the risk of ICU admission in symptomatic pregnant women hospitalized with COVID-19. These three parameters had good sensitivity (62.5%, 87.5%, and 87.5%, respectively) and specificity (84.2%, 70%, and 73.7%, respectively) at the following selected thresholds: 2255 mL, 319 mL, and 14%, respectively.
A CT scan is a useful diagnostic tool for a variety of clinical presentations during pregnancy, such as acute abdominal pain or trauma [Citation18,Citation19]. If a scan is clinically indicated, then it should not be withheld, although some obstetricians remain resistant to the use of this modality in pregnancy due to largely unfounded concerns about radiation exposure. In these circumstances, the clinician has an obligation to thoroughly discuss the potential risks and benefits of any proposed intervention. In reality, the fetal radiation dose during a chest CT without contrast, is estimated at < 1mGy, which falls well below the maximum recommended cumulative exposure [Citation20,Citation21].
Multiple, asymmetric, and peripherally distributed ground-glass opacities with or without consolidation are the typical chest CT manifestations of early COVID-19 [Citation22] and these lesions seem to be more common in infected pregnant women. A systematic review in 2021 found ground-glass opacities in 77.2%, consolidation in 40.9%, and pleural effusion in 30% of infected pregnant patients [Citation23]. In this respect, our findings were quite similar to that of the published data. Ground-glass opacities, consolidations, and multilobar involvement were more commonly seen in those patients admitted to ICU and advanced lung involvement is known to be an important factor associated with poorer prognoses.
Among the currently available scores used to predict the severity of disease are the CT chest severity score (CT-SS) and the CT scan score. In the former, a threshold of 19.5/40 can predict severe disease with a sensitivity of 83.3% and a specificity of 94% [Citation16]. In the second alternative, the prognosis was noted to be very poor with a total score of 18/25 or higher [Citation17,Citation24]. In support of these findings, a previous report regarding Middle East respiratory syndrome-related coronavirus showed the predictive value of CT severity score for prognosis and short-term mortality [Citation25].
To the best of our knowledge, none of these CT scan scores have been formerly validated in pregnancy. Our study has confirmed the utility of a simple tool for predicting clinical deterioration in this obstetric population, by calculating two lung volumes (total and infected volumes) with the aid of artificial intelligence.
An important issue to consider in the COVID-19 work-up is the timing of the CT chest scan. Wang et al. showed that the peak of lung involvement in COVID-19 infection is 6–11 days after the onset of symptoms [Citation26], which corresponds with the transfer into the inflammatory phase. This was also confirmed by Pan et al. who defined four stages of the lung involvement. Stage 1 (within 4 days after onset of symptoms) is characterized by the presence of ground-glass opacities and a CT score of 2 ± 2. In stage 2 (5–8 days), there is an increase in the crazy-paving pattern and the CT score becomes 6 ± 4. In stage 3 (9–13 days), there is an appearance of consolidation and the CT score peaks at 7 ± 4. In stage 4 (≥14 days), there is a gradual resolution of consolidation and the CT score declines [Citation17]. In this study, the period from the onset of symptoms to CT scan was similar in the two groups, mainly during the peak of lung involvement, at stages 2 and 3.
The use of artificial intelligence software to measure the total lung volume, the infected lung volume, and the ratio of the infected to the total lung volumes on CT chest scans, when integrated into a management protocol for pregnant patients hospitalized with COVID-19 may aid in the prediction of deterioration and the need for ICU admission.
At certain gestations and where appropriate, this tool could also aid clinicians in the planning of delivery before the advent of significant clinical deterioration, thus optimizing obstetric management decisions [Citation9].
Several limitations of this study have been identified. Firstly, the number of the study population was quite small. This is explained by the low percentage pregnant women infected with SARS-CoV2 who require hospitalization and even fewer who will eventually benefit from ICU admission. A simple solution to this issue would have been to include multiple centers; however, the use of CT chest scan is not systematically included in the management of infected pregnant women at many centers in our area. The availability of artificial intelligence software is also quite limited. A second inherent weakness is the retrospective design, although, we tried to reduce the risk of bias by the use of routine real-time recording of clinical data in the patient’s electronic medical records and by blinding of the radiologist, who reported the CT chest scans, to the patients’ clinical course and outcomes. Finally, the software was unable to analyze the CT scan of one patient in the control group due to technical issues. This patient had no lung lesions, as reported by conventional analysis.
