Could this new superconductor help downsize MRI scanners?

Magnetic resonance imaging (MRI) scanners are an essential part of the medical toolkit. They allow us to see detailed images of almost any part of the body and are critical for diagnosing a huge range of conditions. At the heart of most MRI scanners is a large coil of superconducting wire. This coil is an electromagnet: applying a current through it creates a magnetic field in its centre. The stronger the current through the electromagnet, the stronger the magnetic field.

When a superconductor is cooled down to below its critical temperature, its electrical resistance drops to zero. As a result, superconductors can carry much higher currents than regular conductors (such as copper). So, in turn, superconducting electromagnets can efficiently produce the ultra-strong magnetic fields needed for MRI scanners to image tissue.

However, producing these currents requires a power supply for the electromagnet, which adds to the MRI’s overall size. With the aim of trying to shrink MRI scanners down, engineers are working on more compact types of superconductor that don’t need a power supply.

“MRI machines are massive and require specialist rooms,” says Dr Mark Ainslie, a materials engineer at King’s College London who specialises in applied superconductivity. With some technical advancement, he adds, it could be possible to create a much more compact MRI machine, potentially small enough for local health centres or GP offices.

From wire coils to solid lumps of material

Ainslie and his colleagues are working on a class of superconductors, called bulk superconductors, that could hold the key. Rather than being shaped into a coil of wire, bulk superconductors take the form of a solid lump of material. An example from Ainslie’s office cupboard resembles an ice hockey puck.

On its own, this particular ice hockey puck is “useless”, says Ainslie. That’s until you apply and remove a magnetic field. It will retain that magnetism and then act like a permanent magnet, just like the ones on your fridge. As long it’s kept cold, below its critical temperature, it will continue to produce this magnetic field.

The challenge for Ainslie and his colleagues over the past decade has been to produce bulk superconductors that can carry high enough, and stable enough, magnetic fields for practical applications such as MRI. A few years ago, Ainslie was part of a team that received a Guinness World Record for trapping the highest ever recorded field in a copper-based superconducting material. But despite its impressive performance, this type of superconductor is very difficult to make: it has a complicated structure and is made by a specialist technique. A small sample takes “at least a week” to grow, making manufacturing them at an industrial scale a tall order.

“There’s a big difference between a material being a superconductor – there are hundreds – and it being a practical engineering material,” he explains. Instead, the team turned to an iron-based superconductor, which can be made using a conventional industrial technique. That means it could potentially be manufactured on a large scale – and made more easily by more people.

A record-breaking magnetic field, thanks to machine learning

Ainslie’s Japanese colleagues built specialised computer software and used machine learning to optimise the processing conditions for the iron-based superconductor, finding this improved its superconducting properties. The team beat the previous record in an iron-based superconductor by almost three times, with a field strength and stability that meets the standards for medical MRI.

The team also dug into the material properties of the samples, to unpick why its performance was so impressive. To their surprise, they saw completely different microstructures for the machine-learning-optimised samples when compared to those where the processing conditions were entirely designed by researchers. In previous studies, researchers had typically observed that when the internal structures of samples were composed of fine grains of crystals, they were better superconductors. In this case, the machine-learning-optimised samples had a mix of large and small grains when viewed under an electron microscope.

“We probably would have looked at [this] normally and thought, that's not good,” explains Ainslie. “It was interesting that the machine learning didn't have any biases that researchers have.” He and his colleagues now hope to determine why this unusual microstructure results in such excellent superconductive properties.

A step towards smaller MRI scanners

With these results in hand, Ainslie and the team believe that producing high-performing bulk superconductors on a larger scale is a step closer for manufacturers.

However, he cautions that there are further challenges to address before more compact MRI scanners are a reality. For example, while the footprint of current MRI scanners may be huge, it can at least scan a person’s entire body. The design of a smaller-scale machine would likely look very different, and rigorous clinical testing would be needed to get it into the right form.

The immediate next step for Ainslie is to show that the magnetic field can be induced in the material without the very specialised equipment ordinarily needed to do so. This equipment is costly and accessible to only a handful of labs, so achieving this step without it will demonstrate to manufacturers that the material is scalable to produce.

But first, one of the field’s most important conferences is coming up in Japan. A critical mass of world-leading experts, including Ainslie and his colleagues, will be putting their heads together on ways that they can accelerate the real-world applications of iron-based superconductors.