Comparison of initial thin-section CT features in coronavirus disease 2019 pneumonia and other community-acquired pneumonia

Background Coronavirus disease 2019 (COVID-19) pneumonia caused similar symptoms to other community-acquired pneumonia (CAP). It is important to early quarantine suspected patients with COVID-19 pneumonia from patients with other CAP to reduce cross infection. The purpose of the study is to review and compare initial thin-section computed tomography (CT) features in patients with coronavirus disease 2019 (COVID-19) pneumonia and other community-acquired pneumonia (CAP).


Background
In the last December of 2019, a novel coronavirus, named as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), emerged in Wuhan, Hubei province, China [1,2]. It has raised world concern as it spread across the mainland of China and other countries around the world within several months.
World Health Organization (WHO) named the disease caused by SARS-CoV-2 as Coronavirus Disease 2019 (COVID-19) on February 12, 2020 [3]. As of March 12, 2020, a total of 80981 con rmed cases in China and 44279 con rmed cases in 117 countries outside China was reported [4]. Common presenting symptoms of COVID-19 pneumonia include fever, cough, fatigue, dyspnea, and diarrhea [5]. The real-time uorescence polymerase chain test of SARS-CoV2 RNA is regarded as the reference standard to make a de nite diagnosis of COVID-19 infection [6]. However, nucleic acid testing has some defects, such as time lag, shortage of supply, and high false negative rate [7]. CT has become an effective modality for the diagnosis and management of patients with COVID-19 pneumonia. There are several studies described the radiological features of COVID-19 pneumonia. Multifocal areas of GGA and/or patches of consolidation with bilateral, subpleural, and lower lung zonal preponderance were reported as the common chest CT features of COVID-19 pneumonia [8,9,10].
The newly onset lower respiratory syndromes with new lung in ltrate on chest radiograph, especially when supported by laboratory ndings and compatible physical examinations, is considered indicative for diagnosing community-acquired pneumonia (CAP) [11]. The clinical and radiological manifestations of COVID-19 may have some overlap with the manifestations of CAP caused by other pathogens, such as In uenza-A viral, Mycoplasma pneumoniae and Chlamydia pneumoniae and Streptococcus pneumoniae [12]. It is very important to early quarantine suspected patients with COVID-19 pneumonia from patients with other CAP to reduce cross infection. However, there are few reports of the comparison of CT features between the COVID-19 pneumonia and CAP caused by other pathogens. In our present study, we compared the pulmonary thin-section CT features of patients with COVID-19 pneumonia to those with other CAP. Methods 1.patient populations 24 consecutive adult patients with COVID-19 pneumonia and 28 consecutive adult patients with CAP caused by other pathogens at our institution between January 17th 2020 and February 28th 2020 were enrolled in our study. The diagnosis of COVID-19 was determined by positive real-time uorescence polymerase chain reaction of the patient's respiratory specimen for COVID-19 nucleic acid. Of the 24 patients, the chest thin-section CT scans were obtained on the same day when the initial throat swab test was performed. 28 patients with other community-acquired pneumonia (CAP) met the criteria of new symptoms of lower respiratory tract infection and new exudative lesions in chest CT, excluding patients with immunode ciency, important organ transplantation, hormone therapy for more than 2 weeks and lung cancer. 23 cases of pathogens were detected by sputum culture, pharyngeal swab, blood culture, serum antibody testing and other etiological examination, including Mycoplasma pneumoniae (n = 11), in uenza virus (n = 5), Klebsiella pneumonia (n = 2), Streptococcus pneumonia (n = 3), Staphylococcus aureus (n = 2). All the patients were excluded from the diagnosis of COVID-19 by twice negative RT-PCR test for COVID-19. Clinical, epidemiological and laboratory characteristics of the 52 patients were collected.

CT examinations
Thin-section CT scans were performed with multi-detector CT scanners (SOMATOM De nition AS, Siemens Healthineers, Germany; Discovery CT750 HD, GE Medical Systems, American; UCT 780 scanner, United Imaging, China). Thin-section CT images were obtained in a supine position during a breath hold at full inspiration. The scan range was from the lung apex to the top of the diaphragm. CT scans were performed with the following parameters: tube voltage, 120keV, tube current, 100-200mAs; pitch, 0.984~1.200; matrix, 512×512; FOV 350 mm×350 mm. The images were reconstructed into 0.625mm, 0.5mm and 1mm slice thickness, with an interval of 0.625mm, 0.5mm and 1mm for GE, United Imaging and Siemens images, respectively. Images were read and captured at lung window setting (window width, 1200-1600 HU; window lever 500 to 700 HU) and mediastinum setting (window width 350-400HU, window level, 20-40HU).

