基于薄层CT病变区肺质量参数与新型冠状病毒肺炎临床特征的相关性

目的：通过薄层CT容积重建计算新型冠状病毒肺炎（COVID-19）病变区的肺质量，并探讨其与早期临床特征的相关性。方法：连续收集2020 年1月19日至2月31日48例COVID-19患者的初诊DICOM格式CT薄层资料，传输至后处理软件进行人工智能自动分割病灶并得到病灶的体积和平均密度，计算病变肺的质量。分析病变肺的质量与性别、年龄、体温、发病时间、实验室检查指标的相关性，比较不同临床特征组别间的质量差异。结果：患者共计48例，年龄28～81（45.2±16.5）岁，男26例，女22例，中位发病时间3 d。病变肺质量与患者的体温、发热天数和C反应蛋白值存在显著相关性（r =0.373、0.360、0.713，均P ＜0.05）。中、高度发热组病变肺质量65.7（41.8，101.7）g]明显大于低度发热组10.0（3.1，49.3）g]和无发热组5.5（1.0，16.0）g]，差异均有统计学意义（P ＜0.05）；发病时间≥5 d组病变肺质量45.7（16.4，86.3）g]大于发病时间≤4 d组10.0（2.6，39.3）g]；C反应蛋白升高组病变肺质量45.4（11.3，88.4）g]大于正常组4.0（1.2，14.2）g]，差异均有统计学意义（P ＜0.05）。结论：病变肺质量与COVID-19患者的临床特征和实验室指标具有显著相关性，可作为反映病毒性肺炎严重程度的CT定量参数之一。

Correlation between quality parameter based on thin-section CT and clinical symptoms of COVID-19 pneumonia

Objective: To calculate the lung quality of CT lesions in patients with COVID-19 pneumonia,and explore its correlation with early clinical features and laboratory indicators. Methods: The DICOM format of thin-section CT of 48 patients with COVID-19 during January 19, 2020 to February 31, 2020 were collected and transmitted to the post-processing software for segmentation of lesions by artificial intelligence. The volume and average density of the lesion were obtained, and the mass of the diseased lung calculated. The correlation between the quality of diseased lungs and gender, age, body temperature, onset time and laboratory tests was explored, and the quality difference among groups with different clinical characteristics was compared. Results: Forty-eight patients, aged (45.2±16.5)years, 26 males and 22 females, had a median onset of 3 days. The lung quality was significantly correlated with the patient’s body temperature, fever days, and C-reactive protein values (P<0.05),with r values of 0.373, 0.360, and 0.713, respectively. The lung quality in the moderate or high fever group was 65.7 (41.8, 101.7)g, which was significantly larger than that in the low fever group 10.0 (3.1, 49.3)g] or the nonfever group 5.5 (1.0, 16.0)g]. The quality in patients with initial CT scan≥5 days from illness onset was 45.7(16.4, 86.3)g, which was significantly higher than patients≤4 days from illness onset 10.0 (2.6, 39.3)g]; the quality in patients with high C-reactive protein was 45.4 (11.3, 88.4)g, significantly higher than patients with normal Creactive protein 4.0 (1.2, 14.2)g]. All differences were statistically significant (P<0.05). Conclusion: The quality of the diseased lung is significantly correlated with the clinical symptoms and laboratory indicators of COVID-19 patients, which can be used as one of the effective CT quantitative parameters reflecting the severity of viralpneumonia.