Facebook’s AI researchers are working with New York University to make magnetic resonance imaging (MRI) up to 10 times faster.
The goal: Through a project called fastMRI, Facebook’s team will develop algorithms capable of filling in the gaps in low-resolution MRI scan, effectively turning them into higher-res ones. This could be important because currently it can take an hour or more to produce a good MRI scan. That is a long time to lie motionless inside an MRI machine, especially for children or patients who are already unwell. The hope is to cut the time down to a few minutes.
How’s it work? MRI scans will be enhanced using what’s known as a generative model. Neural networks will be trained (using anonymized data) to fill in missing or degraded part of MRI scans. Larry Zitnick, the lead Facebook researcher on the project, says the key challenge will be making sure that nothing vital is left out of the touched-up pics.
Why tho? Fear not. Facebook assures us this isn’t a scheme to get users to post their MRI scans along with vacations pics. Rather, the hope is to advance an important machine learning technique, and to help attract AI researchers keen to work on meaningful stuff.
Machine medicine: Indeed, AI is poised to have a big impact on medicine and health care, but some serious challenges remain. It will be vital to make sure patient data is protected, to account for algorithmic bias, and to find ways to explain algorithmic reasoning.
Why Meta’s latest large language model survived only three days online
Galactica was supposed to help scientists. Instead, it mindlessly spat out biased and incorrect nonsense.
DeepMind’s game-playing AI has beaten a 50-year-old record in computer science
The new version of AlphaZero discovered a faster way to do matrix multiplication, a core problem in computing that affects thousands of everyday computer tasks.
A bot that watched 70,000 hours of Minecraft could unlock AI’s next big thing
Online videos are a vast and untapped source of training data—and OpenAI says it has a new way to use it.
Google’s new AI can hear a snippet of song—and then keep on playing
The technique, called AudioLM, generates naturalistic sounds without the need for human annotation.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.