Brain age prediction

Predicting brain age from MRI

The human brain changes structurally throughout life, but the pace and pattern of those changes differ from person to person. In collaboration with Prof. Tammy Riklin Raviv from the School of Electrical and Computer Engineering, the lab uses deep learning to estimate a person's age directly from structural brain scans. A gap between predicted brain age and actual chronological age can point to differences in how a brain is aging, whether that's healthy variation or an early sign of something else. In one study, the lab's model was trained on brain scans from over ten thousand people across a wide age range and predicted age with an average error of just a few years. Beyond making the prediction, the lab also developed a method to identify which brain regions the model relies on most, since deep learning models are often accurate but hard to interpret. That work pointed to fluid filled spaces in the brain, including the ventricles, as carrying a lot of the age related signal, alongside contributions from regions like the thalamus.

Why this matters for Alzheimer's research

Understanding typical brain aging matters because it helps separate what's normal from what isn't. Alzheimer's disease is the most common cause of dementia, and its early stages can be hard to distinguish from ordinary aging. A reliable way to estimate brain age, and to understand which brain features are driving that estimate, could eventually support earlier detection of atypical aging patterns and better targeted research into neurodegenerative disease.