Visual and automatic evaluation of the volume of lung damage on computer tomography with pneumonia caused by COVID-19

Abstract

The paper assessed various signs of lung tissue damage and the extent of its damage using automatic and empirical methods on CT of the chest organs in pneumonia caused by COVID-19. We analyzed 198 CT scans of the chest of patients with confirmed COVID-19 pneumonia of varying severity. The visual assessment was performed by a radiologist with 8 years of experience in thoracic radiology. The presence of CT patterns was assessed: ground-glass opacities, consolidation, reticular changes, “crazy paving”. Automatic analysis of CT scans performed in the “service of automatic diagnosis of patients with COVID-19”, which is determined by the amount of light and the amount of “ground glass” and consolidation. The automated analysis of computed tomograms was carried out in the “Service for automatic assessment of the severity of lung injury in patients with COVID-19”, which determined the volume of the lungs, the volume of the lesion “frosted glass” and consolidation. The average lesion volume in all groups was 19.1 % according to visual analysis. According to automatic analysis, the prevalence was 11.1 %. The visual assessment of the volume of the lesion is more pronounced compared with the automated assessment, in which the information is more objective.

Keywords

• covid-19
• пневмония
• 3d-моделирование
• мскт
• cad
• нейросети