Normative Modeling of Brain Morphometry Across the Lifespan Using CentileBrain: Algorithm Benchmarking and Model...

Area of Science:

• Neuroimaging
• Brain Morphometry
• Developmental Neuroscience

Background:

• Understanding typical age-related brain changes is crucial for identifying neurodevelopmental and neurodegenerative disorders.
• Existing normative models often lack sex-specific considerations, potentially obscuring important biological differences.

Purpose of the Study:

• To establish an empirically validated, sex-specific normative modeling framework for brain morphometry.
• To assess deviations from typical age-related neuroanatomical trajectories and inform future study designs.

Main Methods:

• Comparative evaluation of eight algorithms using regional morphometric data from 37,407 healthy individuals (ages 3-90).
• Optimization involved nonlinear age polynomials and linear global measures as covariates.
• Multivariate Factorial Polynomial Regression (MFPR) was selected as the preferred algorithm.

Main Results:

• MFPR models demonstrated high accuracy across the lifespan and within age-specific bins.
• Models exhibited longitudinal stability over a 2-year period.
• Optimal performance was achieved with sample sizes exceeding 3,000 participants.

Conclusions:

• The developed framework provides reliable sex-specific normative data for brain morphometry.
• This tool can enhance the interpretation of neuroimaging findings and guide future research.
• The MFPR model and associated scripts are publicly available via CentileBrain.