Interpretation of images
The CT images were interpreted independently in random order by two radiologists (with 5 years and 20 years' experience in chest CT imaging, respectively) who were unaware of the underlying diagnoses. The nal decision was reached by consensus when an evaluation differed.
CT images were analyzed for the following presented radiological ndings: (1) density of lesions: pure ground-glass attenuation (GGA), pure consolidation, mixed GGA and consolidation. GGA was de ned as a hazy area showing increased attenuation without obscuring the underlying vascular markings. Consolidation was de ned as parenchymal opaci cation that obscured the underlying vessels [13,14].
(2) shape of lesions: the shape of main parenchymal abnormality (GGA and consolidation) in each patient was classi ed as patches with long axis parallel to pleural, patches with long axis parallel to the bronchovascular, and nodules with round morphology. (3) other ndings: crazy-paving appearance, vascular dilatation within the lesions, tree-in-bud pattern, linear opacity, cavitation, halo sign/reversed halo sign, pleural effusion and mediastinal or hilar lymph node(s) enlargement. Crazy-paving appearance was characterized by a network of thickened interlobular or intralobular interstitial superimposed on a background of GGA [15]. Tree-in-bud pattern was used to describe the appearance of the constellation of small centrilobular nodules and concomitant branching opacities [16]. Linear opacity referred to intralobular septal line and parenchymal bands. Cavitation de ned as a low-attenuation area within pulmonary consolidation [13]. Halo sign was de ned as a CT nding of GGA surrounding a nodule or mass, and reversed halo sign was characterized by a focal GGA surrounded by a ring of consolidation [17,18]. Mediastinal and hilar lymphadenopathy were de ned as lymph node with a short axis dimension of 10mm.
In addition, the distribution patterns of parenchymal abnormality were also analyzed. We recorded whether the main parenchymal lesions locate bilaterally or unilaterally. The cross-sectional distribution of each patient was classi ed as central, peripheral, or randomly distribution. The disease was classi ed as central distribution if the main lesions were predominantly located within the inner third of the lung, peripheral distribution if they were predominantly located within the outer third of the lung, and randomly distribution if the lesions presented with no predominant distribution. The zonal involvement was classi ed as upper, middle, and lower zone by the following three lung levels: the upper zone was de ned as lung eld above the tracheal carina, the middle zone between the carina and inferior pulmonary vein, and the lower zone below the inferior pulmonary vein.

Statistical analyses
Statistical analyses were performed using IBM SPSS Statistics Software (version 25; IBM, New York, USA). Quantitative data were expressed as mean ± standard deviation (minimum-maximum). Counting data were expressed as the count (percentage of the total). The mean ages, the mean days of CT scan after symptoms onset, the white blood cell counts and lymphocyte counts of COVID-19 and other CAP patients were compared using Student's t-test.
The frequencies of crazy-paving appearance and vessel dilatation were signi cantly higher in patients with COVID-19 pneumonia compared with patients with other CAP (p = 0.031 and p = 0.000, respectively). Conversely, the frequencies of pure consolidation, tree-in-bud sign and pleural effusion were signi cantly higher in patients with CAP than those in patients with COVID-19 pneumonia (p = 0.002, p = 0.000 and p = 0.048, respectively). There were no signi cant differences in the shape of main lesions and other CT features, including pure GGA, mixed GGA with consolidation, air bronchogram, linear opacities, halo sign/reversed halo sign, cavitation and lymphadenopathy between the groups of COVID-19 pneumonia and other CAP.
The frequency of bilaterally involvement and peripherally distribution was signi cantly higher in COVID-19 group than in the CAP group (p = 0.029, and p = 0.009, respectively). No signi cant differences in zonal involvement between the two groups were found.

