A 32-channel multi-coil shim setup optimized for the human brain, pushing the limits of shimming at 9.4T
Ali Aghaeifar1,2,3, Jiazheng Zhou1,2, Irena Zivkovic4, Joshi Walzog1, Mirsat Memaj1, Theodor Steffen1, Rahel Heule1, Feng Jia5, Maxim Zaitsev5, and Klaus Scheffler1,3

1Max Planck Institute for Biological Cybernetics, Tübingen, Germany, 2IMPRS for Cognitive and Systems Neuroscience, University of Tübingen, Tübingen, Germany, 3Department for Biomedical Magnetic Resonance, University of Tübingen, Tübingen, Germany, 4C.J. Gorter Center for High Field MRI, Leiden University Medical Center, Leiden, Netherlands, 5Dept. of Radiology, Medical Physics, University Medical Center Freiburg, Freiburg, Germany


Multi-coil shimming is an effective approach to reduce B0 field inhomogeneity. In this work, we optimized a 32-channel multi-coil to fit best for shimming of the human brain. The individual coils are optimized in terms of size and position. The performance is compared with the conventional symmetric design.


Although ultrahigh field offers many advantages, static magnetic field inhomogeneity that scale with the field strength leads to serious challenges1. Signal loss, geometric distortions, banding artifacts and wide spectral linewidth are some of the well-known consequences of the poor B0 inhomogeneity. As an alternative to conventional B0 shimming using scanner’s built-in setup based on spherical harmonic (SH) functions, multi-coil setup is proposed that works more effectively for local inhomogeneities2, where an array of multiple small loops are placed around the volume of imaging. Principally, utilizing more number of loops gives higher degrees of freedom and therefore a better shimming performance3. However, the need for more dedicated amplifiers, challenges with maintenance and control make an increased number of loops highly challenging. As another possibility, optimizations of the coil’s geometry and the position has been proposed for a limited number of slices4. The objective of this work is keeping the number of the coils fixed, but optimize the position and the size of the individual coils that adapt best for shimming of the human brain while practical constraints are taken into account.


The optimization was performed on a 32-channel multi-coil setup. As a starting point, the square-shaped coils with a side length of 60mm were symmetrically positioned in four rows on a cylinder with a diameter of 165 mm. A nonlinear constrained optimization based on sequential quadratic programming (SQP) algorithm was implemented in MATLAB (Natick, MA) aiming to minimize the B0 uniformity for the whole brain and thus to achieve a better global shimming (Figure 1). Each coil has three degrees of freedom, the coil size, the coil position in axial z direction and angular coordinate (θ) on the cylinder. The maximum side length of the coil was limited to 100 mm, and the constraints for the spatial coil location was given by cylinder geometry. Whole brain B0 map of 12 volunteers after shimming with the scanner’s 2nd order SH setup was measured at 9.4T whole‐body Siemens scanner (Erlangen, Germany). The dataset was split into two groups of 8 and 4 maps for training and testing, respectively. Possible current amplitude was chosen during the optimization based on the coil size; smaller coils were permitted to be fed with a higher current (maximum 3.0 A) and vice versa (minimum 1.5 A). Similar to our previous setup5, the RF transceiver coil housed in the multi-coil cylinder, but the cylinder radius was chosen smaller in the current design to shorten the distance between the shim coils and the target. Performance of the optimized 32-channel multi-coil setup was evaluated with ∆B0 mapping and bSSFP sequences after global shimming of the whole brain. Field map was measured with a 3D double echo GRE sequence and the following parameters: TE1/TE2/TR = 2.6/7.4/15 ms, FA = 12o and 2 mm3 isotropic resolution. Images with a 3D balanced SSFP sequence were acquired using sign-alternated RF pulses with the following protocol: TE/TR = 5/10ms, FA = 14o and 1mm3 isotropic resolution. Dummy scans were employed at the beginning to reach a steady state. Both sequences were configured to a low bandwidth readout to minimize eddy current effects.


