Brain age prediction: Cortical and subcortical shape covariation in the developing human brain

Area of Science:

• Neuroscience
• Developmental Neuroscience
• Brain Imaging

Background:

• Cortical development involves complex maturational changes across various brain shape measures.
• Summarizing these changes into a single index, like brain age, is of growing research interest.
• Understanding covariation patterns is key to characterizing individual neurodevelopment.

Purpose of the Study:

• To quantify covariation patterns among multiple cortical and subcortical brain measures.
• To assess the accuracy of these patterns in predicting chronological age, sex, and cognitive ability.
• To validate the reliability and sensitivity of prediction models using independent and longitudinal datasets.

Main Methods:

• Utilized the Joint and Individual Variation Explained (JIVE) method on neuroimaging data from the Healthy Brain Network (HBN) cohort (N=869, ages 5-18).
• Validated findings in the Nathan Kline Institute - Rockland Sample (NKI-RS; N=210).
• Employed ridge regression for age and sex prediction, and assessed intelligence quotient (IQ) correlations.

Main Results:

• Identified significant covariation across cortical thickness, gray matter volume, surface area, curvature, travel depth, white matter, and subcortical volumes.
• Covariation patterns accurately predicted chronological age (r=0.84) and sex (AUC=0.85) in cross-validation.
• These patterns explained 10% of the variance in full-scale IQ (R²=0.10) and demonstrated high reliability in test-retest and longitudinal analyses.

Conclusions:

• Significant covariation exists across diverse brain shape measures and subcortical volumes.
• Distinct covariation patterns, unique to each measure, contribute to predicting age, sex, and cognitive abilities.
• The developed prediction models exhibit high reliability and sensitivity to longitudinal changes in brain development.