INTRODUCTION

Phased array (1) principle has been widely used as transmit and receive (TxRx) array designs at high and ultra-high field MRI systems (2,3). However, phased arrays have been rarely used as transmit coil design at low field due to the strong coupling among the coil elements, which exists not only between the adjacent coil elements, but non-adjacent coil elements. Such coupling can be mitigated through preamplifier decoupling mechanism in a receive array design. In a transmit array design, however, new decoupling mechanisms must be introduced to decouple the coil elements. In this work, we designed and built a 3T 8-channel ¹³C/¹H TxRx array with a novel cable trap decoupling mechanism.

METHOD

Coil Design: Figure 1a shows the schematic diagram of three ¹³C coil elements with cable trap decoupling circuits and Figure 1b shows the schematic diagram of the cable trap decoupling circuit only. Each cable trap decoupling circuit consists of an inductor (L) and capacitors Cp_x11, Cp_x12, Cp_x21, Cp_x22, Cp_x31 and Cp_x32 (x=1, 2, …, as their capacitance values may vary for different coil element decoupling). The cable trap is a parallel LC resonant circuit formed by an inductor, formed by 5 turns of 29-aug micro-coax wound on a 6 mm-diameter hollowed rod, and two capacitors (Cp_x11, Cp_x12, and Cp_x11=Cp_x12) placed at the two ends of a 5mm-wide PCB across the hollowed tube of the rod. The cable trap was placed at the middle point so that the length of the coax cable to the two coil elements to be decoupled is the same, as shown in Figure 1c. The eight 165mm × 113mm ¹³C coil elements were sitting on a 230mm-diameter cylinder, with the nose part raised up by 15mm. Adjacent coil elements were decoupled with each other by optimizing their geometric overlap. Cable trap decoupling circuit was used to decouple non-adjacent coil elements (coil 1 vs. coil 3, and coil 1 vs. coil 4). Eight 145mm × 120mm ¹H coil elements (Figure 1d) were placed on a 260mm-diameter cylinder. The ¹³C and ¹H arrays were co-centered and had 45° offset with each other in XY plane. Adjacent coil elements were decoupled by optimized geometric overlap. Each coil element was connected to its own TR-switch and preamplifier. Cable length was adjusted for preamplifier decoupling. ¹³C and ¹H 1:8 power splitters were used to split the ¹³C and ¹H transmit powers into 8 with 45° offset respectively. Phantom Experiment: SNR was measured in a spherical ethylene phantom to characterize the performance of the ¹³C and ¹H arrays spatially. For ¹³C, a 2D CSI sequence with/without RF excitation was used to calculate SNR voxel-by-voxel by dividing the peak absolute signal by the standard deviation of noise; for ¹H, SNR maps were reconstructed from two GRE sequences with and without RF excitation (4). Proton decoupling performance was evaluated by using the same 2D CSI sequence with and without proton decoupling. In vivo Experiment: Human subject studies were performed following a protocol approved by our institutional review board, and informed consent was obtained from subjects. In vivo ¹H head imaging was acquired with 3D MPRAGE and FLAIR sequences. All scans were performed on a Siemens Prisma 3T MRI system (Siemens Healthcare, Erlangen, Germany).

RESULT AND DISCUSSION

Unloaded to loaded Q ratio is 3.2 for the ¹³C coil element and 5.8 for ¹H. Figure 2 shows the S-parameter matrices and the noise correlation matrix of the ¹H and ¹³C arrays. The coil elements furthest to each other in the ¹³C array, e.g. coil 1 vs. coil 5, have the strongest coupling (max S21 = -11.6dB), as no decoupling mechanism was implemented between them. For the ¹H array, all the S21 is lower than -13dB. Noise correlation matrices also showed that both arrays were well decoupled, with off-diagonal elements less than 0.35 in both. Figure 3 shows SNR maps of the ¹³C and ¹H arrays. Both have higher signals in the peripheral area of the phantom. ¹³C spectra in the same voxel with/without proton decoupling shown in Figure 4 has no noise increase with proton decoupling, demonstrating sufficient decoupling between ¹H and ¹³C. Figure 5 shows high-resolution head images acquired with the 3D FLAIR and MPRAGE sequences.

CONCLUSION

We designed and built a 3T 8-channel ¹³C/¹H array with cable trap decoupling mechanism. Workbench and experiment results show that the coil elements are decoupled very well with each other. We are testing ¹³C sequences to use this array for hyperpolarized ¹³C MRSI and steady-state ¹³C MRS after oral uptake of ¹³C labelled glucose.

Acknowledgements

This work was funded by Cancer Prevention and Research Institute of Texas (CPRIT) RR180056 and was performed under the rubric of the Advanced Imaging Research Center, UT Southwestern Medical Center.