Dynamic inter-regional coordination patterns as specific predictors of consciousness

Background and aims: To date, specific signatures of conscious states in humans remain elusive. Contemporary theories concur that such markers can be traced to temporally evolving brain processes instead of static descriptions of brain activity. Methods: Dynamic fMRI connectivity patterns (states) by means of clustering of phase-based coherence was estimated on 47 healthy and 112 patients diagnosed in vegetative state/unresponsive wakefulness syndrome (VS/UWS) or in a minimally conscious state (MCS). To validate whether the patterns captured properties of awareness, out-of-sample generalization was performed on patients with cognitive-motor dissociation (i.e. lacking overt conscious behaviour yet evidenced using functional neuroimaging), and on anesthetised patients, under the premise that complex signatures would disappear uniformly across all subjects. Results: A pattern of long-range positive/negative coherence had a higher probability of occurring in healthy and MCS patients. A pattern of low inter-areal coordination, mostly similar to anatomy, was more likely to occur in VS/UWS. Inter-state transitioning was flexible for healthy and MCS and more rigid for VS/UWS patients. Unconscious patients were more likely to avoid the exploration of the complex connectivity state. The generalization to cognitive-motor dissociation predicted the occurrence of the complexconnectivity state. The generalization to propofol anaesthesia showed an equalization of occurrence probabilities of all patterns regardless of clinical diagnosis. Conclusion: The dynamics of inter-areal coordination contain information specific to conscious awareness. The rigid and less metastable dynamics in VS/UWS could account for the limited mental capacities in these patients. The minute identification of these patterns and their external manipulation could account for non-invasive restoration of consciousness.