Introduction

In vivo ³¹P MRS/MRSI detects various phosphorous metabolites involving ATP energy and phospholipid metabolisms and NAD redox state in human brain^1-3. The abnormality observed in the ³¹P studies reflects brain diseases^4-6. However, it is challenging to achieve high spatiotemporal resolution owing to a lower ³¹P γ ratio and very low phosphorus metabolites concentration. Previous studies^7-8 have shown that increasing magnetic field strength (B₀) can largely improve the ³¹P detection sensitivity and spectral resolution. However, the improvement depends on many factors including B₀, RF coil quality factor (Q), the longitudinal relaxation time (T₁) and the resonance linewidth (∆V_1/2). In this work, we analyzed and compared the ³¹P-MRSI SNR between 7T and the world’s first 10.5T whole-body human scanner.

Method

RF coil construction: We designed and constructed a loop-dipole ³¹P-¹H probe with passive decoupling, which could be operated at 7T or 10.5T frequencies. As shown in Fig. 1A-B, the ³¹P loop coil has a 7.5cm outer diameter. The ¹H dipole coil was located 2.5 cm away from the ³¹P loop coil plane. One matching circuit was used for the ³¹P loop coil, and two matching circuits for the ¹H dipole coil. The coil Q factor was measured using -7dB from the 0dB baseline of the S₁₁ plot as demonstrated in Fig. 1C-D.
³¹P MRSI Experiment: The ³¹P MRSI experiments were performed on an inorganic phosphate (Pi) phantom (100mM, similar loading effect as human head and T_1,Pi was approximately 300ms at 7T) on 7T and 10.5T human scanners (SIEMENS, Germany) using the dual-channel ³¹P-¹H loop-dipole coil. Figure 2C shows the set-up of the loop-dipole coil loaded with the Pi phantom. B₀ shimming was performed over an 8cm cube at the bottom of the phantom near the coil. Then, the 3D ³¹P chemical shift images (CSIs) with an array of voltages were acquired with an FOV of 15cm covering the entire phantom. At both field strengths, we used the same CSI parameters: TR=1500ms (fully relaxed condition), TE=0.5ms, excitation hard pulse length=500us, and spectral width=5000Hz.
SNR quantification: The ³¹P FID data from each CSI voxel was transformed to a spectrum with a 10Hz line broadening. As show on Fig. 3, for each CSI voxel, the Pi resonance intensities measured with different excitation voltages were fitted to a sine function. The peak intensity determined by the regression of the sine wave function at the flip angle of 90^o presented the ultimate (or largest) Pi signal (S_L) that could be achieved in a particular voxel. We selected 10 voxels from the similar location near the center of the coil sensitive region for both field strengths to calculate the mean S_L values at 7T and 10.5T, respectively; and 8 voxels outside of the phantom acquired using zero excitation voltage (thus no Pi signal) to calculate the standard deviation of the noise. To obtain the ultimate SNR for a selected voxel, the S_L value of each voxel was divided by the mean noise standard deviation.

Result

RF coil performance: As shown in Fig. 1 C-D, the ³¹P loop coil and ¹H dipole coil were able to be tuned and matched to a S₁₁ value of -15dB and -20dB, respectively, at both 7T and 10.5T. For both field strengths, the S₁₂ between the loop and dipole coil was less than -20dB at the ³¹P resonant frequencies, showing an excellent decoupling between the ³¹P and ¹H coils.
SNR comparison between 7T and 10.5T: Figure 3 illustrates one example of sine function fitting to determine the S_L value for a representative CSI voxel at 7T. Figure 4 shows the ³¹P SNR maps of three transversal slices at both field strengths. These SNR maps are overlaid with the 3D-CSI ³¹P spectra acquired using the excitation voltage approximately corresponding to the nominal 90° flip angle for the global FID signal. For the selected 10 voxels, the ratio of the SNRs measured at 10.5 T to 7 T was 1.5 ± 0.2 (see Table 1).

Discussion

This study reports quantitative sensitivity comparison result showing the SNR ratio (=1.5) of ³¹P MRSI imaged at 10.5T versus 7T (with a B₀ ratio of 1.5), thus, an approximate linear relation to B₀ consistent with previous reports^7,8. The linewidth effect at both 7T and 10.5T were similar in this study, and the 1.5s TR allowed fully relaxed condition at least for 7T measurement. Therefore, we can have a ³¹P SNR ratio equation: SNR_10.5T / SNR_7T = (10.5/7)^β Q^0.5, resulting β =1.7 using the parameters listed in Table 1. This power dependence is slightly higher than previously reported work4 where β =1.4-1.5 were found in human brain at 4T and 7T.

Conclusion

We report the first 10.5T study that investigated the B₀ dependence of ³¹P MRSI sensitivity by quantitative comparison between 7T and 10.5T human scanners. The results suggest an approximately 50% SNR gain at 10.5T as compared to 7T. In considering the SNR gain and anticipated improvement of spectral resolution at 10.5T⁸, we expect significant improvements for in vivo ³¹P MRSI human brain application at 10.5T.

Acknowledgements

This work was supported in part by the National Institutes of Health grant: U01 EB026978, R01CA24095 and P41 EB027061.