Connectome Similarity Varies with Global Metastability and Movie Content

Abstract

IntroductionRecent evidence supports that non-linear segments of brain activity carries most of the individual-specific information which allows unique identification based on functional connectomes (e.g. connectome fingerprints) in MEG. In fMRI, metastability has emerged as a reliable and robust summary measure of brain dynamics. Here, we set out to test the hypothesis that metastability is associated with connectome fingerprinting in naturalistic fMRI. We further examined whether this putative association varies with time scales, across brain functional networks, and depending on the movie content. MethodsWe analyzed fMRI data from 86 participants of the Naturalistic Neuroimaging Database. Data was collected during movie-watching of 10 entire movies of different genres. Global and network-level metastability were calculated using an fMRI-appropriate proxy measure based on the variability of spatial coherence. The within-subject stability (Iself) and between-subjects similarity (Iothers) of individual functional connectomes were also calculated. Linear mixed-effects models and appropriate contrasts were used to test associations between metastability and fingerprinting indices. ResultsAt the whole-brain level and considering data from all movies, an association across windows was found between metastability, Iself, and Iothers, with stronger effects observed for larger windows, such as the 600-second window (Iself: b = -0.13, SE = 0.032, p < 0.001; Iothers: b = -0.23, SE = 0.028, p < 0.001). Associations were also observed with distinct magnitudes within the eight large-scale functional networks, with particularly strong effects in the Motor Network (Iself: b = -0.34, SE = 0.027, p < 0.001; Iothers: b = -0.27, SE = 0.021, p < 0.001). Although no whole-brain differences were observed, movie content exhibited distinct association patterns within networks, notably for the prediction of Iself in the Visual Network for 500 Days of Summer ({Delta} = -0.22, SE = 0.11, p = 0.043). For Iothers, significant between-movie differences were found in the AMN ({Delta} = -0.25, SE = 0.062, p < 0.001), Dorsal Attention ({Delta} = 0.16, SE = 0.08, p = 0.038), Ventral Attention ({Delta} = 0.14, SE = 0.060, p = 0.017), and Visual Networks ({Delta} = -0.17, SE = 0.081, p = 0.031). ConclusionMetastability is systematically related to functional connectome fingerprints. This relationship is time-scale dependent, varies across functional networks, and is sensitive to movie content. Our findings bridge nonlinear brain dynamics and connectome identifiability, highlighting naturalistic paradigms as a powerful framework for individualized precision neuroimaging.