A unified framework for association and prediction from vertex-wise grey-matter structure

Area of Science:

• Neuroimaging
• Human Genetics
• Quantitative Psychology

Background:

• Large-scale neuroimaging cohorts enable detailed studies of human variation.
• Understanding the link between brain architecture and phenotypes is crucial.

Purpose of the Study:

• Investigate the association between phenotypes and brain morphology (morphometricity).
• Explore bivariate associations between phenotypes using brain morphology.
• Develop and demonstrate new applications of a linear mixed model (LMM) framework.

Main Methods:

• Utilized a linear mixed model (LMM) framework.
• Analyzed vertex-wise measures of cortical and subcortical morphology.
• Included large datasets from UK Biobank (UKB) and Human Connectome Project (HCP).

Main Results:

• Identified significant morphometricity for 58 out of 167 tested phenotypes.
• 23 phenotypes showed replicated significant morphometricity.
• Body size metrics substantially explained morphometricity, offering insights into psychiatric disorder associations.

Conclusions:

• The LMM framework effectively links phenotypes to brain morphology.
• Body size is a significant factor influencing brain-structure-phenotype relationships.
• The approach allows for phenotype prediction and novel analyses like ROI comparisons.