Maximizing the signal-to-noise ratio (SNR) in in vivo cardiac phosphorus-31 magnetic resonance spectroscopy (³¹P-MRS) remains a major challenge in investigations of human myocardial energy metabolism. Typically, past studies used a ³¹P transmit/receive surface coil positioned over the chest close to the heart. Although a surface coil offers superior SNR compared to a volume coil, it suffers from flip angle variations due to B₁ inhomogeneity and limited penetration depth. This compromises the spatial localization of ³¹P-MRS signal acquisition from the heart, and hampers the quantification of metabolite concentrations. Here, we present a novel coil design consisting of dual-tuned ¹H/³¹P coil array with 4 elements in a geometric alignment tailored to achieve a larger and more homogeneous field of view for in vivo ³¹P-MRS compared to a standard ³¹P transmit/receive surface coil. Moreover, the elements are arranged in quadrature to maximize SNR for ³¹P-MRS, while detuning of the ¹H resonance was integrated to facilitate imaging and B₀ shimming.

Coil design: The proposed coil system consists of 4 dual-tuned (¹H @ 127.8 MHz; ³¹P @ 51.8 MHz) ∅ 15-cm elements in an anterior-posterior 2×2 combination (Figure 1) with active detuning at the ¹H frequency. Quadrature RF transmission and signal reception for ³¹P-MRS was routed via two X-nuclei channels. Four channels were used for ¹H signal reception. The setup was interfaced to a Philips Ingenia 3.0 Tesla MR system (R5.1.8; Philips Healthcare, Best, The Netherlands). The coil’s performance was compared with a standard linear single-turn ³¹P transmit/receive surface coil (∅ 14 cm, Philips).

Phantom studies: Two 9×9 gridded cryogenic storage boxes were used as a spacer between the anterior and posterior elements. To achieve a coil load similar to the human thorax, 500-mL 0.9% NaCl bags were placed on either side of the spacer, directly adjacent to the coil elements (Figure 1). A 3-mL vial filled with 15M phosphoric acid (H₃PO₄) was positioned in the center of the field of view and used for ³¹P B₁ calibrations. Coil reference B₁ and drive scale were adjusted such that a block pulse flip angle sweep yielded maximal signal intensity at the intended 90° excitation angle and zero signal at the intended 180° excitation angle. Next, the receive sensitivity distribution was measured by relocating the 3-mL 15M H₃PO₄ vial to different positions within the grids, and acquiring signal with an adiabatic pulse-acquire sequence.

In vivo ¹H-MRI and ³¹P-MRS studies: In vivo cardiac data were acquired in a healthy male volunteer (29 y/76 kg/1.76 m) positioned supine in the MR scanner. Acquisition of a left-ventricular short-axis cine ¹H-MRI series was performed with a balanced turbo field echo (BTFE) sequence (voxel size: 2.0×1.6 mm; matrix: 176×196; slice thickness: 8 mm; TR/TE: 2.8/1.41 ms; number of averages: 2; SENSE reduction factor: 1.5; 30 heart phases; retrospective ECG synchronization; single breath hold). Cardiac-triggered 3D-CSI was performed to acquire localized ³¹P-MR spectra of the heart (voxel size: 40×40×40 mm; matrix: 8×8×8; block pulse; flip angle: 45°; TR: 1 heart beat; number of averages: 1; 2048 data points; bandwidth 3000 Hz). Acquisition time was approximately 8 min 30 s. For comparison, the same 3D-CSI sequence and settings were used to obtain ³¹P-MR spectra from the same subject with the standard linear single-turn ³¹P surface coil.

Figure 2 shows ³¹P receive sensitivity maps comparing the spatial distribution of SNR between the 4-element ¹H/³¹P coil array (A) and the linear single-turn ³¹P surface coil (B). In the region relevant for cardiac ³¹P-MRS (boxed areas in Figures 1-2), SNR was 24% higher (p = 0.0002, paired t-test) and more homogeneous for the 4-element coil system (32.2 ± 4.8, mean ± SD) than for the linear single-turn surface coil (26.0 ± 9.5). This translated into a higher SNR for in vivo ³¹P-MRS of a 40×40×40 mm cardiac voxel with the 4-element ¹H/³¹P coil compared to the linear single-turn ³¹P surface coil (Figure 3). In addition, cardiac cine ¹H-MRI with the 4-element ¹H/³¹P coil yielded images at a spatial and temporal resolution that will be sufficient for the quantification of cardiac function (Figures 4-5). We demonstrated that with a dual-tuned ¹H/³¹P coil array with 4 elements, it is feasible to achieve a higher SNR and a more homogeneous receive sensitivity in the region of interest for cardiac ³¹P-MRS than with a standard linear single-turn surface coil. Furthermore, the 4 channels available for ¹H signal reception allow for cardiac cine ¹H-MRI with parallel imaging. Combined, the current setup allows for assessments of heart function with ¹H-MRI and myocardial energy metabolism with ³¹P-MRS during a single examination.