Synopsis

In this work, a 2D radial ²³Na sequence with half-VERSE pulses was developed to reduce SAR and achieve ultra-short TEs. Simulations and measurements of the slice profiles showed good agreement between full- and half-VERSE pulses with a maximal deviation in the FWHM of 7%.

In measurements, the TE_min was reduced from 1.41ms to 0.1ms by the use of half- instead of full-VERSE pulses. This resulted in an SNR gain of up to 18% in phantom, 7% in brain and 26% in calf measurements.

Last, rapid single-slice ²³Na images (2.9x2.9x12mm³) were obtained in a clinically feasible acquisition time of 2:56min.

Introduction

²³Na imaging was found to yield information additional to that provided by ¹H MRI¹. Even though 3D measurements are commonly conducted in ²³Na MRI, certain applications benefit from 2D imaging, such as cardiac MRI with higher in-plane resolutions².
The slice-selective pulses used in 2D imaging are limited by SAR and maximum coil voltage constraints, which both can be addressed by the use of Variable-Rate Selective Excitation (VERSE) pulses^2,3.
Another issue of conventional selective excitations are the long TEs, which are (for sinc/VERSE pulses) determined by half the pulse duration, the gradient ramp-down and the time required to rewind half the slice-selection gradient: $$$TE_{min}=\frac{\tau}{2}+T_{rampdown}+T_{rewind}$$$. Due to the short ²³Na relaxation times, reduction of TE yields an increased SNR, which can be realized by the use of half pulses⁴.
In this work, a 2D ²³Na sequence is presented, combining a VERSE routine for reduction of the SAR and peak voltage with half pulses to shorten the minimum TE.

Methods

A spoiled 2D radial sequence with VERSE excitation pulses was implemented. Here, the VERSE (also denoted as full-VERSE) pulses are constructed by processing a sinc pulse with a VERSE routine, as implemented in previous works⁵. After excitation, a slice-selection rewinder is applied, followed by a density-adapted readout gradient⁴, a rewinder in read direction and a simultaneous spoiling gradient, dephasing the magnetization over 4π in slice direction. A schematic sequence diagram is shown in Figure 1.
To reduce the minimum TE, half-VERSE pulses were implemented, constructed by truncating the full-VERSE RF pulse and the corresponding slice-selection gradient after their first half. The duration of the half-pulse is doubled to keep the pulse duration and FA constant. Since the RF half-pulse ends at its maximum (i.e. at isodelay), only the ramp-down momentum must be rewound, which allows TEs below 0.1ms.
Repeating the sequence N times with varying readout directions enables homogenous sampling of the k-space. After a 1s pause, allowing recovery of the magnetization towards thermal equilibrium, the same N spokes are acquired a second time with an inverted orientation of the slice-selection gradient. Combination of the two sets results in N measured spokes in k-space, allowing use of the same reconstruction for both the VERSE and the half-VERSE sequence, which was performed using a NUFFT operator⁵ after application of a Hamming filter.
The theoretical slice profiles of the full-VERSE and half-VERSE pulses were compared using Bloch simulations (τ=1024µs, bandwidth-time-product (BWTP)={2.7,5}, T₁=64ms, T₂^*=55ms, FA=35°, κ=0.5).
All measurements were performed on a 7T research system (Magnetom 7T, Siemens Healthcare) and a cylindrical phantom containing 0.9% NaCl was used. Six vials with Agar concentrations between 2% and 7% were inserted into the phantom, covering a range of relaxation times T₁=35-50ms, T_2s^*=4-11ms, T_2l^*=25-40ms⁷. The slice profiles were measured (FA=35°, TE=5ms, TR={25,150}ms, slice thickness=56mm), which was enabled by inserting phase-encoding gradients⁸ in the sequence. The final profile was determined by averaging the signal in the center of each slice, subtracting the mean noise floor and then normalizing the profile to its maximum.
A 2D measurement (parameters in Figure 3) was conducted in the phantom to investigate the SNR gain due to the half-VERSE pulses using⁹:
$$$SNR=\sqrt{2-\frac{\pi}{2}}*\frac{mean(signal)}{std(noise)}$$$.
Further, the SNR increase was investigated in the brain and calf of two healthy volunteers (parameters in Figure 4). In the head measurements, the SNR was determined in the brain tissue under exclusion of CSF. For calf measurements, muscle tissue was selected, omitting bone and vessels.
Finally, another in vivo measurement was performed in the human calf with a shorter scan duration of TA=2:56min.

