A topography-based predictive framework for naturalistic viewing fMRI

Recent work has shown great interest in understanding individual differences in complex brain function under naturalistic viewing (NV) conditions. However, methods specifically designed for achieving this goal remain limited. Here, we propose a novel approach, called TOpography-based Predictive Framework (TOPF), to investigate individual differences in evoked brain activity on NV fMRI data. Specifically, TOPF identifies individual-specific evoked activity topographies in a data- driven manner and examines their behavioural relevance using a machine learning predictive framework. Our results show that these topographies successfully predict individual phenotypes across cognition, emotion and personality on unseen subjects, and the identified predictive brain regions are neurobiologically interpretable. Further, the prediction accuracy exceeds that of the commonly-used functional connectivity-based features. Conceptually, we highlight the importance of examining multivariate evoked activity patterns for studying brain-behaviour relationships. In summary, we provide a powerful tool for understanding individual differences and brain-behaviour relationships on NV fMRI data.