Cheap and mobile MRI: Maximum health gain for the euro

Imaging & Image Guided Medicine

In sub-Saharan Africa, 100,000 children annually suffer from hydrocephalus. After early diagnosis, this condition can be treated easily by placement of a shunt. However, providing a definitive diagnosis requires an MRI scanner, a device that is not always readily available. Uganda, for example, only has one MRI scanner, which is in a hospital in the capital. If the general population is to benefit from this technology, scanners need to be taken to the people.

TU Delft researchers Martin van Gijzen and Rob Remis, together with Andrew Webb, an LUMC radiologist, set out to develop a cheap, mobile MRI solution based on permanent magnets. This initiative won them an STW OpenMind award, which allows them to apply for a proof-of-principle by the end of 2017.

A full-body MRI scanner is an advanced piece of equipment, using strong magnets and superconductivity enabled by liquid helium cooling. It creates images on the basis of hydrogen atom resonance, and costs about €2 million. The building, electricity supply and regular maintenance by highly skilled staff require another €2 million. Budgets like this mean that it is not feasible to have MRI scanners in large parts of Africa.

“It was Andrew’s idea to start developing a cheap, mobile MRI unit with a permanent magnet which wouldn’t require electricity or cooling,” says Martin van Gijzen, Associate Professor of Numerical Mathematics.

As Andrew Webb explains, “A regular MRI scanner can diagnose many different conditions in all kinds of patients. Here, we want to perform only one dedicated task: to establish whether hydrocephalus is present in small children or not. The permanent magnet is ten times weaker and the images it produces are less detailed, but that may not matter, as long as they are good enough to detect hydrocephalus. That’s what we’re aiming at with this technology.”

Low-end into high-end

At this stage, it is not yet entirely certain that this concept can be realised. “Nobody has proved it yet,” says Webb, “but we understand the principles involved. Thanks to a grant of €50,000 from STW, we hope we can have a proof-of-principle within a year from now.” Webb’s idea was inspired by Professor Steve Schiff from Penn State University, an electrical engineer and paediatrician, who has vast experience of working in Uganda. “Schiff knows about the technological as well as the medical side of the matter,” says Webb. “What’s more, he’s aware of what could work in practice in Uganda. In particular, the hardware should be cheap, compact, and mobile so that it can be used in field clinics, ambulances or even on the back of a pick-up truck. It must also be easy to assemble and easy to repair.”

The more low-end the hardware, the higher the claim on signal processing. This makes high-end signal processing key to the project. “Luckily the imaging task is relatively simple,” says Van Gijzen. “It’s relatively easy to distinguish water from brain matter. Nonetheless, data acquisition becomes entirely different when you work with a permanent magnet.”

Combining approaches

The major obstacle is that the signal from a permanent magnet is far less homogeneous than from a conventional MRI. “We’re trying to turn this drawback into an advantage and to make use of the variation in the field to enhance the imaging,” says Van Gijzen. “With an inhomogeneous field, you can try to find out where the signals are coming from specifically, because the field is inhomogeneous. When you rotate the magnet in patterns and the information is put together, you may receive the spatial information you need for a good enough image.”

The options for a solution are there. “The mathematical tools we have at our disposal nowadays are far more advanced than the Fourier transformation which is the standard for modern MRIs,” says Van Gijzen. “As we’re working with a low-resolution device, the quantity of data will be much less,” says Rob Remis, Associate Professor of Circuits and Systems. “This allows for the application of more sophisticated approaches used in non-Fourier imaging in general. Think, for instance, of imaging algorithms used for CT scans, in astronomy or in geophysical imaging.” Masters’ students are already working on combining multiple low-res approaches to get high-res results, which is the direction the team is aiming for.

High ambitions

The research team is currently making use of a test product with a 4cm-diameter magnet. However, a proof-of-principle prototype will consist of a 15cm- to 20cm-diameter magnet. This object will weigh thirty to fifty kilos excluding the processing equipment. Costs for the complete system should be between €10,000 and €50,000 – preferably at the lower end. The layout of the magnet is not yet definitive. “You can design the magnet in a Lego-like fashion by putting together many small magnets,” says Van Gijzen.

Without ambition no one ever gets anywhere

The design of the magnet, its hardware, data acquisition and processing strategy are scheduled to be finished by the end of 2017. Ideally, the result will not only be a proof-of-principle, but also a working prototype. The next step would be to evaluate and improve that product. Van Gijzen, Remis and Webb know that this schedule is highly ambitious. But they also know that without ambition no one ever gets anywhere.

Interview by: Leendert van der Ent