Over the last decades sodium (²³Na) MRI has evolved into a valuable biomedical application which can give additional information about the stage of a disease or about therapy response¹. Only a few ²³Na abdominal MRI studies have been performed at 7T, mostly using radiofrequency (RF) coils constructed for special applications with a medium field of view (FOV)^2,3. In this work, a versatile ²³Na RF transceive body coil with a large FOV and a high transmit/receive efficiency was designed, simulated, and implemented on a 7T scanner. B₁⁺-maps inside a phantom were obtained from RF simulations and measurements to evaluate coil efficiency. Subsequently, the setup was applied for in-vivo imaging of the human torso.

Hardware design:

A close-fitting low-pass elliptical birdcage coil was designed that provides a high filling factor and allows positioning in the isocenter of the MR scanner (Fig.1a). For comfortable subject positioning, the birdcage is separated into two halves (Fig.1b).

Cylindrical quadrature birdcage coils provide the advantage of a circularly polarized B₁⁺-field. In an elliptical birdcage rotational symmetry is violated, making the generation of a circularly polarized B₁⁺-field challenging. To improve quadrature performance of the elliptical birdcage, legs were spaced 360°/12=30° apart around the periphery of the ellipse⁴. Due to the split design, the left and right feeding legs of the birdcage were shifted to the upper half (Fig.1b), resulting in an additional violation of symmetry.

Four feeding ports (top, left, bottom, right) were implemented that enable manipulation of amplitude and phase at four legs. For preliminary measurements and simulations, equal amplitudes and phase differences of 0°, -90°, -180°, and -270° were applied. In further studies, B₁⁺-homogeneity can be improved by applying optimized phase and amplitude settings.

Coil simulations:

RF simulations were performed in CST Studio Suite 2015⁵. B₁⁺-maps were calculated for a phantom (ε_r’=54 , σ=0.43S/m at 78.6MHz, 23Na-content: w(NaCl)/w_ges=1.8%). A heterogeneous male body model (34y, 1.77m, 72.4kg)⁶ was used to obtain specific absorption rates (SAR) for in-vivo application.

Imaging:

Measurements were performed on a MAGNETOM 7T system⁷. Images were acquired using a density-adapted 3D radial sequence (DA-3DPR)⁸. Image reconstruction and post-processing were performed in MATLAB⁹.

B₁⁺-maps inside the phantom were obtained using the dual angle method¹⁰. Sequence parameters: Nominal isotropic resolution=10mm, reconstructed FOV=(400 mm)³, FA: α₁=55°, α₂=110°, t_pulse(55°)=1ms, t_pulse(110°)=2ms, T_E/T_R=(1.05ms/250ms), #projections=10000, T_AQ(per sequence)=41min40s.

High-resolution 3D in-vivo ²³Na image data of the human torso of a free-breathing healthy volunteer (26y, 1.80m, 75kg) were obtained. The maximum time-averaged input power of the RF coil was determined from the simulated SAR in the heterogeneous body model.

Sequence parameters: Nominal isotropic resolution=4mm, reconstructed FOV=(304mm)³, FA: α=44°, t_pulse=2ms, T_E/T_R=(1.05ms/20ms), #projections=18200, avg=5, T_AQ=30min20s.

Phantom study:

Simulated and measured B₁⁺-maps correspond very well (Fig.2). Mean B₁⁺ and standard deviation (SD) determined in the delineated ROI are 12.2µT (SD=1.4µT) in simulation and 14.1µT (SD=1.4µT) in measurement. B₁⁺-homogeneity can further be improved by applying optimized phases. First simulations (Fig.3) with optimized phases (0°, -73°, -180°, -287°) demonstrate a B₁⁺-distribution which is symmetric with respect to the coil center. Additionally, mean B₁⁺ increases (14.3µT) and SD decreases (1.1µT) compared to the standard phase settings.

In-vivo study:

Fig.4 shows exemplary slices of a high-resolution 3D ²³Na image data set of the human torso. The large FOV of (304mm)³ covers a region from heart to pelvis. High regional ²³Na signals are especially present in the areas of kidney, liver, cartilage, vessels, vertebral disks, spinal canal, and heart. ²³Na signal distribution of left kidney in Fig.4a-c indicates great potential of ²³Na MRI to characterize renal physiology. Although the heart is located at the outer part of the FOV, the left and right ventricles can be discriminated in time-averaged ²³Na signal distributions (Fig.4d-e).

Simulated and measured mean B₁⁺ correspond very well with the target value of 13.57µT (FA=55°). Values of B₁⁺ standard deviations are in an acceptable range. B₁⁺-maps show the expected non-uniformity due to the asymmetry of this elliptical birdcage with two vertically shifted feeding legs. For further studies, B₁⁺-uniformity can be improved by adjusting the phase differences between the feeding ports.

High-resolution ²³Na images show the great potential of ²³Na MRI at 7T. The developed body coil especially offers the possibility to examine ²³Na signal distribution at 7T in organs of the human torso like heart, breast, liver, kidney, and prostate. Additionally, retrospective gating can be used to separately reconstruct different phases of cardiac cycle or respiratory motion¹¹.