4218

Design of shim coils by combination of simple building blocks: Romeo and Hoult revisited
Michael Packer1, Mark Fromhold1, and Richard Bowtell2
1School of Physics and Astronomy, University of Nottingham, Nottingham, United Kingdom, 2Sir Peter Mansfield Imaging Centre, University of Nottingham, Nottingham, United Kingdom

Synopsis

Shim coils can be formed from combinations of small numbers of simple “building block” coils (loops and saddles). In a seminal 1984 paper, Romeo and Hoult, showed how to combine building blocks to produce coils that generate a given spherical harmonic, whilst eliminating contamination from as many unwanted spherical harmonic terms as possible. By applying updated numerical optimisation approaches to this problem, we show how to produce improved zonal shim coils. Designs of Z2, Z3 and Z4 coils formed from three pairs of loops, which are shorter than previously reported coils while having large regions of conformity are described.

Introduction

Shim coil systems form an essential component of human MRI scanners, pre-clinical MR systems and NMR spectrometers1,2. They are also key for any magnetic field generating system in which high field homogeneity is required. In most cases a shim system comprises a set of multiple different coils wound on a cylindrical former, each generating a spatially-varying field that corresponds to a spherical harmonic function. Although shim coils can be formed from distributed wire-paths using methods originally developed for MRI gradient coil design3,4, in many cases shim coils are still formed from small numbers of simple “building block” coils (loops and saddles), since such coils are simpler to fabricate and bring together into a shim set. Methods for combining these building block coils to produce shim coils that generate a given spherical harmonic, whilst eliminating contamination from as many unwanted spherical harmonic terms as possible were described in a seminal paper1 by Romeo and Hoult. The resulting zonal and tesseral coil designs have been widely used. Here we revisit the analysis used in designing these coils and demonstrate that using current numerical optimisation approaches it is possible to produce improved zonal coil designs with small numbers of building blocks.

Methods

We consider coils composed of pairs of wire loops separated by distance $$$2d$$$ on the surface of a cylinder of radius $$$a$$$ (Fig. 1). It is well known that the z-component of the magnetic field from such an arrangement is given by
$$B_{z}\left(\rho,z\right)=\frac{2\mu_0 NI}{\pi a}\int_{0}^{\infty}\mathrm{d}k \ k\begin{bmatrix}\cos\left(\frac{kz}{a}\right)\cos\left(\frac{kd}{a}\right) \\\sin\left(\frac{kz}{a}\right)\sin\left(\frac{kd}{a}\right) \end{bmatrix}I_{0}\left(\frac{k\rho}{a}\right)K_{1}(k) \>\>\>\>\> [1],$$.

where $$$ρ$$$ and $$$z$$$ are cylindrical co-ordinates, $$$N$$$ and $$$I$$$ represent the number of turns and current, and the upper and lower terms represent the cases where the current in the two loops flows in the same or opposite senses. Forming a Taylor series about the origin, we can decompose the field into a series of weighted spherical harmonic terms;

$$B_{z}\left(\rho,z\right)=\frac{2\mu_0NI}{\pi a}\begin{bmatrix}b_0(\frac{d}{a})+\left(\frac{\rho^2}{4a^2}-\frac{z^2}{2a^2}\right)b_2(\frac{d}{a})+\cdot\cdot\cdot\\\frac{z}{a}b_1(\frac{d}{a})+\left(\frac{z\rho^2}{4a^3}-\frac{z^3}{6a^3}\right)b_3(\frac{d}{a})+\cdot\cdot\cdot \end{bmatrix} \>\>\>\>\> [2]$$

where
$$b_{2n}(\chi)=(-1)^n\frac{\partial^{2n+1}}{\partial \chi^{2n+1}}\Bigg(\frac{\pi\chi}{2\left(1+\chi^2\right)^{\frac{1}{2}}}\Bigg) \>\>\>\>\> [4].$$
and
$$b_{2n-1}(\chi)=(-1)^n\frac{\partial^{2n}}{\partial \chi^{2n}}\Bigg(\frac{\pi\chi}{2\left(1+\chi^2\right)^{\frac{1}{2}}}\Bigg) \>\>\>\>\> [3] $$

By combining together $$$P$$$ loop-pairs and choosing appropriate values of the number of turns $$$N_p$$$ and separation $$$d_p$$$ for each pair, $$$P$$$ spherical harmonics can potentially be eliminated from the field, producing a coil that predominately generates the desired spherical harmonics - i.e. we ensure that $$$\sum_{p=1}^P N_pb_{l}\left(\frac{d_p}{a}\right)=0$$$ for an appropriate set of $$$l$$$-values. In general there are multiple combinations of $$$N_p$$$ and $$$d_p$$$ values which will null the unwanted harmonics, and an optimal choice from the set of solutions can be made so as to produce a coil of limited length with maximum efficiency per unit resistance and a reasonable number of turns per loop.

Here we designed zonal coils with three pairs of loops, whose performance was compared with previously described coils formed from two loop pairs1. The addition of one loop pair allowed the nulling of three,rather than two, unwanted spherical harmonics and the additional degrees of freedom were used to design shorter Z2, Z3 and Z4 coils (with a length to diameter ratio less than 1.4) and larger regions over which the generated field follows the desired spatial variation compared with previously described coils1.

