Facebook and NYU School of Medicine’s Department of Radiology recently announced on the Facebook blog that their AI research team has collaborated to create an AI system that can speed up MRI machines. The research project will aid in investigating the use of artificial intelligence (AI) to make magnetic resonance imaging (MRI) scans up to 10 times faster.
MRI scans are excellent tools for doctors, as they provide doctors with images that typically show a higher level of detail related to soft tissues — such as organs and blood vessels. They allow a non- invasive look inside a human patient’s body. But they take a long time, taking anywhere from 15 minutes to over an hour, compared with other imaging devices such as X-rays and Ct scans, which takes less than a second or up to a minute, respectively. These long scan times can make MRI machines challenging for young children and difficult for people who are in pain and are asked to keep still for the entire scan.
Also, many patients dislike MRI scans because it involves lying in an enclosed tube with a loud noise. Additionally, rural areas have faced MRI shortages and in other countries with limited access, causing long scheduling backlogs. The researchers have used artificial intelligence to boost the speed of MRI scanners and make it accessible to a more significant number of patients.
The new system, called fastMRI, lessens the number of time patients must hold their breath during imaging of the liver, heart, or other organs in the abdomen and torso. Enhanced speed could let MRI machines fill the role of X-ray and CT machines for some applications, enabling patients to avoid the ionizing radiation associated with those scans.
Trained radiologists compared knee films from traditional MRI scans with those that were sped up using the fastMRI technology, and radiologists could not differentiate the regular images from those that had been created using the new technology. The research team describe the output as interchangeable. The research team also says that new technology-enabled creating MRI films in 75% less time, in about 15 minutes.
The machine which was in the testing phase for two years is now ready for other groups to test it in preparation for commercial use. And since it works with existing MRI machines and is an open-source, the team believes that radiologists could be using it within two years.