Facebook announces that it was possible to reduce the scan time to 1/4 with ``fast MRI'' technology that accelerates MRI scan with AI
Facebook's AI research department Facebok AI Research (FAIR) and New York University Langone Medical Center use AI technology for ``
FastMRI
https://fastmri.org/
FastMRI breakthrough shows AI-accelerated MRIs interchangeable with traditional MRIs
https://ai.facebook.com/blog/fastmri-breakthrough-shows-ai-accelerated-mris-interchangeable-with-slow-traditional-mris
Facebook and NYU use artificial intelligence to make MRI scans four times faster-The Verge
https://www.theverge.com/2020/8/18/21373335/faster-mri-scans-ai-machine-learning-facebook-nyu-research-clinical-study
In order to have an MRI, the patient needs to sleep for a long time in a large device. However, MRI noises and squeaking noises in a quiet room not only arouse the patient's anxiety, but also cause great stress for those with claustrophobia. Research by FAIR and the New York University Langone Medical Center is expected to solve the problems of MRI by speeding up the work of MRI.
``Fast MRI'' that makes MRI scan 10 times faster reveals a method to dramatically improve the quality of MRI images-GIGAZINE
The fastMRI research team trained machine learning models using low-resolution and high-resolution MRI images. This model can predict what kind of MRI image will be output from the initial stage of MRI scan with low reading accuracy.
The fastMRI research team is already conducting experiments at the clinical level, and when the radiologist made a diagnosis with the predicted image by fastMRI and the actual MRI image, he reported the same diagnostic result.
The images below are images of a young patient with injured knee taken by MRI as usual. Large femur and tibia shows a fracture of.
And the following images are MRI images output at 4 times the conventional speed using fast MRI. The same image as the conventional MRI is output. As research progresses, MRI scans can be done up to 10 times faster, the fastMRI research team said.
Although fastMRI is still a work in progress, training data and models are fully open-accessible and can be implemented without the need to physically modify existing MRI scanners, which we expect to be operational in the next few years. Dan Sodickson, a professor of radiology at the New York University Langone Medical Center, has revealed that he is already in talks with the makers of MRI scanners.
'Fast MRI is a key step in incorporating AI technology into medical imaging,' said Nafisa Yakubova, a researcher at FAIR. 'After long discussions between radiologists and AI engineers, the realization of fast MRI was realized. It has become possible. Complementary expertise is the key to creating such a solution.'
Related Posts:
