NYU and Meta Researchers Find AI Makes MRI Scans 4 Times Faster

Researchers at NYU Langon Radiology have created an artificial intelligence that creates high-quality images in collaboration with Meta AI using only a quarter of the data needed for previous scans.

Artificial intelligence improves the speed and clarity of MRIs compared to traditional scans, according to a recent study by the Department of Radiology at NYU Langone Health and the AI ​​Research Group at Meta. The study, published Jan. 17, demonstrates the power of a machine learning algorithm that can create high-quality MRI scans using as little as one-fourth of the data normally needed, reducing the time it takes to scan to five minutes.

An MRI provides more detailed information than other types of medical scans, such as CT scans and X-rays. They also use non-invasive magnetic fields instead potentially harmful radiation to create images, but it usually takes much longer and expensive. Faster MRIs can reduce the cost of scanning in terms of time and money, making them more affordable for patients.

In addition to facilitating access to the technology, AI assistance could make scanning more convenient for patients, said Patricia Johnson, associate professor at Langone, co-author of the study.

“It will improve the experience,” Johnson said. “You are there for five minutes instead of 20 – you are much less likely to move. That’s why images look better too. They are less likely to get distorted.”

The artificial intelligence model is part of fastMRI, a Langone initiative created in 2018 in partnership with Meta — formerly Facebook — to increase the speed of MRI scanning. The FastMRI artificial intelligence model used a dataset of 298 MRI scans to recognize patterns in the scans, allowing the unscanned parts of the 3D image to be recreated.

Meta invests in AI technologies to support the development of the Metaverse — a virtual world on the Meta social networking platform. In a press release published alongside Meta’s 2021-2022 earnings report, founder and CEO Mark Zuckerberg said that Meta’s investment in artificial intelligence gave positive results.

Meta AI researchers have been working with Langone for two years. do research — who have focused on knee scanning — and plan to continue their research on how AI technology can improve MRI for other parts of the body, including the brain.

During the study, both conventional and artificial MRI scans of the knees of 170 study participants were performed. It was found that AI-assisted scanning was just as helpful as conventional scanning in making a diagnosis and also resulted in higher quality images.

“Our new study translates the results of an earlier laboratory study and applies them to real patients,” said Michael Recht, director of radiology at New York University’s Grossman School of Medicine. said in a statement. “FastMRI has the potential to revolutionize the way we do MRI and increase the availability of MRI to more patients.”

In addition to developing its own algorithms, fastMRI provides anonymous data to other researchers who may not have the same access to MRI scanners that can cost over $500,000. By publishing their data, the program’s researchers hope to stimulate competition and innovation in the field.

“We are so unique that NYU Medical Center has about 50 MRI scanners, which is more than in some countries,” Johnson said. “We just have a unique opportunity to collect large amounts of patient data. We want to promote research in this area.”

Contact Annabelle Wang at [email protected]