Abstract
The purpose of this study was to produce metabolite-specific T(2) and concentration maps in a clinically compatible time frame. A multi-TE 2D MR spectroscopic imaging (MRSI) experiment (multi-echo single-shot MRSI [MESS-MRSI]) deployed truncated and partially sampled multi-echo trains from single scans and was combined with simultaneous multiparametric model fitting. It was tested in vivo for the brain in five healthy subjects. Cramér-Rao lower bounds (CRLB) were used as the measure of performance. The novel method was compared with (1) traditional multi-echo multi-shot (MEMS) MRSI and, as proof of concept, with (2) a truncated version of MEMS, which mimics the MESS acquisition (MESS-mocked) on the original fully sampled MEMS dataset. MESS-MRSI simultaneously yields concentration and T(2) maps with a nominal voxel size of ~2 cm(3) with a 16 × 16 FOV matrix in 7 min scan time. The estimated values not only align well with the equivalent mocked experiment but are also in good agreement with the traditional threefold longer MEMS acquisition. The MESS-MRSI scheme extends former findings for single-voxel MESS, with improvements in CRLB ranging from 17% to 45% for concentrations and 10% to 23% for T(2)s when compared to traditional MEMS. This finding suggests that concentrations and T(2) times can be reliably estimated in a multi-echo spectroscopic imaging exam by trading off spectral resolution (for some of the acquired TEs) with a significant reduction in scan time, as long as (1) an appropriate bidimensional frequency-TE model is deployed and (2) one TE is sampled in full. Thus, high spectral resolution information can be injected to the partially sampled TEs during fitting by prior knowledge from the one fully sampled TE. Tissue-type and regional distributions of 16 metabolite concentrations align well with the literature, and T(2) distributions for five major metabolites are described by region and tissue. The novel MRSI acquisition strategy, based on partially sampled single-shot multi-echo trains twinned to multiparametric fitting, is optimally suited to provide simultaneous 2D concentration and T(2) maps in clinic-compatible scan times. MESS principles allow embedding advanced MRSI techniques to further improve speed, coverage, or resolution. Preliminary findings from a cohort of five subjects reveal correlations between T(2) relaxation times and the relative fraction of gray/white matter, suggesting tissue-type-dependent microstructural changes.