GE Healthcare Reveals Pioneering Artificial Intelligence Tool for Faster MRI

GE Healthcare is promising to slash the time needed for an MRI using its new deep learning (DL) tool. Sonic DL, the name given to their state-of-the-art image acquisition accelerator, can reduce scan times by up to 83% and even capture the motion of a single heart contraction in real time.

The innovation uses artificial intelligence (AI) to reduce scan times and cater to cardiac patients with arrhythmias or breath-holding challenges. Currently, these patients are forced to sit through time-consuming scans that are prone to image degradation and can be challenging for those with severe cardiac conditions. With the new tool, patients no longer need to conduct repetitive breath holds, and images are less likely to be damaged by movement within the MRI machine.

“Sonic DL is a paradigm shift for MR enabling high-quality imaging in a single heartbeat,” said Jie Xue, President & CEO of Global MR at GE HealthCare. “It greatly expands patient access and improves diagnostic value for patients who need it the most but previously couldn’t be scanned successfully. Our industry-leading deep learning technology, AIR Recon DL, has benefited more than 10 million patients. We expect Sonic DL to further extend the lead of GE HealthCare in leveraging AI to advance healthcare and benefit patients worldwide.”

The AIR Recon DL technology that supports Sonic DL was designed on GE’s Edison AI software platform. Essentially, the platform uses an algorithm to improve image resolution without necessitating additional raw data. These capabilities provide a faster acquisition time, which is necessary for patients who cannot lie still on an MRI table.

GE has already received FDA clearance to begin marketing its Sonic DL platform, and in the future, they expect to use AI-supported MRI across additional areas of the body.

Medical imaging and AI have become intertwined in the past several years as researchers find new ways to improve radiologists’ ability to reach diagnostic conclusions. For example, Harvard Medical School and Stanford University announced last year that they had developed a self-learning AI tool that could detect pathologies in chest X-rays as good or better than human radiologists.

Additionally, companies like Google are working with healthcare providers to build extensive image libraries to serve as training sets for AI tools. Researchers expect they could even use AI to create new medical images based on previous scans without requiring additional time in an MRI machine or other medical imaging device.