Artificial intelligence could promote the democratization of MRI machines – 05/09/2024 at 8:12 p.m.

Artificial intelligence could promote the democratization of MRI machines – 05/09/2024 at 8:12 p.m.
Artificial intelligence could promote the democratization of MRI machines – 05/09/2024 at 8:12 p.m.

A man undergoes magnetic resonance imaging (MRI) near Paris, March 22, 2044 (AFP / ALAIN JOCARD)

A magnetic resonance imaging (MRI) machine, built with simple commercially available parts combined with artificial intelligence, matched the performance of high-end MRI machines, according to a study released Thursday that could promote better access to these expensive, life-saving tools.

MRI is a medical imaging technique that provides precise two- or three-dimensional views of the inside of the body, non-invasively. MRI plays a crucial role in locating tumors and treating pathologies of the central nervous system (brain and spinal cord), muscles or heart.

But these sophisticated devices remain rare in developing countries: Africa has less than one MRI machine per million inhabitants, compared to 40 per million inhabitants in the United States and 55 in Japan. The cost of a latest generation MRI scanner exceeds a million dollars.

To tackle this problem, a team from the University of Hong Kong led by researcher Yujiao Zhao managed to build a simplified, low-power MRI machine using commercially available hardware, at a cost of approximately $22,000. They published their results in the prestigious journal Science.

A large consumer of electrical energy, a traditional MRI machine uses a powerful magnetic field produced by superconducting magnets. Liquid helium, a rare and expensive substance, is used to cool the coil of superconducting magnets.

However, the machine created by the team of Hong Kong researchers uses a simple magnet, does not need helium and operates with a power of 1,800 watts, comparable to that of a hair dryer.

To compensate for the loss of contrast in the images obtained, Yujiao Zhao and his team integrated into their system an artificial intelligence (AI) algorithm exploiting a large dataset of high-resolution images of human anatomical structures.

They then tested their machine on 30 healthy adult volunteers, taking images of their bodies, from the brain to the knees. The results of this so-called “ultra-low field” MRI were found to be comparable to those of conventional MRI scanners, 60 times more powerful.

These advances offer hope “of being able to meet the unmet needs of health centers around the world,” commented Ms. Zhao.

-

-

PREV OFPPT and Lafarge Holcim Maroc sign a framework agreement on training in the construction sector
NEXT The Casino group, in the grip of serious financial difficulties, sold 121 stores to Auchan, Les Mousquetaires and Carrefour