The human brain ages like the rest of the body—but not evenly. Some regions wear out faster than others, and a new artificial intelligence tool from the University of Southern California can now show exactly where. Researchers have built an AI that creates detailed maps of brain aging, revealing which parts of the brain look older or younger than they should for a person's actual age.
Led by Associate Professor Andrei Irimia at the USC Leonard Davis School of Gerontology, the team trained their AI using nearly 15,000 brain scans from cognitively healthy adults ranging from teenagers to people in their 100s. With this massive dataset, the model learned what a typical aging brain looks like. It can then compare any new brain scan against that standard and flag regions that seem to be aging faster than expected.
"Not all brain regions age at the same rate," Irimia said. "Some areas appear to be more resilient, while others are more vulnerable to aging and disease."
When the researchers applied their tool to people with mild cognitive impairment or Alzheimer's disease, the differences were striking. These participants showed substantially more widespread patterns of advanced local brain aging, especially in the frontal and temporal lobes—regions involved in decision-making, memory, and other higher-level thinking skills. These are also among the first areas affected by Alzheimer's pathology.
The study also uncovered that the right hemisphere of the brain tends to show slightly more advanced aging than the left, regardless of whether a person is right- or left-handed. Meanwhile, the parietal and occipital lobes, which handle spatial awareness and sensory processing, appeared more resilient.
Traditional brain aging research reduces the entire brain to a single number. Irimia's approach is different—it measures aging at the level of tiny 3D units called voxels, the same units that make up an MRI scan. This gives doctors and researchers a much richer picture.
"This more nuanced understanding of how the brain ages could pave the way for earlier identification of dementia, a better understanding of what factors affect risk and new ideas for treatment approaches," Irimia said.
The research, published in the journal Proceedings of the National Academy of Sciences, drew on six large public datasets including the UK Biobank and the Human Connectome Project. With further development, the tool could one day help doctors spot early signs of neurodegeneration years before symptoms appear—giving patients and families precious time to plan and researchers new clues about what drives these devastating diseases.