MRI Unveils Enhanced Capabilities by Eliminating Shielding and Utilizing Superconducting Magnets

A Cost-Effective, Compact Scanner Utilizing AI for Enhanced MRI Quality

Magnetic resonance imaging (MRI) has dramatically improved healthcare by enabling the acquisition of 3-D images without radiation. However, conventional MRI machines often consume upwards of 25 kilowatts, operating powerful magnets that generate fields up to 1.5 tesla. This substantial energy demand restricts the use of these scanners mainly to dedicated medical facilities and hospital departments.

A research group from the University of Hong Kong has produced a low-power, simplified full-body MRI device. Utilizing artificial intelligence, this innovative scanner operates with a compact 0.05 T magnet and can function using a standard wall outlet, drawing just 1,800 watts during use. The team claims their AI-integrated machine can generate high-quality images rivaling those from today’s high-power MRI machines, potentially enhancing accessibility to MRI technologies globally.

The process of creating images with an MRI involves applying a magnetic field, aligning the body’s protons. The scanner then sends radio waves to disturb this alignment. Once the radio waves cease, the protons realign, emitting signals that the scanner translates into images.

Annually, over 150 million MRI scans are performed worldwide, as reported by the Organization for Economic Cooperation and Development. Despite ongoing advancements, two-thirds of the global population lacks access to clinical MRI procedures, particularly in low- and middle-income countries. For example, the U.S. averages 40 MRI scanners per million inhabitants, while West Africa had just 84 MRI units available for over 370 million people in 2016.

This gap largely arises from the costs and operational needs of traditional MRI systems. These machines require large superconducting magnets, specialized infrastructure, and rooms isolated from radio interference, all contributing to high expenses and limited portability.

Globally, scientists have looked into low-cost MRI alternatives operating at ultra-low-field (ULF) strengths below 0.1 T, which may consume less energy and allow for potential bedside applications. Initially, MRI development was centered around low-field strengths around 0.05 T, until General Electric introduced the first whole-body 1.5 T superconducting scanner in 1983.

Current ULF MRI systems often depend on AI for image reconstruction from the gathered signals but have previously been confined to imaging only single organs or brain structures. The Hong Kong team has now developed a comprehensive ULF MRI system, positioning patients between two permanent neodymium ferrite boron magnet plates. Although these magnets are less powerful than superconductive variants, they are cost-effective, easily sourced, and don’t require cooling.

Remarkably, the energy output from ULF MRI scanners is about a thousandth of that from conventional devices, producing significantly less heat during imaging. The new system is also quieter, which could be beneficial for pediatric patients.

The assembly consists of two units, each comparable in size to a hospital gurney. One unit houses the MRI machine, while the other supports the patient during scanning.

To mitigate radio interference both externally and from the scanner’s components, the researchers implemented 10 sensor coils within and surrounding the device to detect and address disruptive signals. Advanced deep learning methods were integrated for effective image reconstruction, eliminating the need for radio wave shielding and enhancing the scanner’s portability.

In trials involving 30 healthy participants, this new device successfully captured intricate images of the brain, spine, abdomen, heart, lungs, and limbs in under eight minutes, with resolutions around 2mm³. According to a review by Anazodo, the resulting image quality is comparable to that of traditional MRI machines.

Ed Wu, a professor at the University of Hong Kong, regards this work as the start of a multidisciplinary approach to develop a new category of straightforward, patient-focused diagnostic imaging devices powered by computing innovations.

The researchers utilized commercially available electronics, estimating a total hardware cost around US ,000. For context, entry-level MRI machines generally start at approximately 5,000, with advanced systems reaching 0,000 or more.

The prototype’s magnet assembly weighs roughly 1,300 kg, significantly lighter than clinical MRI machines that can weigh up to 17 tons. However, improving hardware could potentially reduce the weight of the assembly to about 600 kg, increasing the machine’s mobility.

While the new scanner won’t replace standard high-magnetic-field MRI technology—like the next-generation MRI systems using 7 T magnets, which provide extremely high resolution—it is designed to support existing MRI technologies in areas lacking high-tech facilities, like intensive care units and community clinics.

In discussions, Anazodo highlighted this Hong Kong advancement is part of a broader trend in developing ULF MRI scanners. For example, researchers at the University of Saskatchewan are working on an even lighter and more portable device aimed for applications in environments like the International Space Station.

The researchers published their findings on May 10 in the journal Science.

This article features as “Compact MRI Ditches Superconducting Magnets” in the July 2024 print issue.