Quantification of CT images in 83 cases of COVID-19 / 预防医学
Journal of Preventive Medicine
; (12): 568-572, 2021.
Article
in Chinese
| WPRIM (Western Pacific)
| ID: wpr-877284
Responsible library:
WPRO
ABSTRACT
Objective@#To quantitatively analyze the chest computerized tomography ( CT ) images of coronavirus disease 2019 ( COVID-19 ) cases by automatic artificial intelligence ( AI ) system, so as to provide the basis for the prediction of severe cases and early clinical intervention.@*Methods@#Eighty-three confirmed cases of COVID-19 from January 23 to February 14, 2020 in Wuchang Hospital of Wuhan were selected and the clinical data were collected. According to the diagnosis and treatment Plan of COVID-19 (seventh trial), the patients were divided into an ordinary group and a severe group. The parameters of chest CT images were quantified by the automatic AI system, and the CT imaging features of two groups were compared.@*Results@#There were 46 cases in the ordinary group and 37 cases in the severe group, with the age of ( 62.68 ±13.69 ) years and ( 50.52 ±12.45 ) years, respectively. The percentages of total pulmonary lesions, the lesion volume of bilateral lungs, the lesion volume of right lower lung, the left lung volume and the right lung volume from -300 to -200 Hu [median (inter-quartile range)] were 19.80% ( 21.69% ), 622.87 ( 1 145.73 ) cm3, 205.73 ( 246.95 ) cm3, 26.50 (21.20) cm3 and 38.02 (48.78) cm3 in the severe group, which were significantly different from 9.78% ( 13.24% ), 333.55 ( 401.77 ) cm3, 126.02 (164.21) cm3, 21.43 (13.11) cm3 and 26.92 ( 18.04 ) cm3 in the ordinary group ( P<0.05 ). The volume of pulmonary lesions reached the peak from 10 to 16 days after infection.@*Conclusion@#The lung lesions in severe cases of COVID-19 are large, especially in the right lower lung, and need to be closely monitored from 10 to 16 days after infection for early warning of severe cases.
Full text:
Available
Database:
WPRIM (Western Pacific)
Language:
Chinese
Journal:
Journal of Preventive Medicine
Year:
2021
Document type:
Article