Results

Figure 2a shows the number of turns and loop positions in the Z2 coil formed from three loop pairs, which was designed to null Z0, Z4 and Z6 terms. A minus sign on the number of turns indicates that the current circulates in the opposite sense. Figure 2c, shows the contours of the field generated by the coil, scaled so that $$$B_z=1$$$ at $$$z=a$$$ and $$$\rho=0$$$. Figure 2e shows the field variation with $$$z$$$ on axis (i.e. when $$$\rho=0$$$). The dotted lines mark the extent of the region within which the generated field deviates by less than 5% from the desired form. The positions of the outermost loops ($$$d_3$$$) are also indicated on this plot. Figures 2b, d and f, show similar plots for a previously described Z2 coil1, designed to null just the Z0 and Z4 terms. The shorter length of the 3-loop-pair coil and the greater extent of the region over which it generates a pure Z2-field is evident. Figures 3 and 4 show similar plots for Z3 and Z4 coils. Figure 5 details the $$$d$$$-values, number of turns per loop, extent of the region of field conformity on axis, efficiency and efficiency scaled by the total number of turns, for each of the coils.

Discussion

The coil lengths are reduced by 20%, 20% and 60% for the Z2, Z3 and Z4 three-loop-pair coils, compared to the corresponding two-loop-pair coils1. The extent of the region of field conformity is also increased by 20% for the new Z2 and Z3 coils. The comparison for the Z4 coil is harder since the previously described coil produces a small Z0 contribution (Figure 4f), but it is notable that the extent of the region of conformity of the new coil design is 60% of the coil length.

Conclusions

By applying current numerical optimisation approaches to the analysis of shim coils formed from simple building blocks, shorter shim coils with larger regions of conformity have been designed. The focus here is on zonal coils, but similar approaches can be used for tesseral coils.

Acknowledgements

This work is supported by the UKRI EPSRC UK National Quantum Technology Hub in Sensing and Timing - EP/T001046/1.

References

1. Romeo, F. and D.I. Hoult, Magnet Field Profiling - Analysis and Correcting Coil Design. Magnetic Resonance in Medicine, 1984. 1(1): p. 44-65.

2. de Graaf, R.A. and C. Juchem, B-0 Shimming Technology, in Magnetic Resonance Technology: Hardware and System Component Design, A.G. Webb, Editor. 2016. p. 166-207.

3. Forbes, L.K. and S. Crozier, Novel target-field method for designing shielded biplanar shim and gradient coils. Ieee Transactions on Magnetics, 2004. 40(4): p. 1929-1938.

4. Poole, M. and R. Bowtell, Novel gradient coils designed using a boundary element method. Concepts in Magnetic Resonance Part B-Magnetic Resonance Engineering, 2007. 31B(3): p. 162-175.

Figures

Figure 1 Single loop coil geometry with radius $$$r=a$$$ displaced from the origin in the $$$z$$$-axis to $$$z=d$$$ shown in cylindrical polar coordinates. The building block for zonal coils consists of a pair of loops at $$$z=\pm d$$$ with current circulating in the same or opposite senses in the two loops.

Figure 2 (a,b) Z2 coil configuration for a 3-loop pair and previously described coil1, respectively, showing number of turns and current direction (left) and loop position $$$d_p$$$ (right). All other specifications are detailed in Fig. 5. (c,d) Contour plots of the normalised magnetic field generated by the coils (a,b), respectively, where the desired harmonic is unity at $$$z/a=1$$$. (e,f) Field profile along the $$$z/a$$$-axis with dotted lines showing the region of 5% conformity while $$$\pm d_{3,2}$$$ markers indicate coil length for the configurations (a,b) respectively.


Figure 3 (a,b) Z3 coil configuration for a 3-loop pair and previously described coil1, respectively, showing number of turns and current direction (left) and loop position $$$d_p$$$ (right). All other specifications are detailed in Fig. 5. (c,d) Contour plots of the normalised magnetic field generated by the coils (a,b), respectively, where the desired harmonic is unity at $$$z/a=1$$$. (e,f) Field profile along the $$$z/a$$$-axis with dotted lines showing the region of 5% conformity while $$$\pm d_{3,2}$$$ markers indicate coil length for the configurations (a,b) respectively.

Figure 4 (a,b) Z4 coil configuration for a 3-loop pair and previously described coil1, respectively, wshowing number of turns and current direction (left) and loop position $$$d_p$$$ (right). All other specifications are given in Fig. 5. (c,d) Contour plots of the normalised magnetic field generated by the coils (a,b), respectively, while the desired harmonic is unity at $$$z/a=1$$$. (e,f) Field profile along the $$$z/a$$$-axis with dotted lines showing the region of 5% conformity while $$$\pm d_{3}$$$ markers indicate coil length for the configuration in (a).

Figure 5 Table detailing coil specifications showing number of turns, $$$N_p$$$, loop positions, $$$d_p$$$, efficiency, $$$\eta$$$, efficiency scaled by total number of turns in the coil, and z-extent of the region of conformity (<5% deviation) in units of $$$a$$$, for designs previously described by Romeo and Hoult1 and optimised 3-loop-pair coils.


Proc. Intl. Soc. Mag. Reson. Med. 28 (2020)
4218