Synopsis

Keywords: Shims, Shims

An MRI scanner equipped with global shim systems fails to reach state-of-the-art when shimming two isolated ROIs simultaneously for two reasons: non-optimal spherical harmonic based shimming routine, and significant high-order B0 inhomogeneities, even though the two-area shimming can be essential in scan scenarios, such as bilateral breasts or dyadic brains. To address these challenges, a hybrid active and passive local shimming (HAPLS) technique is proposed to shim two isolated areas in one FOV simultaneously. Both the simulation and experimental results validated that HAPLS can complementarily address the bifocal and high-order inhomogeneities, and locally improve in vivo B0 homogeneity as well.

Introduction

An MRI scanner equipped with global shim systems fails to reach state-of-the-art when shimming two isolated ROIs simultaneously for two reasons: non-optimal spherical harmonic-based shimming routine, and significant high-order B₀ inhomogeneities, even though the two-area shimming can be essential in scan scenarios, such as bilateral breasts or dyadic brains^1-4. To address these challenges, a hybrid active and passive local shimming (HAPLS) technique is proposed to simultaneously shim two isolated regions of interest areas within the whole FOV.

Methods

All simulations, optimization, and experimental validation were performed on a 3T Siemens Prisma scanner. The top three dominant SH terms of the B₀ field within 50-cm DSV (Figure 1(a)) are all ultra-high-order ones – a(12,0), a(14,0), and a(8,0), and they are at 12.3, 5.84, and 1.94 (in ppm), respectively. Shimming two isolated regions (Figure 1(b)) simultaneously is challenging because: (1) the background B₀ field in both areas contains up to 14th order SH field inhomogeneities; (2) the scanner’s built-in SH-based in vivo shim routines are designed for one volume, so it cannot effectively shim two off-center volumes simultaneously and optimally as shown in Figure 2. The peak-to-peak (pk-pk) ppm for the two targeted imaging regions are 3.39 ppm and 8.92 ppm, respectively. To address the challenge of shimming two-region MRI, a hybrid active and passive local shimming (HAPLS) technique is proposed to shim two isolated regions within one large FOV simultaneously by taking advantage of both PS and AS.
Local Passive Shimming
A local passive shimming (PS) system is constructed by optimized bilateral 20x11 ferromagnetic chip arrays to compensate for the magnet’s significant high-order B₀ inhomogeneities at the boundary of the manufacturer’s specified homogeneous volume, thus locally improving the available FOV. The optimized PS chip arrays were affixed surrounding the two imaging regions with a CNC-machined acetal holder and 3D-printed fixtures as shown in Figure 3 (a-b). For the optimization of PS, the shim field generated by the PS chips was calculated by a current model^5-6, and the thickness of the chip stack was taken as variables to optimize the standard deviation (SD) of the B₀ field by a genetic algorithm in MATLAB.
Local Active Shimming
The active shimming (AS) is used to homogenize the center frequencies of two regions which may be shifted differently after applying PS. Furthermore, AS will deal with the residual lower-order and the additional subject-specific field inhomogeneities. The local AS consists of 40-channel DC loops powered by 64-channel current amplifiers as shown in Figure 3(c-d)^7-9. The volumetric field maps of each shim coil were acquired by the double-echo gradient-echo field mapping sequence, and then used for the optimization of the current distribution of AS. The optimization of AS is conducted on the B₀ field after PS, where the high-order inhomogeneities have been largely removed. The SD across two regions was minimized to get the optimal current distribution by a nonlinear constrained optimization algorithm (fmincon) in MATLAB.

Results

The optimized distributions of the PS chip array, and the optimized current distribution of AS were shown in Figure 3(a,f), respectively. The experimental setup to perform HAPLS was shown in Figure 3(e), and the experimental field maps are shown in Figure 4(a-b). Compared to B₀ field in Figure 2(a), the high-order inhomogeneities in transversal xy-planes were greatly reduced after PS. Compared to no shim, the RMS decreased from 0.473 to 0.255 ppm by 46.1% with HAPLS, and the nonlinearity of the B_Z field along the x-axis was significantly reduced as shown in Figure 4(c). The volume ratio containing MR voxels within a 0.5-ppm frequency span increased from 64.3% to 81.3% by 26.3% as shown in Figure 4(d).

Discussion

To demonstrate the capability of the proposed HAPLS method for shimming two brains in dyadic fMRI, the 3D B₀ maps of two brains were acquired, and then applied HAPLS to them. The PS is the same as that shown in Figure 3(a), and the AS was re-optimized. The B₀ field within two brains before and after applying simulated HAPLS is shown in Figure 5. With HAPLS, the RMS within two brains decreased from 0.461 ppm to 0.211 ppm by about 54.2%, and the 95% pk-pk ppm was reduced from 1.54 ppm to 0.87 ppm by 44%. Due to that the 40-ch AS coil array is more confined on +x and -x regions, the regions close to the isocenter of the scanner were not improved too much. In the dual-head AC/DC coil array we are building, both local FOVs will be 360-degree surrounded, and a better shim performance will be expected.

Conclusion

The hybrid active and passive local shimming was proposed to shim two-region MRI studies with high-order B₀ field inhomogeneities where conventional SH shimming is unable to achieve the required improvements. The PS focuses more on locally improving the high-order field inhomogeneities which are an intrinsic property of the scanner type, whilst the AS concentrates mainly on the residual low-order terms and magnetic susceptibility effects. Both the simulation and experimental results validated the feasibility of the proposed method, and suggested that the HAPLS technique will locally improve in vivo B₀ homogeneity as well.

Acknowledgements

The authors would like to thank for the grants from NSF 1926789 and NIH U24EB028984, also thank the support from Bernd Stoeckel.