1850

Enhancing 2D MRSI: Implementation of CHEmical-shift Adiabatic Pulses (CHEAP) at a 7T Philips platform using Pulseq

Kyung Min Nam¹, Thomas Roos¹, Guodong Weng^2,3, Dennis Klomp¹, Johannes Slotboom^2,3, Jannie Wijnen¹, and Alex Bhogal¹
¹Department of High Field MR, Centre for Image Sciences, University of Medical Centre Utrecht, Utrecht, Netherlands, ²Institute for Diagnostic and Interventional Neuroradiology, Support Center for Advanced Neuroimaging (SCAN), University of Bern, Bern, Switzerland, ³Translational Imaging Center, Sitem-insel, Bern, Switzerland

Synopsis

Keywords: Spectroscopy, Spectroscopy

Motivation: Nuisance signal contamination and challenges associated with implementation involving advanced RF pulses and sequence hinder clinical adoption. Simplified sequence implementation and dissemination are crucial for community-driven advancement and vendor support.

Goal(s): Our goal is to integrate CHEmical-shift selective Adiabatic refocusing Pulses(CHEAP) on the Philips platform via Pulseq open-source platform, creating advanced MRSI sequences that refine metabolite analysis by minimizing unwanted signals.

Approach: Integrating chemically selective adiabatic 2𝜋-refocus pulses in Pulseq achieved optimal spectrum coverage, reducing interference from residual water and lipid signals.

Results: Implementing the CHEAP sequence significantly mitigated interference from residual water and lipid signals, demonstrating its potential for advancing MRSI.

Impact: The implementation of CHEAP sequence via Pulseq promises a standardized, shareable method, fostering collaboration and enabling precise metabolic studies. This advancement in MRS techniques may significantly improve reproducibility across sites and enhance capacity for metabolic profiling in health and disease.

Introduction

In magnetic resonance spectroscopy (MRS), CHEmical-shift selective Adiabatic refocusing RF Pulses (CHEAP^1,2) can be tailored to selectively refocus specific frequency ranges. These precise, narrow-bandwidth pulses are crucial for targeted excitation, which can be applied to effectively suppress contaminating signals from water or extra-cranial lipids. By nature adiabatic RF pulses are insensitive to B1 inhomogeneity, exciting and refocusing the region of interest. CHEAP pulses dephase unwanted signals beyond the intended range, notably reducing water and lipid signals without causing increased Specific Absorption Rate (SAR) or spectral aliasing. Our study replicated a 2D MRSI sequence using chemically selective adiabatic 2𝜋 pulses (2𝜋-CHEAP) at 7T on a Philips platform through the Pulseq open-source framework that was recently implemented at our site³.

Methods

A 7T MR scanner (Philips Medical Systems, Best, the Netherlands) with an 8Tx/32Rx head coil (Nova Medical, USA) using the built-in gradient system (maximum gradient amplitude of 40 mT/m and maximum slew rate of 200 mT/m/msec). Second-order B₀ shimming was performed. Employing the Pulseq framework³, a 2D MRSI sequence with 2𝜋-CHEAP¹ pulses was created. The RF, gradient, and ADC events were based on the complete SLOW-editing² scheme 2 parameters. Imaging parameters for both phantom experiments (sphere phantom A) and a volunteer included FA = 65°, bandwidth = 5.5kHz, and RF pulse duration = 6ms for the sinc-Gauss excitation pulse. Additionally, a 2𝜋-CHEAP pulse used nominal FA = 550°, bandwidth = 0.88kHz, duration = 24ms with and without carrier frequency shift (Δf = 0.00 and 2.90 ppm for phantom, 3.1 ppm for the volunteer). Acquisition parameters were: FOV = 240×240 mm², slice thickness = 10mm, voxel size = 8×8 mm², TR/TE = 1500/68ms, spectral bandwidth = 5kHz, 2048 sample-points, readout oversampling factor = 2, one average, with a 22.5-minute acquisition time. The Philips Pulseq specifics involved RF ringdown time (200 usec), RF deadtime (1100 usec), ADC deadtime (800 usec), block duration raster time (0.1 usec), and RF and gradient raster times (6.4 usec). Data reconstruction used an in-house MATLAB script and the ReconFrame package (Gyrotools, Zurich, Switzerland). Noise covariance matrix calculation⁴ aided in pre-whitening, and Roemer⁵ equal noise channel combination were executed, and metabolite maps were generated from the intensity of each metabolite's peak height.