Discussion
In December 2019, a major outbreak of respiratory disease later named as COVID-19 was reported by health authorities in Wuhan, Hubei province, China [19]. Followed by an exponential growth in a few weeks, the COVID-19 emerged as a new agent of community-acquired pneumonia, attracting extensive growth around the world. Early detection of suspected patients is the important way to control the spread of the disease. However, the caused symptoms of the virus are similar to those of other CAP (e.g., fever, cough, sputum), and screening for COVID-19 among these patients with CAP is not cost-effective and may lead to unnecessary social-disruption. It is very important for clinics to identify the infected patients.
In this study, we comprehensively veri ed the clinical characteristics and CT features of COVID-19 pneumonia and other CAP, aimed to identify common requested clinical data and CT imaging variables that might discriminate COVID-19 patients from newly diagnosed CAP with other pathogens.
We report 24 patients with laboratory-con rmed COVID-19 pneumonia and 28 patients with CAP excluded COVID-19 infection. Current reports show that COVID-19 is generally susceptible. The onset age of COVID-19 in this group is from 14-87 years old, similar to previous study reports that infection can occur at all ages [20]. There was a slight predilection for males over females in patients with COVID-19 pneumonia in our cohort. However, another study reported s striking male predilection with 67(68%) of 99 COVID-19 patients being male [21]. This discrepancy might be due to the small cohorts. No obvious predilection was observed for patients with other CAP in our study, and there was no signi cant difference in gender distribution between the two groups. In this study, we found that there are some similarities in clinical manifestations for patients with COVID-19 pneumonia and patients with other CAP. Most respiratory infections are presented with nonspeci c acute symptoms such as fever, fatigue, cough, sputum, dyspnea, and diarrhea, which make it di cult to differentiate COVID-19 pneumonia from CAP. In our study, the white cell count of patients with COVID-19 pneumonia was signi cantly lower than the patients with CAP, consistent with the laboratory test results for most respiratory viral infections [22].
However, there was no signi cant difference in lymphocyte count between patients with COVID-19 pneumonia and CAP, different from the previous study [9], which may be related to the small sample size and the complex pathogens in patients with CAP.
Radiological examinations play an important role in timely detection and management of pneumonia. Conventional chest radiography is usually the rst imaging technique performed to evaluate for CAP. However, since chest radiography is not sensitive for the detection for GGA, it was not recommended as the rst line imaging modality for COVID-19 pneumonia [23]. National Health and Health Commission of China had recommended the CT manifestations as important evidence of clinical diagnosis in Hubei province [24]. In our study, the characteristic CT ndings in COVID-19 pneumonia include pure GGA or mixed GGA and consolidation, crazy paving appearance and vascular dilatation within the lesions, patches parallel to pleura or nodules with rounded morphology, with a peripheral and lower zone predominance and bilateral involvement. Our results agree with those reported elsewhere [9,10,25]. On the other hand, CT ndings in patients with other CAP include patches of ground glass attenuation or multifocal areas of consolidation along the axis interstitium, tree-in-bud pattern, with a random distribution and middle zone predominance and unilateral involvement. However, these ndings are nonspeci c and can be manifestations of a number of different types of bacterial, viral or atypical pneumonia.
Findings in our study demonstrated that ground glass opacities, mixed ground glass opacities and consolidation, air bronchogram were of consistent common initial features in both COVID-19 and other CAP group. On the contrary, cavitation, halo sign/reversed halo sign, and lymphadenopathy were rare features in both groups. The presence of crazy paving pattern, vessel dilatation, bilateral involvement and peripheral distribution occurred more frequently in the group of COVID-19 pneumonia. However, pure consolidation, tree-in-bud sign and pleural effusion occurred signi cantly more frequently in the group of other CAP. Therefore, these CT features and distribution pattern can be used as markers in differentiating patients with COVID-19 from patients with other CAP during the initial screening. Despite these seemingly positive ndings, one should keep in mind that these CT features are nonspeci c and can be noticed in pneumonias caused by different types of pathogen and other non-infectious diseases.
Therefore, correlation with clinical, epidemiological and laboratory characteristics will play an important role in searching for a diagnosis.
Finally, a number of limitations need to be noted regarding the present study. First, this is a retrospective study with limited cases. Most of the patients with COVID-19 enrolled were mild or general cases. Therefore, it might be di cult to determine what are the characteristic thin-section CT ndings for differentiating COVID-19 pneumonia from CAP caused by other pathogens. Second, we mainly focus on the clinical and imaging features at the initial medical contact, while the dynamic changes with appropriate therapy are de nitely helpful to differential these two diseases. Third, the CT image interpretation was obtained by two radiologists rendering consensus and did not have pathological correlation.
In summary, the main purpose of this study was to characterize and compare the clinical characteristics and thin-section CT features between COVID-19 pneumonia and other CAP in the early course of the diseases. The overlapping thin-section CT features such as pure GGA, mixed GGA with consolidation, indicate di culty in obtaining high accuracy in differentiation between the two groups. The data used in this study was anonymized before its use.

Consent for publication
Not applicable.

Availability of data and materials
All data generated and/or analyzed during the current study are available from the corresponding authors on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Funding
This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors.