qMRI aims to produce measurements directly related to tissue microstructure with high diagnostic and research value that are independent of scanner and acquisition protocol. Maps of the myelin-sensitive longitudinal relaxation rate (R₁) can be calculated from two high resolution 3D FLASH acquisitions with different excitation flip angles. However, this measure will contain bias due to transmit field (B₁⁺) inhomogeneity, an increasing problem at higher field strengths. Additional calibration data are typically acquired to correct for B1⁺ inhomogeneity. However, the reference scan is time consuming and difficult to implement, particularly in a clinical setting. To facilitate qMRI in such a clinical setting, the UNICORT post-processing approach was developed as an alternative that does not require calibration data¹. It employs a probabilistic framework that incorporates a physically informed generative model of smooth transmit field inhomogeneities and their multiplicative effect on R₁ estimates. However, different systems may require different priors for the B₁⁺ bias field, depending on the particular transmit coil used. Here we show that these parameters can be estimated using a linear relaxometry model framework², without the need to acquire B₁⁺ mapping data.

Data were acquired on two 3T Verio systems (Siemens Healthcare) as part of whole brain, multi-parameter mapping protocols³ at two sites without (U. Zuerich) or with (CMU) B₁⁺ mapping data.

(A) Determining Optimal Parameters Without B₁⁺ Calibration:

Data were acquired on three participants with 1mm isotropic resolution⁴. In addition to maps of R₂* and MT, R₁ maps were created using the UNICORT approach while varying the regularisation (-log₁₀(k)=[1:1:5]) and full-width at half maximum (FWHM=[30:10:70]) parameters constraining the bias field and fit. In each case, the UNICORT-corrected R₁ map was combined with the original MT and R₂* maps within the linear relaxometry model². The absolute residuals of the relaxometry model were integrated within a mask defining white matter and cortical gray matter for each UNICORT-corrected R₁ map. The settings with the lowest residuals were deemed optimal, since they best respect the established inter-relation between quantitative parameters.

(B) Testing Optimised Parameters against B₁⁺ Calibration⁵:

Data were acquired on 8 participants with 0.8mm isotropic resolution. Three R₁ maps were calculated: 1) using B₁⁺ calibration data 2) using UNICORT parameters determined to be optimal for a Trio system¹ and 3) using the parameters deemed optimal for the Verio system in step (A). Histograms were used to analyse the error of the UNICORT-corrected R1 maps with respect to the B₁⁺-corrected R₁ maps.

Optimum parameters for the Verio system determined in (A) were k=10^-3 and FWHM=30mm. Example B₁⁺- and UNICORT-corrected R₁ maps are shown in fig. 1 (note the differential anterior-posterior gradient). Histograms of the error across the cohort are summarised in fig. 2. The peak error was reduced from 8.2% to 0.3% using the parameters deemed optimal for the Verio system in (A).By using the linear relaxometry model, optimisation of UNICORT parameters can be achieved without the acquisition of B₁⁺ calibration data, though such data was used for verification here. This approach demonstrates the importance of, and provides a solution to, optimising UNICORT priors for the particular transmit coil used in order to minimise bias in R₁ maps. Nonetheless, maximum accuracy will be achieved through full calibration and correction of the B₁⁺ transmit field and should be the preferred approach where possible.