Spectral Slope and Lempel–Ziv Complexity as Robust Markers of Brain States during Sleep and Wakefulness

Höhn, Christopher; Hahn, Michael A.; Lendner, Janna D.; Hoedlmoser, Kerstin

doi:10.1523/eneuro.0259-23.2024

Cited by 10 publications

(1 citation statement)

References 87 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A new generation of metrics, evaluating more elaborate aspects in EEG signals such as complexity, has recently emerged (e.g., [8][9][10] ). These metrics are grounded in hypotheses positing that consciousness stems from the intricate interactions across the entire brain.…”

Section: Introductionmentioning

confidence: 99%

Registered report: Common signatures of loss of consciousness in human and macaque electrocorticogram

Shimaoka,

Leung,

Price

et al. 2024

Preprint

View full text Add to dashboard Cite

Despite the widespread application of general anesthesia for surgical procedures since the 19th century, a fundamental understanding of its operating principle and a reliable metric for Loss of Consciousness (LoC) remain elusive, leaving unmitigated risks such as overdose or intraoperative awareness. Given that LoC is inducible by the same anesthetics such as propofol in diverse animal species, here we propose that there are corresponding shared neuronal mechanisms that underlie LoC, which manifest as common signatures in neural activity across species. In this project, our goal is to identify shared metrics for detecting LoC induced by propofol in both humans and macaque monkeys without pre-specifying candidate metrics. To facilitate this, we employed highly comparable time-series analysis (hctsa), computing over 7000 univariate time-series features from various disciplines. Initial results from our pilot study (one monkey and one human) will be followed by more extensive analyses in the registered report format.For Stage 1 of the registered report, we first identified hctsa feature(s) that distinguish awake and propofol-anesthetized states from one macaque intracranial data set. For each feature, we trained a nearest median classifier using 200 ms data epochs, and evaluated its classification accuracy with 10-fold cross-validation in the macaque. We subsequently evaluated its classification accuracy in human intracranial data, recorded from one participant with medically refractory epilepsy.Results from Stage 1 provide strong evidence of the existence of univariate features capable of discriminating between awake and propofol-induced unconscious states in both species. For example, a hctsa feature characterizing model fitting performance trained on the macaque frontal lobe achieved significant (p<0.00013) classification accuracy both in macaque (75% accuracy) and human frontal lobe recordings (82%) after a False Discovery Rate correction (q=0.01). Building on Stage 1 findings, Stage 2 will expand the investigation to include recordings from another macaque and ten humans, aiming to validate the generalizability of identified features across species. Ultimately, the project seeks to enhance our understanding of LoC mechanisms, mitigating risks associated with anesthesia administration in surgical procedures both in humans and other species.

show abstract

Section: Introductionmentioning

confidence: 99%