Volumetric quantification of lung MR signal intensities using ultrashort TE as an automated score in cystic fibrosis

Objectives The study aimed to validate automated quantification of high and low signal intensity volumes using ultrashort echo-time MRI, with CT and pulmonary function test (PFT) as references, to assess the severity of structural alterations in cystic fibrosis (CF). Methods This prospective study was performed in a single center between May 2015 and September 2017. Participants with CF completed clinical examination, CT, MRI, and PFT the same day during routine clinical follow-up (M0), and then 1 year after (M12) except for CT. Using MRI, percentage high (%MR-HSV), low (%MR-LSV), and total abnormal (%MR-TSV) signal intensity volumes were recorded, as well as their corresponding attenuation values using CT (%CT-HAV, %CT-LAV, %CT-TAV, respectively). Automated quantifications and visual Bhalla score were evaluated independently by two observers. Correlations were assessed using the Spearman test, comparisons using the Mann-Whitney test, and reproducibility using the intraclass correlation coefficient (ICC). Results A total of 30 participants were enrolled (median age 27 years, 18 men). At M0, there was a good correlation between %MR-HSV and %CT-HAV (rho = 0.70; p < 0.001) and %MR-LSV and %CT-LAV (rho = 0.60; p < 0.001). Automated MR metrics correlated to PFTs and Bhalla score (p < 0.05) while %MR-TSV was significantly different between CF with and without respiratory exacerbation (p = 0.01) at both M0 and M12. The variation of %MR-HSV correlated to the variation of FEV1% at PFT (rho = - 0.49; p = 0.008). Reproducibility was almost perfect (ICCs > 0.95). Conclusions Automated quantification of abnormal signal intensity volumes relates to CF severity and allows reproducible cross-sectional and longitudinal assessment.