MRI is an indispensable tool for radiologists and has emerged as the pre-eminent means of detecting brain injuries or disease. However, the large size and installation complexity of modern clinical MRI scanners mean the installed base of MRI systems remains relatively small and concentrated in developed, urban centres, despite the usefulness of the data they provide. An important step towards improving the accessibility of brain imaging MRI is to dramatically reduce not only the system size, but also the infrastructure needed to support the system.

The size of conventional clinical MRI systems is largely driven by the requirements for a cost-effective magnet to produce a very strong (1.5 – 7 T) and homogeneous (<10 ppm) magnetic field. Counterintuitively, cost of such MRI magnets become higher as they become more compact [1], meaning very compact head/brain imaging magnets with the shoulders excluded from the magnet have not yet been manufactured. However, recent results [2] have demonstrated clinical quality MRI images can be obtained in a less homogenous (< 500 ppm variation) magnetic field than typically targeted in MRI magnet design.

Here, we describe the design and progress in manufacture of a 1.5 T MRI magnet which will create a specific modest uniformity magnetic field over a brain-sized imaging volume. The magnet will be used in an MRI system designed for use with emerging MRI techniques developed at University of Minnesota as part of a grant led by University of Minnesota (USA), funded by National Institutes of Health (U01-EB025153). The magnet is solenoidal but asymmetric, orientated vertically and highly compact since it completely excludes the shoulders from the magnet bore. To improve patient comfort and reduce claustrophobia, a window is provided through the magnet cryostat. To reduce the operational costs of the magnet, and to give the most robust solution possible, we have chosen to wind the magnet from REBCO conductor using a combination of no-insulation (NI)-style coil winding and active quench protection.

As can be seen from Figure 1A, the magnet is unshielded and comprises 23 double pancake coils arranged into five coil packs. The pancakes are graded into 3- and 4-mm width coils where one pack has negative polarity. The magnet uses approximately 14 km of REBCO and is designed to operate at 30 K, conduction cooled by a Cryomech PT-90 single stage pulse-tube cryocooler. To ensure the magnet operates at less than 75% of I_c at 30 K, we have performed full in-field anisotropy measurements using our SuperCurrent system [3] on numerous REBCO samples from our supplier. For this conductor, we found there to be a good correlation between the 77 K self-field performance and in-field performance. We therefore graded and arranged conductor in the magnet using the 77 K self-field value to minimise the in-field variation of I_c within the magnet when the full anisotropy of the conductor is considered.

In a previous study [4], we described our experience making a small magnet using NI-style coils, where we were able to tune the contact resistance of the coils by adjusting the concentration of copper powder in a heavily filled epoxy coil encapsulant. The study showed that, by filling the epoxy with diamond powder of a particular size in addition to the copper powder, we could not only prevent conductor thermal degradation, but also precisely set the gap between adjacent coils within the coil as will be required for winding an MRI magnet. We have continued to refine the filled epoxy system and have found, by using metallic coated diamonds in place of copper powder, we can exert a similar level of control over contact resistance as previously reported, but the contact resistance now has little variation with temperature.

When using NI-style coils in an MRI magnet, a critical parameter is the ramp down time of the magnet. Should an emergency occur, it is important for patient and operator safety to be able to shut the magnet down within 30 s. We therefore developed a lumped-circuit model [5] of the magnet to calculate the contact resistance required to allow sudden discharge of the magnet within 30 s. This resulted in a target contact resistance of 7x10^-6 Ωm². Using a series of small test coils with different concentrations of metallic coated diamonds relative to uncoated diamonds, we were able to determine the required concentration to achieve the target contact resistance. Subsequently, we developed a multi-physics lumped-circuit quench model of the magnet to check the performance of the magnet both during sudden discharge and in the event of quench. In both cases we saw a near-monotonic decay of coil current in all coils and a maximum temperature of approximately 80 K.

At the time of writing, we have wound more than half the coils for the magnet. The coils have been individually tested to ensure no degradation has occurred during winding and their contact resistance is within acceptable limits. To date we have been able to achieve the target contact resistance and the coils appear to be robust to thermal cycling. Other critical magnet components have been completed, including the cryostat and the patient interface to the system, as shown in Figure 1B.

In summary, we have designed and are part-way through constructing a highly compact REBCO MRI magnet for brain imaging. Completion of the magnet in 2021 and demonstration of images by our collaborators at University of Minnesota in 2022 will show clinical MRI may be performed without the requirement for a highly uniform magnetic field. The resulting MRI systems will be much more compact and robust, paving the way to a broader uptake of brain imaging MRI.

References:
[1] Supercond. Sci. Technol. 26 093001 (2013)
[2] Magn Reson Med. Feb;85(2):831-84 (2021)
[3] Rev. Sci. Instrum. 85, 113907 (2014)
[4] IEEE Trans Appl Supercond. 31(5):4601105 (2021)
[5] IEEE Trans Appl Supercond. 27(4): 4603505 (2017)

Keywords: MRI, REBCO, Quench, No-insulation coils