hotspot

How many scientists does it take to build a working MRI from scratch?

New York University researchers Leeor Alon and Tobias Block wanted to find out. The two hosted 50 researchers in New York City last month, and over the course of four days, managed to build a small, low-field MRI. With a few extra days, Alon and Block successfully used the scanner to image an item inside a mystery box: two water-filled syringes. The overarching message of the gathering: MRIs don’t need to be as expensive as they currently are.

advertisement

MRI machines, which typically cost between $1 million and $3 million, are frequently the most expensive pieces of equipment in a hospital. They’re large, heavy, and complex, built with three layers of magnets to image and liquid helium to cool the machines down. Engineers need to maintain the machines and doctors need to interpret their scans. Patients in sub-saharan African countries have particularly difficulty accessing MRIs; the continent has the lowest number of MRI scanners per million people at 0.7, followed by Southeast Asia at 1.1.

“The kit itself is so complicated that there’s so many scientists, electrical engineers, MRI researchers, and clinicians [involved in building and researching the scanner],” said Sola Adeleke, an oncology fellow at Guy’s Hospital in London and an expert in advancing MRI access in Africa. “All of those things are not in enough supply in low and middle income countries.”

Sign up for STAT Health Tech

Tech is transforming health care and life sciences. Our original reporting is here to keep you ahead of the curve.

MRI access is a problem in the United States as well, with scans costing up to $2,850 out of pocket. Patients in rural areas with smaller medical centers have fewer MRI options, which also raises the cost.

advertisement

Three major manufacturers — Siemens, GE, and Philips — have dominated the MRI market for decades. The lack of competition keeps MRI costs high and stymies innovation, Block said, as the companies have little incentive to change their machine design. Block and Alon have obtained scanners from the big three for research projects, but there are often limits to the types of research they want to pursue. Block, for example, would like to automate image capture rather than rely on highly trained staff. But that would require the vendors to open up their source code, which is messy and intricate.

“We have a lot of MRI scanners, but it’s always restrictive,” Block said. “Sometimes you want to try something and they are very hesitant to do that. For them, it’s a lot of investment to modify the systems.”

Advancements in machine learning algorithms and 3D printing may be able to shake up the field, Block and Alon said, bringing MRI costs down and energizing researchers on a budget.

“If we have a low-cost, open source scanner that people can reproduce and build, that can down the road create a lot of new, exciting developments and technology,” Block said.

One way to drive down cost is by producing scanners with lower magnetic fields. Researchers previously pushed for higher-field machines in order to extract crisper images. Experts now argue that low-field images may be just as clinically meaningful, especially with the help of smarter AI.

“You want to treat the patient and you don’t need a pretty picture,” Block said. “If the patient has a stroke and you see that on a cheap scanner, that’s totally fine.”

Related: NYU launched private ChatGPT for its health data, and set its staff loose to experiment

Ruth Carlos, a radiologist at the University of Michigan, agreed that MRI is ripe for disruption from a small, fast-moving team. She pointed to the push for portable CT scanners, calling for a similar transformation in MRI.

“Redesigning an MRI that would have the same capabilities, but are small enough, fast enough, cheap enough that you can put them essentially on the back of a flatbed truck or even in a Winnebago and drive it out to the community,” Carlos said. “That would open up access incredibly.”

At the hackathon last month, some participants stayed until four o’clock in the morning to meet the ambitious timeline of building an MRI in four days. Anja Samardzija, a biomedical Ph.D. student at Yale, worked on the gradient coils that create variation in the MRI’s magnetic fields. She said the hackathon spurred creativity, making the tedious task of gluing metal coils more enjoyable.

The team successfully built a prototype big enough to image a hand, calling it Zeugmatron Z1 after “zeugmatography,” the first name for MRI coined by Paul Lauterbur.

“We integrated all the components, but we had some issue with a cable or a connector that we were not able to troubleshoot,” Alon said. “We were acquiring signals, but much weaker signals than we expected.”

After troubleshooting in the days after the hackathon, the two successfully made the Zeugmatron create its first image. The group recently submitted several abstracts to the leading magnetic resonance conference. The next step is to open the scanner up to the research community. After that, they’ll pursue approval from an institutional review board to start testing it on people. They think they could manufacture this type of machine for under $15,000.

Opening up MRI access for researchers could have a downstream benefit for patients, making cheap and effective MRIs more broadly available across the world. Adeleke emphasized that future hackathons should involve more researchers and health care workers from other countries.

“There can be some skills transference and we can learn from each other,” Adeleke said. “We can learn about local health care challenges, how the health system works in different regions.”

Correction: This article has been updated to clarify that Alon and Block are researchers, not radiologists, and that the scanner successfully created its first image after the hackathon. The hackathon took place over the course of four days, not five.