Introduction

X-nuclear (e.g., ³¹P, ¹⁷O and ²H) MRS imaging (MRSI) could play critical roles for better understanding neuroenergetics under various pathophysiological conditions^1,2. Recently, we have developed a quantitative ²H MRSI technique to simultaneously assess cerebral glucose metabolism and TCA cycle activity at ultra-high-field (UHF)^1,3,4. However, such studies require both ¹H and ²H RF coils for structural and metabolic imaging. In traditional design, the electric decoupling between two coils often degrades the performance for the low-frequency X-nuclear imaging. Additionally, to construct a large size ¹H surface coil operating at UHF, multiple split capacitors are commonly required^1,2,5,6,7. In this abstract, we present a novel design of ¹H-²H dual-frequency coil with single copper wire and a large coil size that can be tuned/matched to either ¹H (~698 MHz) or ²H (~107MHz) Larmor frequency for imaging at 16.4T. Moreover, we demonstrate that the same principle can be extended to build multi-coil arrays with excellent performance.

Methods

Figure 1(a1) shows the prototype of the ¹H-²H dual-frequency coil which is based on the basic LC resonant circuit design and comprised of four capacitors and an open-loop (5.5 cm in diameter). Based on the circuit diagram shown in Fig. 1(b1), resonant frequencies (w₀) of the coil can be estimated according to Eq. (1). With capacitance and inductance appropriately selected by simulation results (example shown in Fig. 3), frequencies near 698 MHz and 107 MHz can be obtained.
$$ \frac{1}{\frac{1}{\frac{1}{\frac{1}{\omega_0L_1}-\omega_0C_4}+\omega_0L_2}-\omega_0C_L}-\frac{1}{\omega_0C_5}-\frac{1}{\omega_0C_S}=0. \qquad (1)$$
The coil was tested on a network analyzer to measure S-parameters at 698 MHz and 107 MHz, and afterward taken to phantom study at 16.4 T using a ~3.5 cm diameter water ball. Global proton FID was acquired (400μs hard pulse, NT=1, TR=2s) for power calibration. The ¹H transmission field (B₁) was mapped using the double-flip-angle method⁸ and 2D GE multiple-slice (GEMS) sequence (TR=6s, TE=2ms, matrix=64×64, FOV=48×48mm², flip angle (FA)=20°&40°). A same-sized traditional ¹H surface coil, split by four identical capacitors (1.3 pF each, see Figs. 1(a2) and (b2)), was used for comparison. Global deuterium FIDs were also collected (200μs hard pulse, NT=50).

To test array coil design, we built a quadrature coil (Fig. 2(a)), comprised of two identical ¹H-²H dual-frequency single-loop coils as shown in Fig. 1(a1). The S-parameters were measured on the bench with a loaded phantom. 2D GEMS (FA=20°, matrix=128×128, FOV=80×80mm²) and 3D chemical shift imaging (CSI) (TR=45ms, matrix=9×9×5, FOV=80×80×80mm³) sequences were employed to acquire proton and deuterium images at 16.4 T using a 4.5-cm water ball containing 0.4% D2O and 77mM NaCl. A 2D-GEMS (TR=70ms, FA=14°, NT=64, matrix=32×32, FOV= 80×80mm², one 80mm slice) deuterium image was also collected.

Results and Discussion

Figure 3 shows the simulation results of the unloaded ¹H-²H dual-frequency coil with the balance capacitor C₅ fixed at 4.7 pF, to illustrate three representative cases with varied C₄ values from 1 to 5 pF. Interestingly, the coil resonant frequency with 1 pF C₄ was too high and unable to reach 698 MHz for ¹H MRI, or too low with 5 pF C₄. In contrast, the C₄ value of 2.2 pF provided an optimal tuning range for both ¹H (698 MHz) and ²H (107 MHz) resonances.

By analyzing global ¹H FID results shown in Fig. 1(c1) and (c2) as a function of RF pulse power, we found that the dual-frequency coil efficiency was about 40~50% of that traditional coil. Figure 4 shows the coronal B₁ maps of the dual-frequency coil (top) and traditional proton coil (bottom), indicating a similar uniformity but different field intensities produced by the two coils. The lower B₁ field of the dual-frequency coil is due to the lower energy efficiency. Since the proton coil is mainly for structural imaging and shimming, the decreased B₁ efficiency won’t undermine its utility for ²H MRSI applications.

Figure 2(b) shows the S-parameters of both ¹H and ²H channels of the quadrature coil, suggesting that the two channels are well decoupled and can be tuned/matched at both frequencies. Representative deuterium images are presented in Fig. 2(c) (sagittal GEMS) and Fig. 2(d) (axial CSI), indicating excellent performance and large imaging coverage.

Overall results suggest that the new coil design provides optimal performance for low-frequency ²H resonator and sub-optimal performance for high-frequency ¹H resonator (~7 times frequency differences) in UHF imaging. This design, simple and robust, allows a large coil size (>5 cm) with an extremely high frequency of 1000 MHz or beyond (see Fig. 3). The single-coil design for two different frequencies significantly simplifies the coil fabrication, especially for dual-frequency coil array to be used in human whole-head X-nuclear MRSI. Compared to conventional volume transmission multi-receive head coil, our design could provide a similar or better performance even at sub-optimal ¹H B₁ efficiency, as a large volume coil could lose a few times of B1 strength in the cortical area while the ¹H surface coil array is placed close to the object.

Conclusion

In this work, we have developed a simple, low-cost RF coil technique to construct dual-frequency surface coil(s) operating at proton and X-nuclear Larmor frequencies. A high-channel dual-frequency transceiver coil array is advantageous to achieve large-volume imaging. The same technique can be applied to other X-nuclear MRI/MRSI, for instance, ¹⁷O, ²³Na and ³¹P, applications.

Acknowledgements

This work was supported in part by NIH grants of R01 MH111413, R01 CA240953, U01 EB026978, S10 RR025031, P41 EB027061 and P30 NS076408.