Multimodal MRI prediction of cognitive functioning across the lifespan: separating between-person differences from...

Area of Science:

• Neuroimaging
• Cognitive Neuroscience
• Gerontology

Background:

• Brain Magnetic Resonance Imaging (MRI) holds potential for predicting cognitive functioning.
• Its clinical utility hinges on accurately capturing stable individual differences for patient stratification and within-individual changes for monitoring prognosis and treatment.
• Understanding the distinct contributions of various MRI modalities is crucial for optimizing their application in aging research.

Purpose of the Study:

• To benchmark the predictive utility of five MRI modalities—task fMRI, functional connectivity (FC), structural MRI (sMRI), diffusion-weighted imaging (DWI), and arterial spin labeling (ASL)—for cognitive functioning.
• To determine the capacity of these MRI markers to explain between-person variance (patient stratification) and within-person changes (prognosis, treatment monitoring) in cognitive aging.
• To investigate the overlap between MRI-derived variance and age-related cognitive variance.

Main Methods:

• Utilized longitudinal data from 450 adults (aged 21-90) from the Dallas Lifespan Brain Study, with up to three waves of data collected five years apart.
• Benchmarked five MRI modalities (task fMRI, FC, sMRI, DWI, ASL) and their combination across 37 cognitive phenotypes.
• Employed variance decomposition and commonality analysis to assess the contributions of MRI markers to cognitive functioning variance.

Main Results:

• A combined MRI marker achieved the highest prediction accuracy for cognitive functioning (R² = .51), outperforming individual modalities, with DWI and FC also showing strong performance.
• MRI markers explained substantial between-person variance in cognition (up to 60.3%) but only modest within-person changes (up to 17.2%).
• Most MRI markers, excluding ASL, demonstrated significant overlap with age-related variance in cognitive functioning.

Conclusions:

• Integrated MRI markers offer high accuracy for stratifying individuals based on cognitive functioning.
• The current MRI markers are more effective for identifying stable between-person differences than for monitoring longitudinal within-person cognitive changes.
• Findings clarify the strengths and limitations of various MRI techniques in the context of cognitive aging research and clinical applications.