Two people can be the same age while their brains look surprisingly different on an MRI scan.
New research suggests that measuring this difference could help doctors estimate who is more likely to develop dementia years later.
Scientists in the Netherlands used computer programs to look at brain scans and estimate how old each person’s brain appeared. They compared that estimate with the person’s actual age to create a simple measure of whether the brain looked unusually old or young.
This idea is often called brain age. A brain that appears older than expected may show patterns of shrinkage or other structural changes that are too small or widespread for a doctor to easily judge by looking at the scan alone.
Doctors already use MRI scans when investigating memory problems. They often visually examine areas of the brain that commonly shrink in Alzheimer’s disease and other causes of dementia.
These visual rating systems are useful, but they have limits. Early brain changes may be slight, and assessments can depend partly on the experience of the person reading the scan.
Computer analysis offers another possibility. Instead of concentrating only on a few brain regions, a program can examine patterns across the scan and compare them with large amounts of information about normal aging.
The new study focused on 799 patients from the Alzheimer Center Amsterdam. They had participated in the Amsterdam Dementia Cohort or the SCIENCe project and had medical information collected between 2002 and 2024.
Just over half of the patients, 412 people, had subjective cognitive decline. They had noticed changes in their memory or thinking, but testing had not found clear cognitive impairment.
The remaining 387 patients had mild cognitive impairment. MCI is more noticeable on testing than subjective decline, but people with MCI can usually continue many everyday activities independently and do not necessarily develop dementia.
The researchers collected each patient’s MRI scan, results from a standard cognitive test and doctors’ visual ratings of brain shrinkage. They then used computer tools called brainageR and cNeuro to calculate how much older or younger each brain appeared than expected.
Over time, the team recorded which patients developed dementia. The results showed a clear relationship between an older-looking brain and a greater chance of progression.
Across all participants, every extra year of estimated brain age above actual age was linked to a 6% increase in the relative risk of dementia. After accounting for traditional MRI ratings, the association remained, at about 4% higher risk for every additional year.
The most interesting result appeared among people with subjective cognitive decline. For them, each extra year that the brain looked older was associated with roughly a 9% higher relative risk of progressing to dementia.
Brain age could also help identify people who appeared less likely to decline. When the brain of a person with subjective cognitive decline looked at least 2.6 years younger than their actual age, the probability of remaining dementia-free was between 91% and 99% across follow-up periods of two to 10 years.
This does not mean that being 2.6 years below the cutoff makes someone immune to dementia. The number came from this particular group of patients and must be tested independently before doctors can rely on it for individual decisions.
The computer measurement was less impressive among people with mild cognitive impairment. Once patients had reached this stage, standard visual MRI ratings already captured much of the useful structural information, so brain age added little.
The findings were published in 2026 in the journal Neurology. The research used data from the Amsterdam Dementia Cohort and SCIENCe project at Amsterdam UMC in the Netherlands.
One appealing feature of the approach is that it could use MRI scans that patients are already having. If the software can eventually be validated and fitted into routine care, doctors might gain extra information without requiring another invasive test.
Still, the study does not show that an older-looking brain causes dementia. Brain age is a statistical estimate, and the relationship could reflect many underlying processes, including normal aging, Alzheimer’s disease, blood vessel damage or other health conditions.
The study also involved patients from a specialist memory clinic, so the results may not apply in exactly the same way to the general population. Computer estimates can also vary depending on the MRI scanner, image quality and method used to calculate brain age.
The most promising use may be among people who sense that their memory is changing before standard tests show clear impairment. In this early group, automated brain-age analysis may reveal subtle warning signs that conventional visual assessment misses.
Overall, the research suggests AI-based brain age could become one piece of the dementia-risk puzzle rather than a stand-alone test. More studies are needed, but the method may eventually help doctors decide who needs closer follow-up and who can be given more reassuring information.
Source: Amsterdam UMC.