Results

Good agreement between the simulated slice profiles of half- and full-VERSE pulses was found as well as to the measured profiles (Figure 2). Slice profile FWHMs of simulations and measurements agree within 7%.
In the 2D measurements (Figure 3), the use of half- instead of full-VERSE pulses allowed a reduction of the minimum TE from 1.41ms to 0.1ms.
A similar SNR for full- and half-pulses was found in compartments with long relaxation times (e.g. Compartment 2), while the SNR improvement by the half-VERSE pulse increased with shorter relaxation times, peaking at an SNR gain of 18% for an Agar concentration of 7%.
Figure 4 shows a representative head and calf measurement conducted with the half-VERSE sequence. SNR gains of 3% and 6% compared to the full-VERSE sequence were found in brain tissue for the two volunteers, while the SNR increases in calf tissue were 17% and 26%.
In Figure 5, the general applicability of the presented 2D ²³Na half-VERSE sequence is shown for short TAs.

Discussion & Conclusion

A novel 2D sodium sequence using half-VERSE pulses was developed, which allowed reduction of TE from 1.41ms to 0.1ms compared to a full-VERSE sequence and therefore resulted in SNR improvements up to 26%. Lower SNR gains in brain tissue compared to calf may be linked to potentially longer relaxation times¹⁰, CSF contributions in the brain or minor slice profile alterations. Nevertheless, a gain in the SNR by use of half-VERSE pulses was found in all measurements. Further, 2D measurement of the calf muscle was possible in 2:56min (2.9x2.9x12mm³), allowing rapid single-slice mapping of ²³Na in clinically feasible scan time.

Acknowledgements

No acknowledgement found.

References

¹ Mark E Ladd, Peter Bachert, Martin Meyerspeer, Ewald Moser, Armin M Nagel, David G Norris, Sebastian Schmitter, Oliver Speck, Sina Straub, Moritz Zaiss; Pros and cons of ultra-high-field MRI/MRS for human application; Prog Nucl Magn Reson Spectrosc; 2018 Dec;109:1-50

² Konstandin S, Schad LR. Two-Dimensional Radial Sodium Heart MRI Using Variable-Rate Selective Excitation and Retrospective Electrocardiogram Gating with Golden Angle Increments. Magn Reson Med. 2013; 70:791–799

³ Conolly S, Nishimura D, Macovski A. Variable-Rate Selective Excitation. J Magn Reson. 1988; 78:440-458

⁴ Konstandin S, Nagel AM, Heiler PM, Schad LR. Two-Dimensional Radial Acquisition Technique With Density Adaption in Sodium MRI. Magn Reson Med. 2011; 65:1091–1097

⁵ Schmitter S, Bock M, Johst S, Auerbach EJ, Ugurbil K, Van de Moortele PF. Contrast Enhancement in TOF Cerebral Angiography at 7 T Using Saturation and MT Pulses Under SAR Constraints: Impact of VERSE and Sparse Pulses Magn Reson Med. 2012; 68:188–197

⁶ Fessler JA, Sutton BP. Nonuniform Fast Fourier Transforms Using Min-Max Interpolation. IEEE Transactions on Signal Processing 2003;51:560-574

⁷ Fabian J. Kratzer, Sebastian Flassbeck, Sebastian Schmitter, Tobias Wilferth, Arthur W. Magill, Benjamin R. Knowles, Tanja Platt, Peter Bachert, Mark E. Ladd, Armin M. Nagel; 3D sodium (²³Na) magnetic resonance fingerprinting for time-efficient relaxometric mapping. Magn Reson Med. 2021; 86: 2412– 2425

⁸ Kratzer FJ, Flassbeck S, Nagel AM, Schmitter S, Bachert P, Ladd ME, Behl NG. 23Na Imaging using VERSE Pulses in a Density Adapted Stack-of-Stars Sequence at 7T. In proc. UHF Workshop of ISMRM 2019

⁹ Hákon Gudbjartsson and Samuel Patz, The Rician Distribution of Noisy MRI Data, Magn Reson Med. 1995 Dec; 34(6): 910–914

¹⁰ Guillaume Madelin, and Ravinder R. Regatte: Biomedical Applications of Sodium MRI In Vivo, Journal Of Magnetic Resonance Imaging, 38:511–529 (2013)