2490

Validation of B0-adjusted and sensitivity-encoded spectral localization by imaging (BASE-SLIM) using High Resolution 3D EPSI
Peter Adany1, In-Young Choi1,2,3, Sean Ellis1, and Phil Lee1,3

1Hoglund Brain Imaging Center, University of Kansas Medical Center, Kansas City, KS, United States, 2Department of Neurology, University of Kansas Medical Center, Kansas City, KS, United States, 3Department of Molecular & Integrative Physiology, University of Kansas Medical Center, Kansas City, KS, United States

Synopsis

B0-adjusted and sensitivity-encoded spectral localization by imaging (BASE-SLIM) provides non-Fourier based localization for arbitrarily shaped compartments. We have extended BASE-SLIM to 3D and compared the outcome of BASE-SLIM reconstruction with that of voxel averaged high resolution 1H MRSI.

Target Audience

Scientists, MR physicists, clinicians and students who are interested in advanced in vivo 1H MRS methods to quantify neurochemicals in the human brain in a region-specific manner.

Introduction

The Spectral Localization by Imaging (SLIM)1 framework provides reconstructions of MR spectra from distinct anatomical compartments without being limited to rectangular voxels as with all Fourier transform based MRSI. We have previously demonstrated that the B0-adjusted and sensitivity-encoded spectral localization by imaging (BASE-SLIM)2, which incorporates B0 and B1 information in spectral reconstruction, could reconstruct MR spectra of gray matter and white matter using 2D MRSI data. Although inherently BASE-SLIM can be applied to 3D MRSI data as 3D data offer better localization performance, it has not been achieved to date. In this study, we extended 2D BASE-SLIM to 3D BASE-SLIM using high resolution 3D MRSI data and compared its performance with ROI averaged (i.e., voxel averaging within ROI) MR spectra from high resolution 3D MRSI.

Methods

Fourteen healthy subjects (36±12 years of age, 6/8 F/M) were studied according to institutional review board approved protocols. MR measurements were performed on a 3 T scanner (Skyra, Siemens) using a 20-channel head/neck array coil. 3D T1-weighted MRI was acquired using a magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence (matrix = 176×256×256, resolution 1×1×1 mm3). 1H 3D MRSI data were acquired using the volumetric echo planar MRSI sequence (TE/TR1/TR2=17/1551/511, matrix = 50×50×18, FOV = 280×280×180 mm3) with GRAPPA (38/50 fill factor) and inversion recovery (IR) pulse for lipid suppression3. BASE-SLIM reconstruction was performed with 50×38×4×16 phase and coil encoding points, comprising the x,y,z and coil dimensions. We have used only 4 out of 14 k-space encodings in z-direction in order to overcome the memroy requirements for BASE-SLIM reconsruction. Anatomical parcellations were obtained using SPM12 and FreeSurfer. Data processing was performed using Matlab and MIDAS software4. MR spectra were quantified using LCModel.

Results and Discussion

Quantification of the NAA and choline (Cho) ratios to creatine (Cr) in the thalamus, caudal cingulate, insula and lingual regions in the left and right hemispheres, including the cortex and white matter structures, using BASE-SLIM were in agreement with those of ROI averaging using 3D MRSI (Table 1). The spectral quality of BASE-SLIM was equivalent to that of ROI averaging (Fig. 3). The BASE-SLIM non-Fourier reconstruction takes advantage of co-aligned high resolution MRI to reconstruct the MR spectra of distinct anatomical compartments rather than rectangular voxels. These results show that BASE-SLIM is also compatible with high resolution 3D MRSI and can provide accurate MR spectra from various anatomically shaped compartments. Althought we used high resolution 3D MRSI data to demonstrate the performance of BASE-SLIM in reconstructing compartmental spectra, BASE-SLIM does not require such high resolution MRSI for accurate compartmental spectral localization and provides greater flexibility to undersample phase encoding dimensions for acceleration. Thus, 3D BASE-SLIM is a promising tool that provides accurate compartment-based spectral localization without prolonged scan time, which is required for high resolution 3D MRSI.

Acknowledgements

The Hoglund Brain Imaging Center is supported by the NIH (S10RR029577) and the Hoglund Family Foundation.

References

1. Hu, X., et al. Magn Reson Med, 1988. 8(3): p. 314-22.

2. Adany, P., et al. NeuroImage, 2016. 134: p. 355-64.

3. Ebel, A., et al. Magn Reson Med, 2005. 53(2): p. 465-9.

4. Maudsley, A.A., et al. NMR Biomed, 2006. 19(4): p. 492-503.

Figures

Brain image segmentation (FreeSurfer) and field of view coverage of MRSI (yellow) using a 3D EPSI sequence. Non-Fourier reconstruction using a modified BASE-SLIM approach was performed, using a 50x50x4 spatial region (shown). Reconstruction used 50x38x4(x16) k-space based on the GRAPPA acquisition and 16 individual receiver coils. The VOI included the thalamus, caudal cingulate, insula and lingual regions among others.

Multi-coil acquisition using GRAPPA: at top, example of 2D GRAPPA EPSI data (first timepoint). Bottom: example MRSI signals from individual receiver coils (top) compared with measured B1 coil sensitivity profiles (bottom). Inclusion of the B1 map in MRSI localization reconstruction can improve the model accuracy significantly. Selected four channels for display from a 16 channel head coil.

Example spectra from EPSI processed using BASE-SLIM, SLIM and MIDAS. The matrix size contained the phase encoding (50×38×4) for SLIM and expanded by the number of coils (50×38×4×16) for BASE-SLIM. The transverse dimension of 4 was obtained by spatial truncation.

Table 1. Ratios of BASE-SLIM and SLIM-processed anatomical compartment reconstructions and traditional ROI voxel averaging (MIDAS software). Quantification was performed on all compartments from all methods using LCModel. The total choline and total NAA ratios to creatine were compared. Identical performance of reconstructions corresponds to a 1.0 ratio.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)
2490