Results

The 2𝜋-CHEAP sequence (Fig. 1) was applied in both phantom experiments (Fig. 2 and 3) and in a volunteer (Fig. 4 and 5). Shifting the carrier frequency (Δf = 2.9 and 3.1 ppm) successfully achieved selective excitation while avoiding pre-pulses for water or fat suppression. Employing just two low-bandwidth adiabatic RF pulses allowed for a short TR (1.5 sec) and minimized SAR. Figure 1 demonstrates the CHEAP sequence diagram introduced in our previous study⁶. Figure 2 shows the spectrum at Δf = 2.9, detecting the ethanol CH3 signal while omitting the water signal. A full bandwidth coverage was achieved at Δf = 2.9 ppm (1.6-4.2 ppm range, Fig 2D, E). In vivo, Δf = 3.1 ppm was used in the 1.8-4.4 ppm range (Fig 4) to avoid strong lipid signals. Residual water signals appeared in the left anterior and lipid signals in the right posterior regions, maintaining high SNR despite acquiring only 1 signal average (Fig. 4D). To handle water and lipid artifacts, a post-processing mask generated a water map without carrier frequency and metabolite maps with the carrier frequency (Fig. 5). Each map was generated from the highest metabolite peak within their range.

Discussion

Our 2𝜋-CHEAP MRSI sequence, created via the Pulseq framework, effectively suppressed unwanted signals (>4.2 ppm and <1.6 ppm), ensuring spectra with flat baselines. However, in vivo scans revealed residual water within the brain, primarily in the centre of the brain, suggesting the need for sequence timing optimization to match hardware specs. Persistent strong lipid signals in areas near the skull could potentially be addressed using additional techniques^7–11 or implementing high-resolution MRSI¹². Although Pulseq files could not be transferred directly from Siemens to the Philips 7T system due to timing differences⁶, sharing scripts of Pulseq files between users can improve pulse sequence design by simplifying adjustments to each system. This approach can improve high-resolution, rapid readout^2,11,13–16 in vivo scans within clinically feasible times.

Conclusion

We have successfully implemented a 2𝜋-CHEAP sequence in the Philips 7T scanner using Pulseq, an open-source framework. This sequence provides comprehensive metabolic information by adjusting the carrier frequency within its spectral bandwidth, effectively mitigating strong residual water and lipid signal contamination. This Pulseq-coded pulse sequence can be effortlessly shared among diverse institutions with different MR platforms and software versions, which is an important step forward in more reproducible MRS research.

Acknowledgements

This work was (partially) funded by Spectralligence (EUREKA IA Call, ITEA4 project 20209) and an NWO VIDI grant awarded to the senior author (VI.Vidi.223.085).

References

1. Conolly S, Nishimura D, Macovski A. A selective adiabatic spin-echo pulse. J Magn Reson. 1989;83(2):324-334. doi:10.1016/0022-2364(89)90194-7

2. Weng G, Radojewski P, Sheriff S, et al. SLOW: A novel spectral editing method for whole-brain MRSI at ultra high magnetic field. Magn Reson Med. 2022;88(1):53-70. doi:10.1002/mrm.29220

3. Roos THM, Versteeg E, Klomp DWJ, Jeroen CWS, Wijnen JP. Open-source Pulseq sequences on Philips MRI scanners. arXiv. 2023.

4. Pruessmann KP, Weiger M, Scheidegger MB, Boesiger P. SENSE: Sensitivity encoding for fast MRI. Magn Reson Med. 1999;42(5):952-962. doi:10.1002/(SICI)1522-2594(199911)42:5<952::AID-MRM16>3.0.CO;2-S

5. Roemer PB, Edelstein WA, Hayes CE, Souza SP, Mueller OM. The NMR phased array. Magn Reson Med. 1990;16(2):192-225. doi:10.1002/mrm.1910160203

6. Nam KM, Weng G, Lee NG, et al. Development of a 2D MRSI sequence with Chemical-Shift Selective Development of a 2D MRSI sequence with Chemical-Shift Selective Adiabatic Pulses (2pi-CSAP) using Pulseq at 7T. In: Proc 32nd Annu Meet ISMRM. 2023; # 3938. ; 2023.

