UF Health researchers find new method to use MRIs to delve into secrets of brain aging

GAINESVILLE — Life experience continually shapes the health of the human brain, for better or worse, like a bank receiving a lifetime of deposits and withdrawals.

Have a poor diet? That’s a withdrawal. Enjoy dancing? Here’s a deposit slip. Scientists think hundreds of factors influence brain health, including education, depression, stress, disease, social engagement and exercise.

Those experiences biologically slow or speed up brain aging. The trick is reading that clock.

University of Florida Health aging researchers have demonstrated the feasibility of using common clinical brain MRIs to peer inside the organ and extract its biological age, or brain age. That knowledge might one day be used as an early signal of cognitive or physical decline, allowing earlier interventions.

Their study was published late last year in Scientific Reports – Nature.

Until now, these brain age estimates could only be obtained from higher-resolution and far more expensive research-grade brain scans rarely used clinically. Using lower-resolution clinical MRIs will allow a look at brain age in MRIs already being done to diagnose injury and disease.

To UF Health researchers, this is a step toward personalized medicine, when doctors can read the brain like a fingerprint and prescribe preventive therapies tailored to the individual.

“We’re coaxing the brain to give up its secrets by developing a new tool so that physicians can one day easily use artificial intelligence to understand the invisible toll or benefit our lives exact on the brain,” said the study’s lead author, Pedro A. Valdes-Hernandez, Ph.D., an assistant professor in the UF College of Dentistry and a neuroimaging expert.

On a team led by UF Institute on Aging researcher Yenisel Cruz-Almeida, Ph.D., Valdes-Hernandez used a publicly available AI model developed elsewhere that was trained on 14,000 brain MRIs to predict brain age. UF Health researchers then “retrained” the model using AI techniques by having the model analyze an additional 6,281 non-research-grade MRIs under multiple MRI configurations.

He said the AI then creates a high-resolution MRI from a clinical scan. The resulting brain age estimate is nearly as accurate.

“It’s a bit of alchemy,” said Valdes-Hernandez. “We’re bringing these artificial intelligence tools together in a new way. The beauty of this research is that much of this is based on existing knowledge.”

Cruz-Almeida said the research is a crucial step as it could eventually allow doctors to tap existing MRIs ordered, for example, to diagnose a brain injury after a concussion, to identify those at high risk of unrelated brain disorders or physical decline.

“We want to detect problems in the brain as early as possible,” said Cruz-Almeida, the study’s senior author and an associate professor in the UF College of Dentistry. “People are already getting MRIs anyway to rule things out. So, the money to get the MRI has already been spent. Why not use them to also provide early predictors of potential cognitive issues down the road?”

While dementia and Alzheimer’s are irreversible, Cruz-Almeida said, lifestyle changes like a better diet and exercise can possibly slow decline.

Each year of added brain age, she noted, is predictive of a 6% increase in the risk of death.

Brain age estimates are not yet used clinically to predict cognitive or physical decline. Researchers acknowledge that accuracy must be improved. But they envision a day when their model could be a diagnostic tool, like an EKG or blood test.

And that could be extremely useful. One day, Cruz-Almeida said, she hopes doctors will use an MRI to monitor brain health the way a general practitioner might order a blood test to check someone’s cholesterol.

The National Institute on Aging supported the study.