Introduction:

Deuterium MRS imaging (DMRSI) is a promising tool with strongly increasing popularity for imaging cerebral glucose metabolism including the tricarboxylic acid cycle (TCA cycle)[1]. DMRSI offers higher sensitivity and less technique complexity than ¹³C MRS; it is also less invasive and more cost-effective than the clinically available, but ionizing-radiating imaging modality - fluorodeoxyglucose (FDG) positron emission tomography (PET). Through administration of deuterated glucose, the Warburg effect map (Lac/Glx) generated by DMRSI revealed high contrast between tumor and normal tissue[2,3]. An 8TxRx/2Rx 2H phased array at 9.4T[4] showed higher DMRS signal in the peripheral area and relative weak signal in the center of a head-size deuterated water phantom. As ²H coil B₁ field wavelength is seven times bigger than that of proton in the same field strength, the majority of ²H B₁ field is spatially distributed close to the conductor of the coil. This deteriorates the ²H B₁ field penetration in the imaging phantom as compared to that of a 1H coil, and dominates the coupling between 2H array coils.

In this study, we report a novel combination of 4 channel ²H loop coils and 4 channel ¹H microstrip elements (i.e., 4-ch ²H TxRx/4-ch ¹H TxRx head array coil). Four large ²H loops were overlapped with each other, providing an excellent ²H signal coverage and uniform distribution inside a head-size phantom.

Method:

The four channel ²H loop coils (width x height: 22cmx14cm) are overlaid around a head helmet of size 23.5cmx19.5cm (Fig. 1). The non-adjacent ²H loops are decoupled using a resonant inductive decoupling circuit (RID) [5] elongated by coaxial cable at both ends, and the adjacent ²H loops are decoupled using the geometric overlapping technique. Four ¹H microstrip elements (15 cm long) [6], each having a shorter physical length than that of a traditional ¹H dipole at 7T, are positioned in the middle of the ²H loops. The ¹H microstrip is mechanically easy to install, and shows less coupling with other electronic components (diodes, preamps, etc.).

A head-size lightbulb-shaped phantom (17cm in length, and 17 cm in diameter) with similar loading as the human head was used for loading the array coil in all bench tests and imaging experiments. The four channel ²H coils were driven in a circularly polarized (CP) manner, with each of the 4 elements transmitting equal magnitudes with the appropriate phase shift, and each channel receives the ²H signal independently. A high-resolution chemical shift imaging (CSI) sequence was used to acquire DMRSI data: FOV 20x20x20 cm3, TR = 540 ms, matrix size = 17x17x13 (nominal size of cylindrical voxel = π0.6x0.6x1.5 cm3), TA = 8:54 min. For processing the ²H CSI spectra, a 10-Hz line broadening was applied, number of points for the left shift was 3, the zero-order phase (rp) for each 2H channel is shown in Table 1. The phase corrected CSI of four ²H channels were added together to generate the combined ²H CSI. The signal-to-noise ratio (SNR) for each CSI voxel was calculated as the resonance intensities of the combined CSI spectra divided by the noise level of the combined CSI spectra. The four ¹H microstrip elements were driven in CP mode, and the combined images are co-registered with the ²H CSI images.

Results:

Fig. 1 shows the prototype of the 4-ch ²H TxRx/4-ch ¹H TxRx head array coil and the set up for performing ¹H MRI and ²H MRSI at 7T. The S parameters of the ¹H microstrip and ²H loop coils are shown in Fig. 2, indicating excellent decoupling with the S12 of ~-20dB or better among all ²H loop coils.

Fig. 3A shows good signal coverage of four ²H channels combined CSI overlay on co-registered proton images across a 13.8 cm slab (nine CSI slices, each with 1.54 cm thickness; axial orientation). Fig. 3B shows phase corrected individual CSI of four ²H channels in the same slice location at the center of Fig. 3A. Each ²H channel demonstrates consistent water signal distribution and spectral phases across the entire CSI after phase correction, indicating excellent decoupling between 2H coils. Table 1 summarizes the Q factors and global FID noise, amplitude and SNR of each ²H channel. The nominal 90-degree voltage for each channel was ~100 volts at 0.5 ms pulse length, and the zero-order phase shift (rp) for each channel was close to 90 degree apart. The four channels’ combined B₁⁺ and SNR maps are shown in Fig. 4A and Fig. 4B, respectively. The majority of measured SNRs of the ²H signal originating from the natural abundance HDO content are above 100 for the nominal voxel size of 1.7cc (Fig. 4B).

Conclusion:

A novel head array coil by integrating 4-channel ²H loop coils and 4-channel ¹H microstrip elements reported in this work provides excellent DMRSI and proton signal coverage in a head-size lightbulb-shaped phantom. Each of the ²H coil element shows similar and excellent performance regarding Q factor, nominal 90^o voltages and noise level; and is well decoupled to each other. This multinuclear array coil will be highly valuable for imaging human brain glucose metabolism at 7T under normal and diseased state.

Acknowledgements

This work was supported in part by NIH grants of R01 CA240953, U01 EB026978, R01 MH111413, P41 EB027061.