Introduction

²³Na imaging generally suffers from low SNR due to the lower gyromagnetic ratio compared to ¹H as well as the lower natural abundance in the human body. Therefore, previously a combined ²³Na-¹H transceiver array consisting of eight ²³Na-loops and eight ¹H-dipoles for imaging the body at 10.5T was proposed¹ to take advantage of the expected increase in SNR afforded by 10.5T. Here, we demonstrate the validation of the electromagnetic (EM) model of this array and present the first ²³Na images acquired at 10.5T in the human body.

Methods

MRI experiments were conducted on a whole-body 10.5T scanner (Siemens Healthcare, Erlangen, Germany). The coil array was connected to the scanner via two 8-way splitters with fixed relative phases between the outputs. Separate TR-switches for sodium and proton were used, each optimized for the respective frequencies of 118Mhz and 447MHz.

Coil validation
To validate the EM model of the coil, experimental data was acquired in a polyvinylpyrrolidone (PVP)-agar phantom model (c.f. Tab.1) with a cross section approximating a human torso that contained a small container filled with oil to obtain a phase reference for MR thermometry². All MR experiments were conducted on a whole-body 10.5T magnet.
Relative B₁⁺ maps were acquired for both nuclei on a channel-wise basis³. In addition, absolute B₁⁺ mapping was performed for ²³Na on a series of GRE images with varying flip angle, while the AFI⁴ method was used for ¹H. For both nuclei, the absolute B₁⁺ maps were acquired for two different B₁⁺ shims (CP⁺ and CP²⁺) by changing the order the elements were connected to the 8-way splitters. Relative and absolute B₁⁺ maps were combined to obtain channel-wise absolute B₁⁺ maps. Based on these, channel-wise real-valued scaling factors were determined to rescale the simulation data.
Using the same two B₁⁺ shims, experimental SAR distributions were acquired based on MRT data. To this end, a multi-echo GRE (TE=5ms/10ms/15ms) acquisition was performed (using the ¹H elements) before and after a ten-minute long heating sequence (P_avg,23Na=145W; P_avg,1H=119W). All temperature data was corrected for a DC offset determined in the small oil container. SAR was calculated based on the measured temperature difference as . For comparison, the simulated SAR was averaged over the same voxel volume (4x4x4mm³).
Regions of low B1+ and SAR were excluded when calculating the normalized root mean square error (NRMSE) between experimental and simulation data.
To determine safe operating power limits, EM simulations were performed in the Duke, Ella, and Fats29 model (Sim4Life, Zurich Medtech, Zürich, Switzerland) with the arrays positioned to target the kidneys. Using the fixed relative phases of the 8-way splitters, peak 10g SAR was evaluated for the CP⁺ shim for both nuclei across all three models. Based on the maximum peak 10g SAR values, total power limits of 71W for ²³Na and 30W for ¹H were determined.

In-vivo study
Imaging was performed on a healthy male subject targeting the kidneys and spine, following an IRB approved protocol with an FDA Investigational Device Exemption. T1 weighted gradient echo images with fat suppression were acquired using the ¹H elements during breath hold. In addition, a 3D radial density adapted sequence⁵ was used for ²³Na imaging. Imaging parameters are summarized in Table 2.

Results

A comparison of the simulated and experimental channel-wise B₁⁺ maps is shown in Figure 1 together with the channel-wise scaling factors: 0.60±0.03 for ²³Na and 0.35±0.07 for ¹H (Fig.1c). The B₁⁺ maps displayed in Figure 2 correspond to a NRMSE of 0.14/0.17 and 0.39/0.24 for the CP⁺/CP²⁺ mode for ²³Na and ¹H, respectively. A similar comparison of the SAR maps is shown in Figure 3, yielding NRMSE of 0.35/0.35 and 0.28/0.18.

In-vivo results of a coronal slice targeting the kidneys, as well as a sagittal slice through the spine are summarized in Figure 4.

Conclusion

In this work we successfully validated the EM model of a combined ²³Na-¹H transceiver array consisting of eight ²³Na-loops and eight ¹H-dipoles for imaging the body at 10.5T and subsequently acquired the first in-vivo human ²³Na images at 10.5T. While the initial in-vivo study shows promising results, an in-depth analysis is still required to quantify the expected SNR gains at 10.5T as well as potential limitations of ²³Na body imaging with regard to total power and SAR

Acknowledgements

Funding was provided by NIH P41 EB027061