AI-enabled measurements of 'local brain aging' offer detailed insights on dementia and more

USC researchers have created an AI-driven model that generates detailed maps illustrating how various parts of the brain age differently. Published in the Proceedings of the National Academy of Sciences, this new approach, led by Associate Professor Andrei Irimia of the USC Leonard Davis School of Gerontology, leverages magnetic resonance imaging data from nearly 15,000 cognitively healthy individuals.

By comparing brain structure to patterns observed in healthy individuals across the lifespan, this study goes beyond traditional methods that typically offer a single brain age estimate. Instead, it measures local brain age at the voxel level, yielding a richer, more nuanced picture of structural aging throughout the brain.

The model was trained using MRI scans from 14,748 cognitively normal adults aged 19–100, sourced from six major public datasets. It was then tested on MRI scans from over 1,900 additional participants, including those with mild cognitive impairment and Alzheimer's disease. The findings reveal distinct patterns of accelerated aging in brain regions known to be early targets of neurodegeneration.

Regions like the frontal and temporal lobes, crucial for decision-making and memory, appeared biologically older than areas like the parietal and occipital lobes, involved in spatial awareness and sensory processing. Additionally, the right hemisphere generally exhibited more advanced aging than the left, a pattern consistent regardless of handedness.

In individuals with mild cognitive impairment or Alzheimer's disease, specific brain structures such as the hippocampus, amygdala, and several deep brain regions associated with memory and cognitive processing displayed significantly older local brain ages. Notably, higher local brain age was linked to poorer cognitive performance, particularly in Alzheimer's patients, indicating that regional brain aging could become more informative as neurodegeneration progresses.

The detailed anatomical maps produced by this model could enhance our understanding of why some individuals experience faster cognitive decline in specific areas. However, the researchers emphasize that this tool is currently a research instrument, requiring further validation with more diverse clinical datasets before potential clinical application.

Despite this, the study highlights a significant advancement in neuroscience by moving beyond a single measure of brain age and toward a more precise understanding of both healthy aging and neurodegenerative diseases.