Evaluation of behavioral variance/covariance explained by the neuroimaging data through a pattern-based regression

Area of Science:

• Neuroscience
• Cognitive Science
• Psychiatry

Background:

• Neuroimaging studies often focus on limited brain regions, hindering a holistic understanding of behavior.
• Existing methods struggle to capture the overarching relationship between brain activity and complex human behaviors.

Purpose of the Study:

• To develop and validate a voxel-based pattern regression method for analyzing brain-associated variance (BAV) across behaviors.
• To investigate shared neural underpinnings between different behavioral measures using neuroimaging data.
• To provide a robust statistical tool for advancing the study of human behavior through neuroimaging.

Main Methods:

• Proposed a voxel-based pattern regression technique to assess brain-wide associations with behavior.
• Validated the method's reliability and accuracy through comprehensive simulations.
• Applied the approach to large-scale neuroimaging datasets (IMAGEN and HCP) encompassing diverse behavioral measures.

Main Results:

• The voxel-based pattern regression demonstrated high reliability and accuracy in simulations.
• Intelligence scores showed significant brain-associated variance with gray matter volume in both adolescent and adult datasets.
• Identified shared neural architectures underlying depression and externalizing problems in adolescents, suggesting early developmental links.

Conclusions:

• The proposed voxel-based pattern regression offers a powerful tool for exploring brain-behavior relationships.
• The method can uncover latent correlations between multiple behaviors, providing novel insights.
• This approach enhances our ability to understand the neural bases of human behaviors using neuroimaging data.