Facebook, NYU join together to speed up MRIs

Facebook and the NYU School of Medicine are joining together to look into speeding up MRI scans through artificial intelligence.

In a new project called fastMRI, the social media giant and medical school hope to broaden access to MRIs and advance AI. They will present their work as open-source so other researchers can use it. Some patients are not able to stay still for the duration of an MRI, and some areas lack sufficient MRI machines, which hinders access.

Sped-up MRIs could also take the place of X-rays and CT scans in some cases, reducing patients' radiation exposure.

Right now, an MRI takes between 15 minutes and 60 minutes. Researchers from the NYU Radiology Department's Center for Advanced Imaging and Research and from Facebook's Artificial Intelligence Research group think they can change that by applying AI. If they can reduce the amount of data an MRI machine must capture—filling in the missing details with data generated by artificial neural networks—they could shorten the time an MRI requires.

Facebook's Artificial Intelligence Research group chose to work with the Center for Advanced Imaging and Research because it was looking for "projects in which AI could have significant real-world impact," and the center was already working on using AI to speed up MRIs, according to a news release. The collaboration will allow Facebook's group to apply what it knows about deep learning and computer vision to NYU's knowledge of imaging science.

"Our collaboration is one between academia and industry in which we can leverage our complementary strengths to achieve a real-world result," said Dr. Michael Recht, chair of the radiology department at NYU.

NYU isn't yet ready to share a project timeline with the public, Recht said. But eventually, fastMRI could lead to other advances in medical imaging, including improvements to CT scans.

But whether the project will actually improve outcomes is unclear, said Paul Clark, director of healthcare research for Digital Reasoning. "This is a pure technology in search of a solution play," he said. "What would likely happen is an increased number of MRIs at the same cost per MRI, leading to greater revenue for MRI companies and hospitals and greater expenses for insurance companies and patients."

The NYU collaboration is one of Facebook's first forays into healthcare, according to Katelijn Vleugels, CEO of Klue, which makes behavior-tracking software.

"Facebook has a strong AI capability with image processing, so it does make sense," she said.

Other industry leaders expected Facebook to get into healthcare first through advertising and data analytics. The company had even asked some U.S. hospitals for anonymized patient data for a research project. It put the project on hold after the Cambridge Analytica data leak was made public.

The NYU collaboration won't involve any Facebook data.

Since then, Facebook has been relatively quiet in healthcare, while other tech giants—Alphabet and Apple—have continued making inroads into the industry, with work by both to use application programming interfaces to move data around.

Google also recently joined with Amazon, Microsoft, IBM, Oracle and Salesforce to push for greater interoperability.

NYU made waves in the medical community last week when it announced it would give all current and future students full tuition scholarships to attend medical school.