Repeatability and reproducibility of longitudinal relaxation rate in 12 small-animal MRI systems

Abstract

Background: Many translational MR biomarkers derive from measurements of the water proton longitudinal relaxation rate R₁, but evidence for between-site reproducibility of R₁ in small-animal MRI is lacking.

Objective: To assess R₁ repeatability and multi-site reproducibility in phantoms for preclinical MRI.

Methods: R₁ was measured by saturation recovery in 2% agarose phantoms with five nickel chloride concentrations in 12 magnets at 5 field strengths in 11 centres on two different occasions within 1-13 days. R₁ was analysed in three different regions of interest, giving 360 measurements in total. Root-mean-square repeatability and reproducibility coefficients of variation (CoV) were calculated. Propagation of reproducibility errors into 21 translational MR measurements and biomarkers was estimated. Relaxivities were calculated. Dynamic signal stability was also measured.

Results: CoV for day-to-day repeatability (N = 180 regions of interest) was 2.34% and for between-centre reproducibility (N = 9 centres) was 1.43%. Mostly, these do not propagate to biologically significant between-centre error, although a few R₁-based MR biomarkers were found to be quite sensitive even to such small errors in R₁, notably in myocardial fibrosis, in white matter, and in oxygen-enhanced MRI. The relaxivity of aqueous Ni²⁺ in 2% agarose varied between 0.66 s^-1 mM^-1 at 3 T and 0.94 s^-1 mM^-1 at 11.7T.

Interpretation: While several factors affect the reproducibility of R₁-based MR biomarkers measured preclinically, between-centre propagation of errors arising from intrinsic equipment irreproducibility should in most cases be small. However, in a few specific cases exceptional efforts might be required to ensure R₁-reproducibility.

Keywords: Biomarker; Error propagation; Hardware stability; MRI; Phantom; Relaxation time; Reproducibility.

Fig. 1

R₁ measurements (logarithmic axis) for each of centres A–L. Each centre made measurements on five 2% agarose phantoms with different Ni²⁺ concentrations. The six horizontal lines represent R₁ values calculated from the field-dependent relaxivities as explained in Table 2. There are two groups of three data points for each phantom at each centre representing, respectively, days 1 and 2, and RoIs (X,Y,Z) = (0,0,0), (10,0,0) and (0,0,12). Error bars are T₁ fit errors from ParaVision.

Fig. 2

Plot of [Ni²⁺] relaxivities in 2% agarose against field strength. Closed circles: this work, 19.36 ± 1.20 °C. Open circle: data from initial 1 5 T characterization of the phantom materials (see supplementary material), 21.5 °C. Standard error of fit is shown, although for B₀ between 1 5 T and 7 T the standard errors of between 0.19% and 0.48% are not evident as they are smaller than the size of the symbol. Other symbols: estimated from literature. +, parameter c₁ in [3], 22 °C. −, estimated, with standard error, from Fig. 1 in [9], 22 °C. ×, estimated, with standard error, from Fig. 4 in [15], 20 °C. ◇, ◻︎, estimated from Fig. 2 in [16], 19 °C and 22 °C respectively.