Propagation Mapping: A Framework for Modeling Whole-Brain Propagation Patterns of Task-Evoked Activity

Human brain mapping has traditionally relied on univariate approaches to characterize regional activity, whereas more recent work focuses on interactions between regions to capture network-level organization. Despite their parallel development, growing evidence suggests that integrating both approaches is critical for a comprehensive understanding of task-evoked brain activity. The present study introduces propagation mapping, an extension of activity flow mapping that models task-evoked brain activity as the propagation of regional signal amplitudes along whole-brain topological routes. This study aims to evaluate propagation maps as reliable neurobiological features for neuroimaging research. Using functional connectomes and structural covariance network derived from a large normative sample (n=1,000), propagation patterns of task-evoked activity were accurately captured (average R2 = 0.947, MAE=0.155, and RMSE=0.229) across 94 participants. Mapping performance remained stable across different task contrasts, parcellation atlases, and signal intensity and spatial distance between regions. Similar performance was observed at both the subject and group levels in an independent sample using the amplitude of low-frequency oscillations during resting-state (n=189). Importantly, despite its reliance on normative connectomes which could homogenize subject-specific variance, propagation mapping instead redistributed individual variance along propagation routes (Cohens d = 0.10, p=0.17). As a biologically comprehensive representation of brain organization, propagation mapping offers a powerful and user-friendly alternative to traditional regional analyses and provides new avenues for discovery in neurological and psychiatric neuroimaging research.