Correlation of MR pulmonary perfusion in patients with COVID-19 with quantitative assessment of acute phase CT images

Abstract

INTRODUCTION: In the last decade, there has been an increased interest in new diagnostic techniques for assessing quantitative values in radiology. In particular, accurate quantitative values may be useful to assess anatomical or physiological changes in the lungs in patients with previously treated COVID-19 infection. OBJECTIVE: To test a quantitative semi-automated algorithm for CT imaging in patients with confirmed COVID-19 infection and to compare the results to MR lung perfusion after coronavirus infection. MATERIALS AND METHODS : The data from 100 chest CT scans of patients with COVID-19 were retrospectively analyzed. 3D segmentation of the lungs was carried out with automatic counting of the number of separated pixels in each slice. For quantitative data analysis, classification based on the density value of each pixel according to the Hounsfield scale was used. The obtained data were compared with quantitative parameters of pulmonary MR perfusion in these patients. Statistics . Generalized additive model with beta distribution, Spearman correlation coefficient was used, Benjamini-Yekuteli correction was used to correct obtained p-values. Comparisons were determined as statistically significant when p<0.05. RESULTS : There was a correlation between quantitative CT data (fractions of pixels corresponding to non-ventilated and hypo-ventilated lung tissue) and the distribution of CT data into groups according to an empirical visual scale. We obtained a correlation between the functional perfusion parameters and the CT images: rMTT — 0.35 (p=0.001), rPBF — 0.23 (p=0.038) and rPBV — 0.35 (p=0.001). DISCUSSION: Using the algorithm of quantitative semi-automatic processing of CT-images suggested in this work allows to obtain numerical data, objectively reflecting percentage of affected lung tissue, that is especially relevant for diagnostics of COVID-19 pneumonia. The obtained correlation between functional perfusion parameters and CT picture can be potentially a marker of the lung pathological changes after COVID-19 pneumonia, that requires further investigations. CONCLUSION: Quantitative processing of CT-images allowed to correctly compare the CT scans of lung lesions in COVID-19 with MR lung perfusion data after COVID-19 infection which could potentially be of prognostic value.