7. Haupt CI, Schuff N, Weiner MW, Maudsley AA. Removal of lipid artifacts in 1H spectroscopic imaging by data extrapolation. Magn Reson Med. 1996;35(5):678-687. doi:10.1002/mrm.1910350509

8. Bilgic B, Chatnuntawech I, Fan AP, et al. Fast image reconstruction with L2-regularization. J Magn Reson Imaging. 2014;40(1):181-191. doi:10.1002/jmri.24365

9. de Graaf RA, Brown PB, De Feyter HM, McIntyre S, Nixon TW. Elliptical localization with pulsed second-order fields (ECLIPSE) for robust lipid suppression in proton MRSI. NMR Biomed. 2018;31(9):1-8. doi:10.1002/nbm.3949

10. Boer VO, Van De Lindt T, Luijten PR, Klomp DWJ. Lipid suppression for brain MRI and MRSI by means of a dedicated crusher coil. Magn Reson Med. 2015;73(6):2062-2068. doi:10.1002/mrm.25331

11. Nam KM, Hendriks AD, Boer VO, Klomp DWJ, Wijnen JP, Bhogal A. Proton metabolic mapping of the brain at 7 T using a two-dimensional free induction decay–echo-planar spectroscopic imaging readout with lipid suppression. NMR Biomed. 2022:1-13. doi:10.1002/nbm.4771

12. Ma C, Lam F, Johnson CL, Liang ZP. Removal of nuisance signals from limited and sparse 1H MRSI data using a union-of-subspaces model. Magn Reson Med. 2016;75(2):488-497. doi:10.1002/mrm.25635

13. Nam KM, Gursan A, Bhogal AA, et al. Deuterium echo-planar spectroscopic imaging (EPSI) in the human liver in vivo at 7 T. Magn Reson Med. 2023;1-12. doi:10.1002/mrm.29696

14. Posse S, Tedeschi G, Risinger R, Ogg R, Bihan D Le. High Speed 1H Spectroscopic Imaging in Human Brain. Magn Reson Med. 1995;33:34-40. http://meteoreservice.com/PDFs/Posse95.pdf.

15. Furuyama JK, Wilson NE, Thomas MA. Spectroscopic imaging using concentrically circular echo-planar trajectories in vivo. Magn Reson Med. 2012;67(6):1515-1522. doi:10.1002/mrm.23184

16. Adalsteinsson E, Irarrazabal P, Topp S, Meyer C, Macovski A, Spielman DM. Volumetric spectroscopic imaging with spiral-based k-space trajectories. Magn Reson Med. 1998;39(6):889-898. doi:10.1002/mrm.1910390606

Figures

Figure 1. A schematic illustration showing the 2D 2𝜋-CHEAP MRSI pulse sequence at 7T using Pulseq. RF events are depicted in blue, GR events in varied colors (Gz: green, Gx: blue, Gy: violet), and ADC events in orange, presented along a timeline. Crusher gradient amplitudes and durations were determined based on parameters from the SLOW-editing full scheme 2. Notably, RF and ADC events share the same axis, and the sequence does not include water suppression pulses (e.g., WET or VAPOR).

Figure 2. Spherical phantom A shows acetate and ethanol metabolites using the 2D 2𝜋-CHEAP MRSI sequence. (A) Displays a reference spectrum from a single voxel in the PRESS sequence. The red box in the phantom image marks the volume of interest in the PRESS sequence. (B) The reference axial image aligns with the 2D 2𝜋-CHEAP MRSI sequence. 2D spectra (C) were acquired using the 2𝜋-CHEAP with a Δf = 2.1 ppm. Red letters denote the Anterior-Posterior and Left-Right directions. (D, E) Show representative spectra from two locations (blue and orange squares).

Figure 3. The 2D spectra (A) were acquired from a phantom without a carrier frequency shift (Δf = 0 ppm), displaying the full spectrum (B) at the central location (green square) within the phantom. Metabolite maps for water (C), acetate (D), and methanol (E) were generated by computing peak intensities from the acetate (1.88 ppm) and methanol (3.16 ppm) signals in the spectra, whereas the water map reflects data obtained without a carrier frequency shift.

Figure 4. Acquired in vivo, the spectra are presented. Panel (A) shows the reference axial brain image (top left), while (B) illustrates the spectrum without a carrier frequency shift, capturing the full spectrum from the brain center. Spectra with a carrier frequency shift (C) reveal the full spectrum (D, top right) from a central location of the brain, zooming in from 0 ppm to 4.2 ppm (orange box) to display key metabolites. Even in these spectra (C), used to avoid residual water and lipid signals, relatively strong lipid (E) and water (F) signals persist in regions close to the skull.

Figure 5. Metabolite maps including in vivo water are displayed. Panel (A) shows the reference axial image, with the orange box representing the field of view (FOV). Panel (B) displays the water map obtained from the water signal without a carrier frequency shift. Panels (C) through (E) show the NAA, Cr+PCr, and Choline maps, respectively, generated from spectra acquired with a carrier frequency shift (Δf = 3.1 [ppm]) for each corresponding metabolite range. The range for each metabolite is indicated on each map (top, yellow), along with the carrier frequency shift value (bottom, grey).

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)

1850

DOI: https://doi.org/10.58530/2024/1850