Dynamic Fingerprinting of the Human Functional Connectome

Introduction: Resting-state functional connectivity (FC) has distinct, personalized patterns that could serve as a unique fingerprint of each individual’s brain. While previous brain fingerprinting methods have used FC maps over a scanning session (static method), it has been shown that the brain is a dynamic system that switches between several metastable states, each of which has a different FC map. Taking the dynamic nature of brain connectivity into account will likely lead to more subject-specific information and better individual identification. Methods: In this article, we derived the state-specific FCs using sliding window correlation and clustering and evaluated their performance in individual identification and cognitive score prediction. Results: The resultant dynamic fingerprints outperformed the static fingerprints in identification accuracy. Furthermore, some of the brain states were more accurate in predicting cognitive scores, indicating that connectivity in some brain states is informative of cognitive abilities, possibly useful as biomarkers for brain disorders. Discussion: These findings suggest that incorporating dynamic information captures subject-specific connectivity features that are not present in static FC alone. The observation that specific states contribute more to cognitive prediction further highlights their potential utility as biomarkers for brain disorders.