Ways of transformation of typical X-ray signs of community acquired pneumonia of viral etiology (COVID-19) according to radiomics data

Abstract

For 3 years since the announcement of the coronavirus disease (COVID-19) pandemic, doctors around the world have been studying the complications caused by different strains of SARS-CoV-2. To study the structure of the lung parenchyma in patients with a complicated course of community acquired viral pneumonia of COVID-19 and different ways of transformation, the most informative is the digital software processing of computed tomography (CT) images of the chest organs (CT).
 Objective — to investigate the ways of transformation of typical radiological signs in patients with community-acquired pneumonia of viral etiology (COVID-19) and the possibility of their transformation into bronchioloalveolar cancer (BAC) by the radiomics method.
 Materials and methods. Chest CT data in the dynamics of 112 patients with a complicated course of community-acquired viral pneumonia COVID-19 were analyzed. Chest CT was performed on an Aquilion TSX-101A Tochiba scanner (Japan) with subsequent digital software processing of CT images using the Dragonfly program from Obyect Research Systems (ORS), Montreal, Canada. The diagnosis of BAC was made based on the data of the pathomorphological examination. Transbronchial biopsy of lung tissue was performed during diagnostic fibrobronchoscopy.
 Results and discussion. As a result of the analysis of possible ways of transformation of typical X-ray changes of COVID-19 community-acquired pneumonia, we identified 3 main ways. In 71 (64.0 %) subjects, according to the chest CT scan, there was gradual resorption of pathological changes and recovery of the lung parenchyma. In 35 (31.2 %) patients, the formation of signs of «vanishing lung syndrome» was detected. 5 (4.5 %) patients were diagnosed with BAC according to the CT scan and pathomorphological examination. Digital software processing of chest CT in dynamics allows to track the process of transformation of the lung parenchyma structure in patients with a complicated course of COVID-19 community-acquired viral pneumonia into BAC and in some cases to confirm the secondary nature of the oncological process.
 Conclusions. Digital software processing of the chest CT data is a highly informative research method that clearly reflects the morphological structure of the lung parenchyma and allows diagnosis and differential diagnosis of diseases.