Conference Agenda

Understanding the neural signatures of consciousness requires characterizing its dynamic functional architecture. Here, we analyze resting-state fMRI dynamics across different loss-of-consciousness (LoC) modalities to uncover consciousness-specific signatures. To overcome limitations of traditional sliding-window approaches, we employed phase-based dynamic functional connectivity analysis, applying the Hilbert transform to BOLD signals and clustering phase coherence matrices using K-means. This approach enabled us to differentiate conscious from unconscious states based on the dynamical properties of functional networks.

We analyzed two independent datasets: (1) 16 healthy participants in awake state, undergoing propofol anesthesia and recovery and (2) 18 participants in awake and N3 sleep. Across both unconscious conditions, we observed:

Increased structure-function coupling, with functional connectivity aligning more closely with structural connectivity.

Reduced Shannon entropy in brain state distributions, indicating diminished network diversity.

Decreased Kolmogorov-Sinai entropy in state transition dynamics, reflecting constrained functional exploration.

These findings generalize previous results from patients with disorders of consciousness [1], demonstrating robust signatures of LoC under both anesthesia and N3 sleep, despite differing acquisition parameters and mechanisms of consciousness alteration. The conscious state exhibited richer exploration of functional configurations, while unconscious states showed restricted dynamics dominated by structure-function coupling [2].

Our methodology proved robust across parameter variations (k = 3–10) and independent of global signal regression. These findings refine our understanding of consciousness as a dynamic process and establish generalizable neural markers of consciousness.

[1] Demertzi et al., Sci. Adv. 5, eaat7603 (2019). DOI: 10.1126/sciadv.aat7603

[2] Castro et al., Commun Biol 7, 1224 (2024). DOI: 10.1038/s42003-024-06858-3

Introduction: The critical brain hypothesis suggests optimal brain function emerges at the edge of chaos— a state balancing stability and flexibility that supports adaptability and is thought to underlie recovery potential. Neuroimaging criticality features have shown promise in assessing levels of consciousness. This study explores the comparative prognostic value of criticality-related features recorded during drug-free and GABAergic sedation states in severely brain injured pediatric and adult cohorts, aiming to identify optimal conditions for assessing recovery of consciousness (ROC) across the lifespan.

Methods: We analyzed two datasets comprising 32 pediatric (mean age: 11.3 ± 3.2 years) and 24 adult (mean age: 55.0 ± 19.2 years) patients with severe brain injuries. From electroencephalogram recordings, we extracted spectral power and criticality features such as aperiodic slope, Hurst exponent, Lempel-Ziv complexity and permutation entropy. Recovery was classified based on the Glasgow Outcome Scale-Extended 3-months post-injury. Machine learning models were trained separately on features from drug-free and GABAergic sedation state recordings (propofol or midazolam) to predict recovery.

Results: Sedation state models outperformed drug-free state models in predicting ROC. Across both cohorts, sedation state models achieved higher predictive accuracy (area under the curve (AUC): 0.82 ± 0.04) than baseline (AUC: 0.70 ± 0.06).

Conclusion: These findings suggest that brain dynamic patterns under GABAergic sedation provide greater prognostic potential for ROC than a drug-free state. This has strong implications for prognostic tools that can be implemented in clinical practice in pediatric and adult cohorts. This work advances our understanding of consciousness and recovery dynamics across the lifespan.

Advances in neurology intensive care unit have increased the survival rate of patients after lesional coma. However, their outcome remains unpredictable, highlighting the need for therapeutic approaches to recover consciousness and facilitate functional recovery to return to an inter-individual communication and a valuable social life. Nowadays, our understanding of the neural mechanisms involved in consciousness remains limited. Numerous studies highlighted the key role of various deep brain structures, including thalamus and basal ganglia, involved in the "mesocircuit" hypothesis for the neural network modulating the states of awareness. The goal of this study is to provide causal inference in this mesocircuit model by modeling the brain activity disturbances induced by intralaminar nuclei inactivation (i.e. centromedian-parafascicular complex [CM-PF] and centrolateral [CL]) and compared to dorsomedial thalamus [DM]. We conducted reversible electrical stimulations in two Macaca fascicularis for nucleus screening and pharmacological disruptions of the GABAergic transmission (using muscimol and bicuculline microinjections) into pre-selected targets. The behavioral results obtained impaired attentional tasks. Brain imaging results in PET-MR with [18F]-FDG identified the CM/PF complex as a thalamic territory relevant to impact attentional processes and to modify the activity of a large set of cortical and subcortical regions of the mesocircuit, including the medial frontal cortex (cingulate cortex and SMA) and the anterior striatum (caudate nucleus and putamen). These results highlight the role of the CM/PF complex at crossroad of ascending reticular stimulations and cortico-subcortical loops modulation, allowing to focus therapeutic intervention upstream, downstream or in this thalamic hub.

