Introduction

Ultra-high field (UHF) MRI systems, 7 T (≥300 MHz) or higher, have the benefit of increased signal to noise ratio allowing finer structures at the mesoscopic scale to be visualized and smaller physiological effects to be detected ¹. However, UHF comes with additional challenges including non-uniform radio frequency (RF) fields, greater susceptibility artifacts, and larger energy deposition ². Phantoms are required in order to test quantitative pulse sequences, and it becomes increasingly important that artifacts in phantoms be as similar as possible to those observed in vivo. In this study, a phantom was designed to evaluate chemical exchange saturation transfer (CEST), magnetization transfer (MT), and T₁ mapping sequences on UHF MRI systems.

Methods

Phantom Design: Figure 1 shows a phantom schematic. A spherical geometry was required for B₀ shimming and to fit a head coil. The fill port locations were chosen to reduce air bubble formation, which would cause field inhomogeneities. Three compartments were included to allow ranges of T₁, CEST, and MT parameters similar to those observed in vivo. The phantom was 3D printed (Viper si, 3D Systems) out of Accura ClearVue.
Sample Preparation: Agar (1% mass/volume) provided MT contrast, Copper (II) sulfate pentahydrate (CuSO₄·5H₂O) to affect T₁, and ammonium chloride (NH₄Cl) to provide CEST contrast. Compartments were filled as follows:

1. 0.5mM CuSO4·5H2O + 100mM NH4Cl
2. 0.5mM CuSO4·5H2O + 50mM NH4Cl
3. 1.0mM CuSO4·5H2O + 50mM NH4Cl

MRI Acquisition: 3T scanning was performed using Achieva system (Philips Medical Systems, Best, The Netherlands) using a Q-body transmit coil and a SENSE-Head-8 receive coil. Following a structural 3D FLAIR sequence, a single slice through all three compartments was selected for quantitative imaging. Second-order, pencil-beam shimming was performed over the slice of interest. Saturation-transfer weighted images were obtained with saturation transfer preparation (four 242.5ms block pulses with 2.5ms gaps; amplitudes and offsets in Table 1) and turbo field echo (TFE) readout (TFE factor = 26). In addition to the CEST- and MT-sensitive images, a WAter Saturation Shift Referencing (WASSR ³) spectrum was obtained to calculate a B₀ map and a WAter frequency Shift B₁ Amplitude (WASABI ⁴) was used to calculate a B₁ map. A T₁ map was calculated from five IR-prepared TSE sequences (TR/TE = 5000/6.2ms; TI = 50 - 4500ms; TSE factor = 10).
7T scanning was performed using Discovery MR950 system (GE Healthcare, Waukeshaw, WI) using a 32-channel receive head coil and a 2-channel transmit birdcage head coil. A 3D FLAIR sequence was used for structural scanning. Saturation-transfer weighted images were acquired using the epiCEST sequence with saturation transfer preparation (one block pulse of 1500ms; amplitudes and offsets in Table 1) and a single-shot EPI readout. Three images were acquired consecutively at each saturation offset, and only the third image was used in the analysis because it was in saturation steady state. The GE Field Map protocol was used to produce B₁ and B₀ maps. The standard “Autoshim” procedure was used to provide x-, y- and z-gradient shims over the whole phantom.

Results

Figure 2 shows imaging results from 3T. 3D FLAIR (Fig. 2a) shows the three phantom compartments. The T₁ map (Fig. 2b) shows the lower T₁ in the compartment with higher concentration of CuSO₄·5H₂O, and the higher T₁ in the other two compartments. B₀ and B₁ maps (Fig. 2c&d) show the variation in B₀ and B₁ across the phantom.
Figure 3 shows imaging results from both phantom and human brain acquired at 7T. The variation in B₀ is similar in the phantom and in the brain (Fig. 3b&f), except near the sinus cavities which cause larger field variations in the brain. The B₁ maps (Fig. 3c&g), display a similar pattern in both phantom and brain with higher flip angles near the center, and lower flip angles near the edges.
Figure 4 shows CEST spectra from brain and from phantom compartments with high and low CEST contrast. The size of the CEST peaks are comparable to that observed in vivo.

Discussion

The similarities in variation and pattern in B₀ and B₁ fields between phantom and brain data indicates that both will require a similar degree of correction.
MT and CEST contrast are both present in the phantom, in addition to having T₁ values in a similar range to that of brain. This is important as the CEST signal is affected by both MT and T₁. The amount of CEST contrast is similar to that in brain (Fig. 4) which will allow the variability due to imaging artifacts to be estimated. The variation in T₁ values in the center and right-most compartments (Fig. 2b) are the result of small effects of NH₄Cl on T₁. It is possible to adjust the amount of CuSO₄·5H₂O in order to match the T₁ values perfectly, but this is unlikely to be necessary.

Conclusions

The three-compartment spherical phantom is a durable, user friendly, and customizable research tool which successfully mimics the CEST, MT, and T₁ properties of the human brain as well as having similar variations in B₀ and B₁. The multi-compartment phantom is ideal for development and testing of quantitative CEST sequences.

Acknowledgements

We thank the Canadian Institutes for Health Research (PJT156252) for financial support.