Purpose

³¹P MRS can spectrally resolve phospholipid metabolites involved in phospholipid metabolism which are altered in many cancers¹. 7 Tesla facilitates the detection of PMEs (phosphocholine (PC), phosphoethanolamine (PE)) and PDEs (glycerophosphocholine (GPC), glycerophosphoethanolamine (GPE)) with increased SNR and spectral resolution. Multi-echo MRSI allows T₂‐weighted SNR enhancement, for an increased metabolite sensitivity, or T₂ information per metabolite^2,3. In MR imaging, a widely incorporated method to increase SNR per unit of time is the use of three-dimensional FSE with modulated refocusing flip angles (FA)^4,5. This employs a long echo train length (ETL; more than 100 in the brain) using variable, low FAs . We aimed to transfer this method to ³¹P MRS imaging. We simulated the use of variable FAs FSE in the ³¹P regime as a new approach for gaining SNR per unit of time within specific absorption rate (SAR) limits. The results were compared with a multi-echo technique with full refocusing pulses. However, the block pulses for this purpose require a high B₁ (i.e. much higher than 15μT which is maximal available with our ³¹P body-coil) to excite the frequency range of interest. As a solution, we have designed a dual-band refocusing pulse to be used at a maximum available B₁ of ~15μT⁶. This pulse selectively hits the two frequency ranges of interest (PDEs and PMEs) with the potential utilization for multi-echo sequences with more than 100 echo pulses. Finally, we implemented the dual-band pulse in a multi-echo MRSI sequence and validated the approach in a phantom and in vivo.

Methods

Simulations
We simulated the effect of extending the ETL on the PDE and PME signals by using refocusing FA modulated FSE technique⁵. T₁ and T₂ of the metabolites used in simulation are listed in Figure 1A. TR=5.5s, nominal FA=15, maximum FA=130, and echo spacing=15ms. Weighted average SNR_wa was calculated as in Eq. 1.

$$$S_{wa}=S_{0}\frac{1+2\sum_1^nS_{i}w_{i}}{1+2\sum w_{i}}, SNR_{wa}=\frac{S_{wa}}{\sigma_{wa}}=SNR_{0}\frac{1+2\sum_1^nS_{i}w_{i}}{\sqrt{1+2\sum w_i^2}} $$$ Eq.1

Where S₀, SNR₀, w_i, S_wa and σ_wa represent free induction decay (FID) signal, FID SNR, signal weight, weighted average signal and noise, respectively. The signal itself was used as signal weight. To compare, SNR_wa and SAR of a multi echo sequence with full refocusing 180s with the same echo spacing of 15ms and the same number of echo pulses (100) was calculated as previously described². FA squared was calculated as an alternative to SAR.

Data Acquisition
MRS measurements were carried out on a 7Tesla MR system (Philips, Best, NL) equipped with a double tuned ¹H/³¹P head coil. We acquired spectra from a phantom containing PME, PDE and inorganic phosphate (Pi) and from a healthy volunteer (female, 28yrs). The AMESING² sequence (multi-echo) was modified such that the excitation was performed by a block 90° pulse, followed by 180° Shinnar-LeRoux dual-band refocusing pulses (7ms) at 14.8 μT, with two refocusing bandwidths of 166Hz to hit the PDE and PME at 3.0-3.5 ppm and 6.2-6.8 ppm, respectively. A single FID by means of a pulse-acquire and 5 and 9 full echoes in one k‐space step where acquired while k‐space data were sampled spherically. As a reference, we performed the same experiments on a phantom using the sequence with adiabatic pulses requiring high B₁ of ~100μT, which could only be obtained on the phantom surface with a homebuilt dual-tuned surface coil set-up^2,7. (B_1,rms⁺)² of the dual-band and adiabatic pulses were calculated as:

$$$SAR\propto(B_1,rms^+)^{2}=\frac{1}{T_{dur}}\int_{0}^{T_{dur}}B_1^+(t)^{2}dt$$$ Eq.2

All multi-echo MRSI experiments were performed with a matrix size of 8×8×8, isotropic resolution 20 mm, 1 average, 256 data points, spectral width 8200 Hz, ΔTE 45 ms (3-fold larger than simulated to minimize truncation artifacts in spectral domain), TR 5s (in vivo) and 4s (phantom).

Results

Figure 1 shows the simulated signals of PDE and PME by using modulated FA FSE approach. It produces a refocusing flip angle train that rapidly brings the signals into pseudo-steady-state (PSS) conditions. Figure 2 shows SNR comparison for GPC signal. Multi-echo method with a number of 20 full refocusing 180 pulses resulted in SNR weighted average of 2.13 (considering T₁ relaxation effect). FSE with modulated FAs gave an SNR weighted average of 1.76 at the same FA squared of 673. The spectra acquired from the adiabatic and dual-band ME implementations are shown in Figure 3A and B. Using just 9 pulses increased SNR by more than a factor of 2 (Figure 3D). Figure 4 shows ³¹P spectra from the brain with PDE and PME signals being refocused by each echo pulse.

Discussion and Conclusion

Simulations showed that we can achieve higher SNR with a full 180 multi-echo sequence than a modulated FAs FSE sequence. Compared to the imaging FSE implementation, the echo spacing in our ³¹P FSE is much longer (15ms vs 3ms) resulting in signal loss due to T₂ relaxation. Using the very low power dual-band pulses, within a TR of 5s the number of echo pulses could be increased when compared to adiabatic pulses ((B_1,rms⁺)² of 0.246µT² compared to 62.277µT²), which would result in higher SNR. In addition to increased SNR, T₂ information of the metabolites can be estimated with this implementation. The in vivo results indicated the feasibility of using low power ³¹P dual-band pulses in multi-echo approaches instead of high power demanding adiabatic pulses.

References

[1] Glunde K, Bhujwalla ZM, et al., Choline metabolism in malignant transformation. Nat. Rev. Cancer, 2011; 11: 835–848.

[2] van der Kemp WJM, Boer VO, et al., Adiabatic Multi‐Echo P Spectroscopic ImagiNG (AMESING) at 7 tesla for measuring transverse relaxation times and regaining sensitivity in tissues with short T2* values. NMR Biomed. 2013; 26:1299–1307.

[3] van Houtum Q, Welting D, et al., Low SAR 31P (multi-echo) spectroscopic imaging using an integrated whole-body transmit coil at 7T. NMR in Biomed. 2019;

[4] Busse R, Hariharan H, et al., Fast Spin Echo Sequences With Very Long Echo Trains: Design of Variable Refocusing Flip Angle Schedules and Generation of Clinical T2 Contrast. MRM, 2006; 55:1030–1037.

[5] Busse R, Brau A, et al., Effects of Refocusing Flip Angle Modulation and View Ordering in 3D Fast Spin Echo. MRM, 2008; 60:640–649.

[6] Klomp DWJ, van de Bank BL, et al., 31P MRSI and 1H MRS at 7 T: initial results in human breast cancer. NMR Biomed, 2011; 24: 1337–1342.

[7] Shams Z, Wiegers EC, van der Kemp WJM, Klomp DWJ, Wijnen JP, ³¹P dual-band pulse for selective refocusing of phosphomonoesters and phosphodiesters at low B1 and low SAR. In Proceedings of the 31^st Annual Meeting ISMRM, 2021; 4569.