Multivariate Developmental Pattern of Cortical Thickness, Brain Functional Connectivity and Behaviour: A Multi-Block Partial Least Square Discriminant Analysis

The intricate relationship between emotional and behavioral traits and brain anatomy and function during childhood and adolescence is crucial for elucidating the neurobiological basis of diverse psychopathological disorders but it still largely unknown. In this study, we aimed to quantitatively evaluate the multivariate patterns of association between brain activity, cortical brain characteristics, and core clinical and behavioral features in a sample of forty subjects with age ranging between 11-21 years who were referred for emotional-behavioral problems. By exploiting a multi-block partial least square discriminant analysis (MBPLS-DA) framework, we attempt to explain the multivariate relations among the Child Behavior Checklist (CBCL) and brain structure and function assessed via magnetic resonance imaging (MRI). Based on individual importance of each block, our findings showed a prominent contribution of regional cortical thickness data in explaining the variance of CBCL scales. By analyzing each block's loadings individually, a prominent positive association was found among regional cortical thickness in paracentral, lateral occipital cortex, precuneus and CBCL composite scores of psychotic, autism spectrum disorder risk profiles and withdrawn scale. Furthermore, the same CBCL scores were also found positively associated with functional connections involving regions of anterior and median cingulate cortex, caudate nucleus, inferior parietal lobule and angular gyrus. Our investigation suggests a key role of brain structure and function in shaping behavioral traits that might be predispose to psychopathology during childhood and adolescence. The proposed MBPLS-DA application promotes broader adoption of this technique across diverse and multimodal MRI and clinical datasets.