Purpose

To develop a robust design of a human head double-tuned (DT) ³¹P/¹H array, which provides good performance at both ³¹P and ¹H frequencies for chemical shift spectroscopic imaging (CSI) at 9.4T.

Introduction

Imaging other than hydrogen nuclei, i.e. X-nuclei, is often difficult due to lower SNR. Therefore, an SNR enhancement at ultra-high field (UHF, >7T) is very critical for X-nuclei imaging. To provide both high-resolution anatomical human brain imaging and B₀ shimming, double-tuning of the coil is necessary. X-nuclei CSI of a human brain is potentially valuable for diagnostics of many diseases. CSI benefits from a whole-brain coverage and a high transmit (Tx) performance not only at X- but also at ¹H-frequency, which at UHF requires using a multi-element RF array (1-4). It is rather difficult to optimize the DT-array at both frequencies at the same time. Therefore, often the performance of the X-nuclei portion is optimized while the ¹H-array performance can be compromised (2). In this work, we developed a novel DT-array design, which provides good performance at both ³¹P and ¹H frequencies.

Methods

It is well known that increasing the number of surface loops in a human head array improves only the peripheral SNR, while the central SNR doesn’t substantially change (5-7). High peripheral SNR can be harmful for CSI due to contamination of near-cortical voxels by a strong fat (¹H) or muscle PCr (³¹P) signal. To minimize this effect, we limited the number of loops in the ³¹P-array to 10, i.e 8 17-cm long transceiver (TxRx) surface loops circumscribing the head and 2 receive (Rx) “vertical” loops placed at the superior location of a head (Fig.1A). The ¹H-portion of the array (Fig.1B) consists of 10 TxRx-elements, i.e. 8 11-cm long surface loops circumscribing the head and 2 “vertical” loops at the superior location. Both the ³¹P-array and ¹H-array are placed in a single layer on the surface of the holder at the same distance to the load (Fig.1C), which provides high loading and, thus, a high Tx-performance for both arrays. Such arrangement is feasible because of the small number of elements in the ³¹P-array. With a higher element count, the ¹H-array is commonly placed in the second layer at larger distance to the sample, which decreases the ¹H Tx-efficiency (2). To minimize interaction between adjacent ³¹P and ¹H surface loops, ¹H-loops are shifted by half-loop size (Figs.1C,2A). Placements of ³¹P-traps into ¹H-loops and ¹H-traps into ³¹P-loops minimize residual interaction. During transmission both arrays are driven in the quadrature mode using two home-built Wilkinson splitters both placed in a single box (Fig.2B). We compared the new array to a single-tuned (ST) ¹H-array (8) of similar size with 8 surface loops (10-cm long). Electromagnetic (EM) simulations of the transmit B₁⁺ and local specific absorption rate (SAR) were performed using CST Studio Suite 2015 (CST, Darmstadt, Germany) and the time-domain solver based on the finite-integration technique. Three voxel models were used, i.e. a head/shoulder (HS) phantom (e =58.6, s=0.64S/m), and two virtual family multi-tissue models, “Duke” and “Ella”. Experimental B₁⁺ maps at ¹H-frequency (9) and ³¹P-frequency (10) were obtained as previously described (9,10). All data were acquired on a Siemens Magnetom 9.4T human imaging system.

Results and Discussion

Figs.3A and 3B show simulated and experimentally measured B₁⁺ maps obtained using both the ¹H and ³¹P portions of the new DT-array and phantoms. Simulated and experimental data match each other well. Figs.3C and 3D show SNR maps obtained using the ¹H/³¹P-array at both frequencies. As seen from Fig.3 addition of “vertical” cross-loops (Fig.1) improves both B₁⁺ and SNR distributions at the superior location of the head. Fig.4 shows in-vivo results obtained using both the new DT-array and the ST-array at ¹H-frequency. Similar to the phantom data, addition of the cross-loops substantially improves the brain coverage. It is also of importance that the Tx-efficiency of the DT-array is very similar to that of the ST-array. Averaged over 120-mm transversal slab B₁⁺ measured 9.3μT/√kW and 9.4μT/√kW for the DT and ST arrays, respectively (Figs.4B,D). Averaging over 40-mm slab produced higher B₁⁺ for the ST-array, i.e. 11.8μT/√kW (ST array) versus 10.7μT/√kW (DT array). Finally, Fig.5 demonstrates CSI images obtained using the new DT-array.

Conclusion

We developed a novel ³¹P/¹H DT-array for CSI of a human brain at 9.4T. Placing both ³¹P and ¹H loops in a single layer at the same distance to the sample provides for high Tx-efficiency at both frequencies without compromising SNR near the brain center at the ³¹P-frequency. Addition of the cross-loops at the superior location of a head improves the brain coverage at both frequencies.

Acknowledgements

No acknowledgement found.

References

1) Adriany G, Van de Moortele P-F, Ritter J et al. Magn Reson Med 2008;59:590-597. 2) Avdievich NI. Appl Magn Reson 2011;41(2):483-506. 3) Gilbert KM, Belliveau JG, Curtis AT et al. Magn Reson Med 2012;67:1487–1496. 4) Shajan G, Kozlov M, Hoffmann J et al. Magn Reson Med 2014;71:870–879. 5) Wiggins GC et al. Magn Reson Med 2009;62:754–762. 6) Vaidya MV et al. Conc Magn Reson Part B 2014; 44B(3):53-65. 7) Lattanzi R, Wiggins GC, Zhang B, Duan Q et al. Magn Reson Med 2018;79(3):1789-1803. 8) Avdievich N.I., Giapitzakis I.A., and Henning A. NMR in BioMed 2018, 31(2):1-13. 9) Yarnykh VL. MRM 2007;57:192-200. 10) Allen SP, Morrell GR, Peterson B et al. Magn Reson Med 2011;65(4):1125–1130.