Repeatability of Phase‐Resolved Functional Lung (<scp>PREFUL</scp>)‐<scp>MRI</scp> Ventilation and Perfusion Parameters in Healthy Subjects and <scp>COPD</scp> Patients
Gesa H. Pöhler, Filip Klimeš, Lea Behrendt, Andreas Voskrebenzev, Cristian Crisosto González, Frank Wacker, Jens M. Hohlfeld, Jens Vogel‐Claussen
Abstract
Background Free‐breathing phase‐resolved functional lung (PREFUL)‐MRI may be useful for treatment monitoring in chronic obstructive pulmonary disease (COPD) patients with dyspnea. PREFUL test–retest reliability is essential for clinical application. Purpose To measure the repeatability of PREFUL‐MRI ventilation (V) and perfusion (Q) parameters. Study Type Retrospective and prospective. Population A total of 28 COPD patients and 57 healthy subjects. Field Strength/Sequence 1. 5T MRI / 2D spoiled gradient echo imaging. Assessment V and Q lung parameter maps based on three coronal slices were obtained at baseline and after 14 days (COPD patients) or after a short pause outside the scanner (healthy subjects). Regional ventilation (RVent) and imaging flow volume loops by cross‐correlation (ccVent) were quantified. Q was normalized to the signal of the main pulmonary artery (Q N ) and quantified (Q Quant ). Pulmonary pulse wave transit time (pPTT), voxel‐by‐voxel (regional), and whole lung (global) ventilation defect percentage based on RVent (VDP RVent ) and ccVent (VDP ccVent ), perfusion defect percentage (QDP), and ventilation/perfusion match based on RVent (VQM RVent ) and ccVent (VQM ccVent ) were calculated. Statistical Tests Regional V and Q were analyzed globally for each subject. Each parameter's median of scans 1 and 2 were assessed by Wilcoxon sign rank test. A parameter's repeatability was analyzed by Bland–Altman analyses, coefficients of variation, intraclass correlation coefficients (ICC), and power calculations. The regional voxel repeatability was examined by calculating the Sørensen–Dice coefficient. Results There was no bias and no significant differences between the first and second MRI for any parameters ( P > 0.05). Coefficient of variation ranged from 2.26% (ccVent) to 19.31% (QDP), ICC from 0.93 (QDP) to 0.60 (pPTT), the smallest detectable difference was 0.002 ccVent. Regional comparison showed the highest overlap (84%) in VDP RVent in healthy voxels and the lowest (53%) in VDP ccVent defect voxels. Data Conclusion V and Q PREFUL‐MRI parameters were repeatable over two scan sessions in both healthy controls and COPD patients. Level of Evidence 2 Technical Efficacy Stage 2