Conclusion
Total lung volume, infected lung volume, and infected-to-total lung volume ratio measured by artificial intelligence software on chest CT may predict ICU admission in symptomatic pregnant women hospitalized with COVID-19. Further large prospective studies are needed to validate the use of these parameters.
Author contributions
Conceptualization (D.A.B., F.D.L.); Data curation (D.A.B., F.D.L.); Formal analysis (D.A.B., F.D.L., A.C., J.C.J, M.M.C.); Funding acquisition (J.C.J., M.M.C.); Investigation (D.A.B., F.D.L., A.C.); Methodology (D.A.B., F.D.L., A.C.); Project administration (J.C.J., M.M.C.); Resources; Software (D.A.B., F.D.L.); Supervision (J.C.J); Validation (J.C.J., M.M.C.); Visualization (D.A.B., M.D.R.); Writing – original draft (D.A.B., F.D.L., A.C); Writing – review & editing (J.C.J., M.M.C.).
Supplemental Material
Download Zip (10.3 MB)Disclosure statement
No potential conflict of interest was reported by the author(s).
Data availability statement
The datasets analyzed during the current study are available from the corresponding author upon reasonable request.
Additional information
Funding
References
- Badr DA, Mattern J, Carlin A, et al. Are clinical outcomes worse for pregnant women at ≥20 weeks’ gestation infected with coronavirus disease 2019? A multicenter case-control study with propensity score matching. Am J Obstet Gynecol. 2020;223(5):764–768. doi: 10.1016/j.ajog.2020.07.045.
- Allotey J, Stallings E, Bonet M, et al. Clinical manifestations, risk factors, and maternal and perinatal outcomes of coronavirus disease 2019 in pregnancy: living systematic review and meta-analysis. BMJ. 2020;370:m3320. doi: 10.1136/bmj.m3320.
- Badr DA, Picone O, Bevilacqua E, et al. Severe acute respiratory syndrome coronavirus 2 and pregnancy outcomes according to gestational age at time of infection. Emerg Infect Dis. 2021;27(10):2535–2543. doi: 10.3201/eid2710.211394.
- Fang Y, Zhang H, Xie J, et al. Sensitivity of chest CT for COVID-19: comparison to RT-PCR. Radiology. 2020;296(2):E115–E117. doi: 10.1148/radiol.2020200432.
- Ai T, Yang Z, Hou H, et al. Correlation of chest CT and RT-PCR testing for coronavirus disease 2019 (COVID-19) in China: a report of 1014 cases. Radiology. 2020;296(2):E32–E40. doi: 10.1148/radiol.2020200642.
- Durhan G, Ardalı Düzgün S, Başaran Demirkazık F, et al. Visual and software-based quantitative chest CT assessment of COVID-19: correlation with clinical findings. Diagn Interv Radiol. 2020;26(6):557–564. doi: 10.5152/dir.2020.20407.
- Ufuk F, Demirci M, Uğurlu E, et al. Evaluation of disease severity with quantitative chest CT in COVID-19 patients. Diagn Interv Radiol. 2021;27(2):164–171. doi: 10.5152/dir.2020.20281.
- Ardali Duzgun S, Durhan G, Basaran Demirkazik F, et al. AI-based quantitative CT analysis of temporal changes according to disease severity in COVID-19 pneumonia. J Comput Assist Tomogr. 2021;45(6):970–978. doi: 10.1097/RCT.0000000000001224.
- Badr DA, Carlin A, Finianos ES, et al. COVID-19 infection during pregnancy: optimal timing for delivery. Revue Médicale de Bruxelles. 2022;43(4):424–430.
