Multi-Cohort, Multi-Sequence Harmonisation For Cerebrovascular Brain Age
Understanding how our brains age is crucial, especially in relation to cognitive health. Researchers often calculate a “brain age” — how old a person’s brain appears based on scans, compared to their actual age. A difference here can sometimes indicate a higher risk of cognitive decline. Adding information about blood flow in the brain, obtained through a specialized MRI technique called Arterial Spin Labelling (ASL), can make these brain age predictions even better and more sensitive to our cardiovascular health, a key factor in brain aging.
However, a challenge arises when combining data from multiple research studies, as different MRI scanners and settings can introduce inconsistencies in the ASL measurements. To tackle this, a recent study investigated various techniques to make these brain imaging data consistent, a process known as harmonization.
The researchers analyzed a large collection of brain scans from over 2600 individuals, including standard structural images and those specifically measuring brain blood flow. They applied several harmonization methods to features extracted from these scans, such as the volume of different brain tissues and the rate of blood flow.
The findings showed that these harmonization techniques successfully reduced unwanted variations in ASL features between different study groups. Crucially, these methods led to more accurate brain age estimations, especially when predictions relied solely on the blood flow data. An interesting discovery was that methods specifically designed for ASL data performed best, highlighting the importance of considering the unique characteristics of how this type of data is collected.
This work is a significant step forward, demonstrating that by standardizing brain imaging data, particularly for blood flow, we can gain more reliable and consistent insights into the complex relationship between cardiovascular health and brain aging across diverse populations.
Source: link to paper