Introduction

Low-to-mid field systems have nearly been abandoned for high-field, high performance ones. However, recent improvements in MR hardware (receiver arrays, power systems, shim arrays, field-monitoring), as well as algorithms and cheap computation (parallel imaging, compressive sensing, artifact corrections) are stimulating a resurgence in interest of upgrading these lower-end magnets with modern technology to achieve high-performance at significantly lower cost, better accessibility, improved portability and simpler siting requirements. In recent years we have seen several research programs and new commercial companies displaying dedicated low-field or specialized systems for extremities, head and pediatric covering fields between 6.5mT through 0.35T, 0.5T, 0.55T and 1T. All the above is nicely reviewed in [1].

Permanent magnets are a class of low-end, inexpensive magnets with the advantage of requiring no power and no cooling to maintain the field. Dedicated permanent magnet MRI can be small, shielded, safe and easy to site. However, these systems can suffer from field inhomogeneity, drift and hysteresis. Localized non-linear shim arrays [2] are a possible solution as demonstrated by [2, 3]. Since they are placed close to the subject, shimming can be achieved with little current, thus making the system simple and affordable.

Permanent magnets in a Halbach configuration generate a uniform field that is orthogonal to the bore [4]. This enables the use of very efficient solenoid RF transmit and receive coils. Since solenoids produce very uniform field inside, and little outside, we look at constructing a shim loop array that covers the solenoid [5]. Each of the elements are orthogonal and naturally decoupled from the RF coil, and the array takes little space and elements are maximally close to the subject. Figure 1 demonstrates the motivation for our work. A field map in a phantom of an Aspect Imaging (Shoham, Israel) Wrist II 1T system shows significant field inhomogeneities with global linear shims. Simulations of a 16 channel local shim array can provide substantial slice-by-slice local field homogeneity corrections in all directions.

The aim of the study is to demonstrate the efficacy of using a localized shim loop array, wrapped around a solenoid RF coil both in simulations and practice. In addition, we aim to demonstrate minimal interaction between our shim array and the RF system.

Methods

Methods: All studies were performed using an Aspect Wrist II 1T MRI system, using a wrist solenoid transmit receive coil. B₀ field mapping was performed using a product B₀ mapping 3D RF spoiled gradient echo sequence.

Simulations: A Biot-Savart simulation was written to estimate the B₀ field for an arbitrary number of loops and sizes [6]. The code was implemented in Matlab. In addition, a constrained Least-squares shimming procedure for determining the shim-currents, given each loop contribution to the field map as well as targeted maps was implemented in Matlab using CVX [7, 8].

Implementation: A 1-sided 6 channel shim loop array (Figure 1a) was constructed using 1.25µm thick, copper tape on a polycarbonate plate. These were arranged in 2x3 configuration covering 100 cm² coronal area of the coil. Each loop was driven with direct current generated by a lab-power-supply via shielded twisted-pair wires fed into the bore (Figure 1a).

Experiments: Calibration was performed using a CuSO₄ doped water-bottle phantom for field mapping. The background B₀ map was obtained, followed by 6 scans to map the field generated individually by each loop. The loop current in the calibration was set to 1A. Shim-currents were then applied, based on the optimization to locally shim specific coronal slices. In addition, simulations were also performed to validate the shimming performance for 6-channels, and predict the shimming ability for 16-channels. To measure the effect of the shim array on SNR, we computed the unloaded Q-factor of the RF coil with-and-without the shim array. Spin-echo images of a calibration phantom were acquired for image SNR comparison.To measure the effect of the shim array on B₁ we ran a product B₁ mapping sequence with and without the shim-array. B₁ were computed via the Aspect Imaging software.

Results and Discussion

Figure 2 shows good correspondence between the simulated and measured B₀ field maps of individual loops in the array. Figure 3 shows localized shimming results of two coronal slices, demonstrating improved homogeneity over global linear shimming, and good correspondence to simulations. 16-channel array simulations demonstrate significant future shimming capability. These results are re-emphasized by displaying the histograms of the maps in Figure 4. SNR and B₁ measurements indicate minimal coupling between the RF coil and the shim array. Measured SNR was reduced by 10%, and the B₁ field map did not show any observable difference (Figure 5).

Conclusion

We demonstrated that localized shim arrays can be integrated with solenoid RF coils, with minimal interactions for shimming B₀ inhomogeneities in permanent magnet systems. Simulations show significant improvements with 16 channels, which we aim to implement next.

Acknowledgements

We would like to thank Aspect Imaging for their support with hardware and software tools, and Jason Stockmann for providing shim boards and documentation.