- Vlachodimitropoulou Koumoutsea E, Vivanti AJ, Shehata N, et al. COVID-19 and acute coagulopathy in pregnancy. J Thromb Haemost. 2020;18(7):1648–1652. doi: 10.1111/jth.14856.
- Wang H, Gao Y, Wu B. Drug therapy strategies for COVID-19 infection during pregnancy. Int J Clin Pharmacol Ther. 2021;59(3):175–181. doi: 10.5414/CP203817.
- Di Renzo GC, Giardina I. Coronavirus disease 2019 in pregnancy: consider thromboembolic disorders and thromboprophylaxis. Am J Obstet Gynecol. 2020;223(1):135. doi: 10.1016/j.ajog.2020.04.017.
- Horby P, Lim WS, Emberson JR, et al. Dexamethasone in hospitalized patients with Covid-19. N Engl J Med. 2021;384(8):693–704. doi: 10.1056/NEJMoa2021436.
- Dawes NW, Dawes GS, Moulden M, et al. Fetal heart rate patterns in term labor vary with sex, gestational age, epidural analgesia, and fetal weight. Am J Obstet Gynecol. 1999;180(1 Pt 1):181–187. doi: 10.1016/s0002-9378(99)70172-9.
- Nielsen Jeschke K, Bonnesen B, Hansen EF, et al. Guideline for the management of COVID-19 patients during hospital admission in a non-intensive care setting. Eur Clin Respir J. 2020;7(1):1761677. doi: 10.1080/20018525.2020.1761677.
- Yang R, Li X, Liu H, et al. Chest CT severity score: an imaging tool for assessing severe COVID-19. Radiol Cardiothorac Imag. 2020;2(2):e200047. doi: 10.1148/ryct.2020200047.
- Pan F, Ye T, Sun P, et al. Time course of lung changes at chest CT during recovery from coronavirus disease 2019 (COVID-19). Radiology. 2020;295(3):715–721. doi: 10.1148/radiol.2020200370.
- Badr DA, Al Hassan J, Salem Wehbe G, et al. Uterine body placenta accreta spectrum: a detailed literature review. Placenta. 2020;95:44–52. doi: 10.1016/j.placenta.2020.04.005.
- Ramadan MK, Sammak B, Mohsen A, et al. Fundal placenta percreta presenting as acute abdomen: a case report and the role of CT-scan in management. Int J Womens Health Wellness. 2019;5(108):2135–2474.
- Tremblay E, Thérasse E, Thomassin-Naggara I, et al. Quality initiatives: guidelines for use of medical imaging during pregnancy and lactation. Radiographics. 2012;32(3):897–911. doi: 10.1148/rg.323115120.
- Patel SJ, Reede DL, Katz DS, et al. Imaging the pregnant patient for nonobstetric conditions: algorithms and radiation dose considerations. Radiographics. 2007;27(6):1705–1722. doi: 10.1148/rg.276075002.
- Ng MY, Lee EYP, Yang J, et al. Imaging profile of the COVID-19 infection: radiologic findings and literature review. Radiol Cardiothorac Imag. 2020;2(1):e200034. doi: 10.1148/ryct.2020200034.
- Oshay RR, Chen MYC, Fields BKK, et al. COVID-19 in pregnancy: a systematic review of chest CT findings and associated clinical features in 427 patients. Clin Imag. 2021;75:75–82. doi: 10.1016/j.clinimag.2021.01.004.
- Francone M, Iafrate F, Masci GM, et al. Chest CT score in COVID-19 patients: correlation with disease severity and short-term prognosis. Eur Radiol. 2020;30(12):6808–6817. doi: 10.1007/s00330-020-07033-y.
- Das KM, Lee EY, Enani MA, et al. CT correlation with outcomes in 15 patients with acute Middle East respiratory syndrome coronavirus. AJR Am J Roentgenol. 2015;204(4):736–742. doi: 10.2214/AJR.14.13671.
- Wang Y, Dong C, Hu Y, et al. Temporal changes of CT findings in 90 patients with COVID-19 pneumonia: a longitudinal study. Radiology. 2020;296(2):E55–E64. doi: 10.1148/radiol.2020200843.