Estimating ventilation correlation coefficients in the lungs using PREFUL‐MRI in chronic obstructive pulmonary disease patients and healthy adults

Abstract Purpose Various parameters of regional lung ventilation can be estimated using phase‐resolved functional lung (PREFUL)‐MRI. The parameter “ventilation correlation coefficient (Vent‐CC)” was shown advantageous because it assesses the dynamics of regional air flow. Calculating Vent‐CC depends on a voxel‐wise comparison to a healthy reference flow curve. This work examines the effect of placing a reference region of interest (ROI) in various lung quadrants or in different coronal slices. Furthermore, algorithms for automated ROI selection are presented and compared in terms of test–retest repeatability. Methods Twenty‐eight healthy subjects and 32 chronic obstructive pulmonary disease (COPD) patients were scanned twice using PREFUL‐MRI. Retrospective analyses examined the homogeneity of air flow curves of various reference ROIs using cross‐correlation. Vent‐CC and ventilation defect percentage (VDP) calculated using various reference ROIs were compared using one‐way analysis of variance (ANOVA). The coefficient of variation was calculated for Vent‐CC and VDP when using different reference selection algorithms. Results Flow‐volume curves were highly correlated between ROIs placed at various lung quadrants in the same coronal slice ( r > 0.97) with no differences in Vent‐CC and VDP (ANOVA: p > 0.5). However, ROIs placed at different coronal slices showed lower correlation coefficients and resulted in significantly different Vent‐CC and VDP values (ANOVA: p < 0.001). Vent‐CC and VDP showed higher repeatability when calculated using the presented new algorithm. Conclusion In COPD and healthy cohorts, assessing regional ventilation dynamics using PREFUL‐MRI in terms of the Vent‐CC metric showed higher repeatability using a new algorithm for selecting a homogenous reference ROI from the same slice.