A team at the USC Leonard Davis School of Gerontology has built an artificial intelligence model that maps aging region by region across the brain, rather than reducing a person's brain to a single "brain age" number. The work, led by Associate Professor Andrei Irimia and published in Proceedings of the National Academy of Sciences, also links those regional patterns to changes in cognitive function across the lifespan.
The model was trained on magnetic resonance imaging from close to 15,000 cognitively healthy people, which gave it a baseline for what any given brain structure typically looks like at a given chronological age. When the researchers applied it to scans from people with mild cognitive impairment and Alzheimer's disease, it picked out distinct patterns of accelerated aging in the regions already known to be hit early in neurodegeneration.
That granularity is the point. As the researchers note, brain regions do not age in lockstep - some appear resilient while others are more vulnerable to aging and disease. A single summary score can hide that variation, whereas a map showing which areas look older than expected offers a way to pinpoint where trouble is starting and how it tracks with cognitive change.
For clinicians and researchers, the appeal is a more informative readout from imaging that many patients already receive. The study demonstrates that the approach can distinguish patterns associated with mild cognitive impairment and Alzheimer's disease; how such maps might be used in individual patient care is a question for further research.