A research team from Facebook AI and NYU Langone Health shared interesting research showing how AI could change the dynamics of MRIs and the patient experience.
Based on the research data they share, AI can dramatically reduce the time to MRIs – a process that would run four times faster.
As the document mentions, this is due to a new approach and the way they use AI to modify the scanning process:
Unlike most medical imaging AI projects, which attempt to use AI to automatically review images for abnormalities, fastMRI is using AI to create images in a new way that requires much less data.
Applying this model, the images created use 4 times less data than the traditional scanning machine process, which can take up to an hour to gather all the raw data for MRI. Less data, less time in the process.
To test this dynamic, a team of radiologists reviewed AI-generated images and traditional MRIs, and couldn’t tell the difference. The images created by AI do not lose quality or stop being accurate, compared to the traditional method, despite using only 25% of data.
This new method opens up a range of possibilities. Not only does it improve the patient experience, since it will take only a few minutes to spend in an uncomfortable process, but it could also replace other types of study in urgent cases.