Introduction

GRASE is the gold standard sequence to quantify myelin from T2 maps, however it requires long scan times. Quantitative multiparametric MRI can be a faster alternative with additional quantitative values.

We aim to evaluate in a phantom the accuracy and repeatability of T2 maps obtained from 3D-GRASE, with different acceleration settings and from alternative multiparametric sequences.

Methods

The EUROSPIN Phantom with 12 vials with different T2 values was scanned using a GE PET/MR 3T system. Seven Spin Echo images TE=[10,30,80,130,180,300,500]ms for reference T2 values (Reference), 3D-GRASE with and without parallel imaging and compressed sensing acceleration, 2D-MAGiC, 2D-QTI, and 2D-MRF were acquired. 3D-GRASE, MAGiC, QTI, MRF were scanned twice to assess repeatability. The T2 maps were generated automatically on the scanner (MAGiC, QTI, MRF) or offline using Matlab (Reference, GRASE). To compare the T2 maps quantitatively, we calculated the percentage difference (Δ%) with respect to the mean value (μ) of the Reference for each vial: 100*(μ_seq– μ_ref)/μ_ref. To evaluate the test/retest repeatability, the Coefficients of Variation (CoV) were calculated.

Results

Table 1 shows for each sequence the median and quartile range of Δ% and CoV over all vials, and scan times. Figure 1 shows the T2 maps.

Conclusions

Compared to the Reference, 3D-GRASE tends to overestimate the T2 values, while multiparametric approaches underestimate them. Among the multiparametric sequences, MAGiC is the most accurate and repeatable.

Based on these preliminary results, we will optimize the sequences for the T2 ranges of brain in terms of efficiency.