Introduction

Ultra-high field MRI (≥7T) has the potential to become a clinically viable tool due to its higher intrinsic SNR and CNR compared to lower magnetic field. This increase in SNR and CNR allows for higher spatial resolutions and higher sensitivity in image acquisition than conventional clinical MRI. However, one of the main technical challenges with increasing field strength is the static magnetic field inhomogeneity. Variations in the B₀ field lead to local frequency offsets that can cause artefacts during MRI and MRS measurements.¹ The magnetic susceptibility differences between tissue interfaces is one of the leading factors for local magnetic field inhomogeneities due to the linear increase in field offsets caused by tissue susceptibility at higher fields. To reduce the spatial variations, higher order spherical harmonics are needed in addition to the standard hardware B₀ shimming methods of the scanner, which would require a high number of shim coils and channels.² An alternative is the use of local shim coil arrays that can provide the high order of spatial field variation albeit confined to a small region. Moreover, local shim coils would couple less to the conductive bore thereby may have negligible eddy currents thus could generate spatially varying magnetic fields that counteract the B₀ field variations in real time. Here, we demonstrate that B₀ field variations in the liver can be substantially reduced using a local array of only 16 shim coils, when compared to the standard hardware correction method at 7T.

Methods

A whole-body 7T MR scanner equipped with a multi-transmit RF system (Achieva, Philips Health Care, Cleveland, OH, USA) was used to acquire B₀ maps of the liver in five healthy volunteers. Eight transceiver fractionated dipole antennas with 16 additional receive loops interfaced to 8 parallel 2kW peak power amplifiers, were positioned symmetrically around the body at the position of the liver.³ RF phase shimming was performed to maximize transmission efficiency and optimize B₁⁺ field homogeneity in the liver. Five dual-echo B₀ maps (GE, 386×410×180mm³ FOV, 6×6×6mm³ voxel size, FA=4°, TR=6ms, TE=1.413ms, ΔTE=1ms) were acquired during free breathing as well as in inspiratory and expiratory breath-holds. The local shim coil array consisted of 16 circular loops, each with a diameter of 5cm. The coil diameter was optimized using simulation libraries provided by MR Shim GmbH (Reutlingen, Germany) between the range of 1-10cm. Each loop consisted of 20 coil windings of enameled copper wire with a diameter of 0.56mm.⁴ After the diameter was optimized, electromagnetic simulations were performed using 30 slices of the gradient dual-echo field maps to calculate the magnetic field that is generated by a constant electric current through the shim coil array using the Biot-Savart Law.⁵ The coils were arranged in two rows of eight around the body (Figure 1). These simulations were validated afterwards with a measurement in the 7T MRI scanner. B₀ maps with equal sequence parameters were acquired in a phantom (octagon, 39×40×20cm, filled with polyvinylpyrrolidone) using a single shim coil driven with a current of 8A. The difference between the simulation and the measured B₀ map was calculated as the RMSE and the correlation with the Pearson correlation coefficient.⁶

Results

The simulations showed that the use of the local array of shim coils led to 65±8.3% improvement of the B₀ homogeneity during free breathing, 80±4.5% improvement during inspiratory state, and 75±12% improvement during expiratory state when compared to unshimmed B₀ map acquired during free breathing (Figure 2, 3). Compared to the field homogeneity when using the standard hardware-based method up to second order terms, the B₀ variation in the liver can be reduced by 9±5.1% in free breathing, 46%±16 in inspiratory state, and 18±12% in expiratory state. Comparison of the simulations with the phantom 7T measurements, shows clear B₀ differences (Figure 4, 5). The difference is close to zero further away from the shim coil. The Pearson correlation coefficient between simulations and measurements is 0.70 and the RMSE is 196Hz with the maximum B₀ field variation value measured at 5.34×10³Hz and the maximum of the simulation at 1.66×10³Hz.

Discussion

Substantial gain in field uniformity can be obtained with organ specific local B₀ shim coils when compared to generalized static spherical harmonic shimming. In our study, we simulated dynamic shimming by using breath-hold, while real-time shim updating was not implemented. Direct use of simulated fields in real time will cause a mismatch as shown by the phantom measurements that differed from the simulations. While this can be prevented by obtaining fast B₀ calibration maps from all shim coils, the magnitude of field variation is similar, thereby the simulation results do provide validation for the power requirements of the coils and amplifiers.

Conclusion

This study shows that local B₀ field variations in the liver at 7T can be reduced by using an array of local shim coils. B₀ shimming improves by 9±5.1% in free breathing up to 46% in inspiration when using a local shim coil array in comparison to standard hardware shimming B₀ shimming. As local shimming becomes more important at ultra-high magnetic field strengths, the use of embedding local shim coils in RF receiver arrays may become attractive, particularly for larger organs such as the liver.

Acknowledgements

No acknowledgement found.