A new simulation- and AI-based approach reduces the time and data needed for brain MRI scans

Researchers from the Institute for Neurosciences (IN) have developed a method using artificial intelligence and simulations to obtain detailed brain information from MRI scans more quickly with far less data.
The new technique reduces scan time for advanced MRIs by up to 90% while preserving accuracy, making neuroimaging more efficient and accessible.
Instead of training models with real patient data, the method uses simulations based on the physics of the diffusion process in brain tissue, which reduces dependence on patient availability and privacy concerns.
The approach can drastically lower the number of measurements required, achieving a high level of accuracy using only 10% of the usual data quantity.
This breakthrough may translate to a significant reduction in scan time, potentially reducing 40-minute scans to around 8 minutes.
The method also has potential applications in early diagnosis for neurodegenerative diseases like Alzheimer’s and reanalyzing decades-old MRI data for new insights.

You may have missed