Figure 2 shows coils arrangement on a cylinder outer surface before and after optimization as well as the experimental realization of the optimized coil. Axes are according to the Siemens MR scanner’s gradient coordinate system. Among 32 coils, the size of 22 coils increased after optimization. Looking at the coils distribution, the highest density is above the frontal cortex, where the most severe field inhomogeneity in the human brain exists. Figure 3 compares B0 inhomogeneity in simulation after shimming with scanner’s 2nd order SH, symmetric 32-channel multi-coil and optimized 32-channel multi-coil for one of the test data. The optimized multi-coil improve the field inhomogeneities by 18.3% better (from 60.7 Hz to 49.6 Hz) compared to the symmetric multi-coil. Figure 4 shows the experimental result of the global shimming with a prototype of the optimized multi-coil in a healthy volunteer at 9.4T. B0 perturbation is well suppressed in the pre-frontal cortex with the optimized multi-coil even though the shimming is restricted to global shimming (no slice-specific shimming is performed). Figure 5 shows the impact of the shimming with the optimized multi-coil for banding artifacts reduction in balanced SSFP images. Banding artifacts well express the spots with poor B0 shim. Shimming with the optimized multi-coil setup successfully mitigated the banding artifacts in the frontal brain region close to the sinuses.


No acknowledgement found.


1. Uğurbil K. Imaging at ultrahigh magnetic fields: History, challenges, and solutions. Neuroimage. 2018;168:7-32. doi:10.1016/j.neuroimage.2017.07.007.

2. Juchem C, Nixon TW, McIntyre S, Boer VO, Rothman DL, De Graaf RA. Dynamic multi-coil shimming of the human brain at 7 T. J Magn Reson. 2011;212(2):280-288. doi:10.1016/j.jmr.2011.07.005.

3. Stockmann JP, Witzel T, Keil B, et al. A 32-channel combined RF and B0 shim array for 3T brain imaging. Magn Reson Med. 2016;75(1):441-451. doi:10.1002/mrm.25587.

4. Zivkovic I, Tolstikhin I, Schölkopf B, Scheffler K. B0 matrix shim array design-optimization of the position, geometry and the number of segments of individual coil elements. In: 33rd Annual Scientific Meeting of the European Society for Magnetic Resonance in Medicine and Biology. ; 2016.

5. Aghaeifar A, Mirkes C, Bause J, et al. Dynamic B0 shimming of the human brain at 9.4 T with a 16-channel multi-coil shim setup. Magn Reson Med. 2018;80(4):1714-1725. doi:10.1002/mrm.27110.


Figure 1. Flowchart of the optimization process. Optimization started from a symmetric design, and eight whole brain B0 maps were used as training dataset. Map of the individual coils was calculated at every iteration and shimming was performed on all training data. Performance of the optimization outcome was validated later with the test data.

Figure 2. Coils arrangement on the cylinder surface, symmetric (left) and optimized (right) design. Both design consist of 32 square-shaped coils. The coils are arranged in four rows for the case of symmetric design with the side length of 60 mm and 4.7 mm gap between two adjacent coils. The minimum and the maximum side length are 30 mm and 100 mm respectively for the optimized design. Description of the coordinate system respect to the patient is as following: +Y = Anterior, -Y = Posterior, +Z = Feet, -Z = Head, +X = Left, -X = Right

Figure 3. A) The changes in the standard deviation of B0 over the whole brain during optimization progress. The first iteration is performed with the symmetric design and resulted in STD = 60.7 Hz which at the last iteration decreased to 49.6 Hz. B) Comparison of the shimming performance after global shimming with scanner’s 2nd order SH, symmetric 32-channel multi-coil and optimized 32-channel multi-coil. The results are obtained in simulation. Comparison of the individual slices may show no improvement for a few slices after shimming with multi-coil since the shimming is performed in a global fashion.

Figure 4. Experimental results of the B0 shimming with the prototype of the optimized multi-coil at 9.4T. The whole brain was first globally shimmed with the scanner’s 2nd order SH setup.

Figure 5. Comparison of balanced SSFP images with two shimming approaches. Reduction in banding artifacts is apparent after shimming with the optimized 32-channel multi-coil setup.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)