Disorders of consciousness (DoC) occur after cerebral injuries in strategic brain areas involved in the mesocircuit. The assessment of these disorders is based on a behavioural gold-standard scale called Coma Recovery Scale-Revised (CRS-R) to assess the state of consciousness and the functional abilities of brain-damaged patients. However, several studies pinpoint a misdiagnosis rate of around 40% in distinguishing between unresponsive wakefulness syndrome (UWS) and minimally conscious state (MCS) patients. Thus, it is necessary to create new tools to refine, and objectify the clinical assessments during CRS-R. The present study combined eye-tracking with the CRS-R assessment to identify different visual explorations between UWS and MCS patients. We recruited DoC patients with no sign of awakening 48h after sedation cessation in Intensive Care Unit. All patients were equipped with a wearable eye-tracker to perform a quantitative assessment of eye movements during CRS-R. We measured the number of saccades, the duration of fixation, and the variation of pupil diameter during the exploration of objects, the patient's face reflected in a mirror, and the physician's face performing the assessment. Additionally, we studied the effect of familiarity by presenting images provided by patients' relatives (familiar) and unknown images (non-familiar). Our preliminary results indicate that the stimulus social nature was relevant as the responses elicited by faces was different between MCS and UWS patients. Notably, pupillary reactivity varied between MCS and UWS patients during social stimuli. These findings suggest that eye-tracking metrics could objectify CRS-R assessments and might enhance the reliability of DoC evaluations.

Introduction: Coma and other disorders of consciousness are common clinical manifestations of severe traumatic brain injury (TBI). Recovery of consciousness is often hard to measure at the bedside, and while neurophysiology dynamics may inform recovery assessment, advancements in understanding its mechanisms remain limited.

Method: We recorded invasive electrocorticography (ECoG) in patients undergoing hemicraniectomy following severe TBI using a 6-electrode strip array. Twelve hours of continuous ECoG data (recorded within 72hours from TBI) were analyzed. Neurophysiological dynamics were evaluated using Shannon Entropy for complexity, and measures of synergetic and redundant information flow, accounting for spatial distribution across electrodes over time. To evaluate the spatial distribution of inter-region information flow, we used Dynamic Time Warping to measure regional-information flow similarity across electrodes and time scales. Poor outcome(PO) and good outcome (GO) was defined as the ability to follow commands prior to discharge using Glasgow Coma Score motor exam. We performed a two-sample t-test between outcome groups.

Results: We analyzed data from 12 (6 GO). The GO group showed higher entropy [GO= 1.50e4±0.44e4, PO=0.96e4±0.55e4, p(t)=0.05(1.78)] and enhanced synergetic information flow in the alpha and beta bands. Spatial distribution of information flow over time revealed greater alpha-synergy [GO=1.78±0.63, PO=0.49±0.32; p=0.0006] and beta-synergy [GO=2.02±1.2, PO=0.54±0.31; p=0.0093] similarity in the GO group, suggesting a spatial-temporal gradient in information flow fluctuation over time in patients recovering consciousness.

Conclusion: Spatial-temporal changes in brain complexity and region-specific information flow precede consciousness recovery after severe TBI. Larger studies are needed to validate these findings, considering scalp EEG and sedative effects.

Introduction

In the search for EEG markers of human consciousness, alpha power has long been considered a reliable marker which is fundamental for the assessment of unresponsive patients from all etiologies. However, recent evidence questioned the role of alpha power as a marker of consciousness and proposed the spectral exponent and spatial gradient as more robust and generalizable clinical indexes. In this study, we investigated etiology-specific differences in clinical markers of level of consciousness.

Methods

We analyzed an existing dataset containing high-density resting-state EEG recordings from 303 patients with a disorder of consciousness. Patients were split in three groups according to their etiology (anoxic brain injury, non-anoxic injury and other). Patients’ level of consciousness was assessed using the Coma Recovery Scale-Revised. We compare a set of candidate EEG makers: 1) absolute, relative and flattened alpha power; 2) the posterior-anterior ratio; 3) the spectral exponent; and 4) Lempel-Ziv complexity. Analysis of diagnostic value was performed using Mann-Whitney-U-test and Spearman rank test.

Results

Diagnostic value of EEG features varied between etiologies. Alpha power had a higher diagnostic importance for anoxic compared to non-anoxic patients. Meanwhile, the spectral slope showed diagnostic value for non-anoxic patients only. The diagnostic value of the posterior-anterior alpha ratio vanished for all etiology groups.

Conclusion

Grouping unresponsive patients from different etiologies together can confound or obscure the diagnostic value of EEG markers of consciousness. Our study highlights the importance of analyzing different etiologies independently and emphasizes the need to develop clinical markers which better account for etiology-dependent differences.