In a recent publication appearing in the Proceedings of the National Academy of Sciences (PNAS) journal, researchers show that a new, AI-based technology can interpret patterns of structural change in brain scans, accurately estimating brain age associated with cognitive decline. Chronological age, measured in days, months and years, is the strongest predictor of Alzheimer’s disease (AD) risk. However, biological age can also be a strong contributing factor – but may be very different from our chronological age, depending on our lifestyle choices, and whether we are living with disease. In their recent article, a team of researchers led by Andrei Irimia of the University of Southern California (US) describe a convolutional neural network (CNN), AI-based tool that can calculate brain age estimates using magnetic resonance imaging (MRI) scans. This tool was developed using collated MRI images from over 4,600 cognitively normal participants, testing it on MRI scans from a further 1,170 participants cognitively normal participants.
The researchers then used the CNN to estimate the brain ages of 650 cognitively normal adults from the CamCAN study, 359 ADNI study participants with Alzheimer’s dementia, and 351 participants with mild cognitive impairment due to AD. To check whether the brain age estimates were linked to cognitive and functional changes experienced by people with cognitive impairment, they cross-referenced the brain age estimates to cognitive and functional test scores. These analyses revealed a lower correlation of test scores with chronological age, suggesting that the biological age estimates calculated by the new tool may better reflect brain function. The researchers were also able to detect different patterns in biological aging in people with and without cognitive impairment, pointing to a potential use of this tool in identifying people who may be at greater risk of AD and dementia. https://www.pnas.org/doi/abs/10.1073/pnas.2214634120
