Standardizing quantitative muscle MRI: an inter-vendor evaluation in healthy adults

Quantitative muscle MRI (qMRI) is applied to detect muscular alterations, particularly in neuromuscular diseases (NMDs). As studies on rare NMDs benefit from multi-center designs, inter-vendor reproducibility must be assessed to ensure data comparability. However, manufacturer-related variability has been limited investigated. This study evaluates inter-vendor reproducibility of qMRI parameters in healthy subjects.

Ten healthy volunteers were investigated within 1 week on two 3T MRI systems (Siemens, Philips). Twelve leg muscles were analyzed using Dixon (fat fraction, FF), T2 mapping (water T2, wT2), and diffusion-weighted-sequence (diffusion parameters: fractional anisotropy (FA), mean, radial, axial diffusivity [MD, RD, λ₁]). Reproducibility was assessed using intraclass correlation coefficient (ICC), within-subject coefficients of variation, t tests, and Bland-Altman analysis.

ICCs demonstrated moderate to excellent reproducibility for most parameters (ICC_FF = 0.923, ICC_wT2 = 0.792, ICC_FA = 0.777, ICC_MD = 0.708, ICC_RD = 0.850), except for λ₁ (ICC = 0.361). Significant inter-vendor differences were observed for all parameters (p < 0.05), with systematic biases ranging from 44.2% for FF to 3.6% for FA.

qMRI parameters were generally reproducible in healthy volunteers. However, absolute values differed between manufacturers and require correction. These differences should be considered to minimize manufacturer-related discrepancies and to facilitate data pooling in multi-center and longitudinal studies.