Group-specific discriminant analysis enhances detection of sex differences in brain functional network lateralization.

Abstract

Background: Lateralization is the asymmetry in function and cognition between the brain hemispheres, with notable sex differences. Conventional neuroscience studies on lateralization use univariate statistical comparisons between male and female groups, with limited and ineffective validation for group specificity. This article proposes to model sex differences in brain functional network lateralization as a dual-classification problem: first-order classification of left versus right hemispheres and second-order classification of male versus female models. To capture sex-specific patterns, we developed an interpretable group-specific discriminant analysis (GSDA) for first-order classification, followed by logistic regression for second-order classification.

Findings: Evaluations on 2 large-scale neuroimaging datasets show GSDA's effectiveness in learning sex-specific patterns, significantly improving model group specificity over baseline methods. Major sex differences were identified in the strength of lateralization and interaction patterns within and between lobes.

Conclusions: The GSDA-based analysis challenges the conventional approach to investigating group-specific lateralization and indicates that previous findings on sex-specific lateralization will need revisits and revalidation. This method is generic and can be adapted for other group-specific analyses, such as treatment-specific or disease-specific studies.