Impact of covariates on brain imaging biomarkers
Approved Research ID: 97117
Approval date: December 14th 2022
When a brain disease is suspected, a doctor must evaluate whether imaging findings are due to any disease, e.g., early Alzheimer's disease, or could be explained just by normal variability. Various factors, such as age, sex, ethnicity, or even the time of day when the imaging was done have impact on the anatomy of the brain but are not directly related to any disease. The doctor must consider such confounding factors in the interpretation.
We will model the impact of various confounding factors on imaging biomarker values from brain MRI images and propose necessary methods to remove their impact enabling easier and more accurate interpretation in clinical practice. These improved models can then be used to differentiate normal findings in an MRI scan from findings that require further investigation. We will perform the corresponding analysis also for the rate of change between images from two time points. This allows the detection of unusually fast changes, which can be indicative of something other than normal aging.
The results of the project will increase the scientific and clinical knowledge of the impact of confounding factors on the brain. In addition, it helps to develop better artificial intelligence models and tools for clinical practice helping doctors to diagnose and treat their patients better.
The expected duration of the project is 2 years.