Advanced magnetic resonance spectroscopic neuroimaging: Experts' consensus recommendations

Abstract

Magnetic resonance spectroscopic imaging (MRSI) offers considerable promise for monitoring metabolic alterations associated with disease or injury; however, to date, these methods have not had a significant impact on clinical care, and their use remains largely confined to the research community and a limited number of clinical sites. The MRSI methods currently implemented on clinical MRI instruments have remained essentially unchanged for two decades, with only incremental improvements in sequence implementation. During this time, a number of technological developments have taken place that have already greatly benefited the quality of MRSI measurements within the research community and which promise to bring advanced MRSI studies to the point where the technique becomes a true imaging modality, while making the traditional review of individual spectra a secondary requirement. Furthermore, the increasing use of biomedical MR spectroscopy studies has indicated clinical areas where advanced MRSI methods can provide valuable information for clinical care. In light of this rapidly changing technological environment and growing understanding of the value of MRSI studies for biomedical studies, this article presents a consensus from a group of experts in the field that reviews the state-of-the-art for clinical proton MRSI studies of the human brain, recommends minimal standards for further development of vendor-provided MRSI implementations, and identifies areas which need further technical development.

Keywords: brain; magnetic resonance spectroscopic imaging; review.

FIGURE 1

MRSI obtained at 7 T for a subject with an anaplastic oligoastrocytoma. Shown are the T1-weighted postcontrast MRI, total N-acetylaspartate (tNAA), total choline (tCho), total creatine (tCr), glutamate (Glu), glutamine (Gln), myo-inositol (mIns) and sum of lipids (Lip). The single-slice FID-MRSI was acquired in 6 min with 6-fold accelerated phase-encoding, voxel size of 3.4 x 3.4 x 8 mm³ and TR/AD of 600/1.3 ms. From Trattnig et al

FIGURE 2

Volumetric MRSI measurements of 2HG at 3 T. (A) 2HG maps for a glioblastoma, postsurgery. Maps were obtained using PRESS at TE = 97 ms with 3D phase encoding and are superimposed on the FLAIR MRI. From Choi C et al. (B) 2HG maps for a glioblastoma obtained using an editing measurement based on MEGA-LASER at TE = 68 ms and 3D stack-of-spirals. The red contours indicate the tumor margins on FLAIR image while the blue and green contours show the radiotherapy dose. From Jafari-Khouzani et al

FIGURE 3

Whole-brain high-resolution metabolite maps taken at 7 T using FID-detection (acquisition delay 1.3 ms, TR = 280 ms), concentric-ring k-space sampling and reconstruction to 80 x 80 x 47 voxels. Total acquisition time was 15 minutes. Additional details can be found in Hingerl et al

FIGURE 4

Example of the use of quality maps to identify regions with spectra of inadequate quality, for two slices from a volumetric EPSI acquisition at 3 T, for TE = 50 ms. Shown are (A) the postcontrast T1-weighted MRI and (B) the Cho map, which shows increased signal corresponding to the location of a glioblastoma, together with several other bright signal regions. In (C) are shown the spectral quality maps, with white regions corresponding to a spectral linewidth of ≤13 Hz and gray regions for voxels with a linewidth of >13 Hz. Poor quality voxels can be removed (D), although spatial information is more clearly conveyed when combined as an overlay image with the MRI (E)

FIGURE 5

(A) Example volume-selected 3 T MRSI result for a glioblastoma showing the contrast-T1 MRI, the Choline-to-NAA Index (CNI) map, and the NAA and Cho metabolite maps. The selected volume is indicated by the yellow rectangle. The CNI map, shown as a color overlay, identifies all voxels with significant differences of the ratio of NAA and Cho. The color bar represents the numerical values of the CNI map. (B) Example Z-score maps for NAA, Cho and Cho/NAA at a time point of 1.7 months following a traumatic brain injury of moderate severity (Glasgow Coma Score 13). The color overlays represent the significance of the difference for the single subject values relative to mean values from an age-matched control group of 25 subjects, with decreased value for NAA and increased values for Cho and Cho/NAA. Adapted from Maudsley et al

FIGURE 6

Illustration of the recommended classes of acquisition methods for increasing levels of complexity of the spectral information. The color indicates the level of expertise required, ranging from sequences that are fully integrated into clinical protocols to specialized sequences that require specific research experience. Observations indicated by the numbers are as follows: (1) whole-brain acquisitions are susceptible to increased contamination from extracranial lipids; therefore, results from spectral fitting of lactate are labeled LL (Lipid+Lactate). (2) Whole-slice or whole-brain acquisitions benefit from using higher spatial resolutions^,, and are therefore not optimum for detection of low SNR signal components for ≤3T measurements. (3) Whole-brain acquisitions have large global B₀ inhomogeneities and quantitative analysis of resonances close to water and lipid may be impacted in some brain regions. (4) Measurements of compounds that have significant spectral overlap are widely implemented using frequency-spectral editing methods, which are most reliably implemented using volume-selective measurements^,,

FIGURE 7

Single slice (top) and whole brain multi-slice (bottom) ¹H FID MRSI acquired at 9.4 T. Scan time was 11 minutes for one slice (TR = 220 ms, 3.1 x 3.1 x 10 mm) and 25 minutes for the whole-brain scan (TR = 300 ms, 10 slices, 3.2 x 3.2 x 8mm, 7-fold acceleration). Modified from Nassirpour et al