2020/12/03 by Gabriel Schamberg, Schamberg, Gabriel, S. Chakravarty +5
Neuroscience · Chemistry · Medicine · #EEG and Brain-Computer Interfaces #Electrochemical Analysis and Applications #Anesthesia and Sedative Agents
paper · pdf · doi:10.48550/arxiv.2012.02246
Burst suppression is an electroencephalography (EEG) pattern associated with\nprofoundly inactivated brain states characterized by cerebral metabolic\ndepression. Its distinctive feature is alternation between short temporal\nsegments of near-isoelectric inactivity (suppressions) and relatively\nhigh-voltage activity (bursts). Prior modeling studies suggest that\nburst-suppression EEG is a manifestation of two alternating brain states\nassociated with consumption (during a burst) and production (during a\nsuppression) of adenosine triphosphate (ATP). This finding motivates us to\ninfer latent states characterizing alternating brain states and underlying ATP\nkinetics from instantaneous power of multichannel EEG using a switching\nstate-space model. Our model assumes Gaussian distributed data as a broadcast\nnetwork manifestation of one of two global brain states. The two brain states\nare allowed to stochastically alternate with transition probabilities that\ndepend on the instantaneous ATP level, which evolves according to first-order\nkinetics. The rate constants governing the ATP kinetics are allowed to vary as\nfirst-order autoregressive processes. Our latent state estimates are determined\nfrom data using a sequential Monte Carlo algorithm. Our\nneurophysiology-informed model not only provides unsupervised segmentation of\nmulti-channel burst-suppression EEG but can also generate additional insights\non the level of brain inactivation during anesthesia.\n