Topological analysis of differential effects of ketamine and propofol anaesthesia on brain dynamics

Abstract

Research has found that the vividness of conscious experience is related to brain dynamics. Despite both being anaesthetics, propofol and ketamine produce different subjective states: we explore the different effects of these two anaesthetics on the structure of dynamic attractors reconstructed from electrophysiological activity recorded from cerebral cortex of two macaques. We used two methods: the first embeds the recordings in a continuous high-dimensional manifold on which we use topological data analysis to infer the presence of higher-order dynamics. The second reconstruction, an ordinal partition network embedding, allows us to create a discrete state-transition network, which is amenable to information-theoretic analysis and contains rich information about state-transition dynamics. We find that the awake condition generally had the ‘richest’ structure, visiting the most states, the presence of pronounced higher-order structures, and the least deterministic dynamics. By contrast, the propofol condition had the most dissimilar dynamics, transitioning to a more impoverished, constrained, low-structure regime. The ketamine condition, interestingly, seemed to combine aspects of both: while it was generally less complex than the awake condition, it remained well above propofol in almost all measures. These results provide deeper and more comprehensive insights than what is typically gained by using point-measures of complexity.

OFF-periods reduce the complexity of neocortical activity during sleep

“…RQA also has parallelisms to topological data-analysis, such as persistent homology, which relies on studying the topology of high-dimensional cloud of points (manifold) corresponding to the system's evolution in phase-space [19,38]. For example, the anterior nucleus of the thalamus shows a ring-like structure in phase space during Wake and REM sleep, but not during SWS [38].…”

“…To study the topology of the neural manifolds during the states of sleep and wakefulness, we employ similar procedures as [19,38]. Briefly, we bin the spike data to 100-ms firing count bins and then reduce its dimensionality by means of the isomap algorithm to a 3d representation.…”

“…Complexity in brain dynamics has been suggested as a necesary condition for consciousness [1][2][3]. Accordingly, the complexity of electroencephalographic (EEG) signals is observed to decrease during unconscious states, such as during sleep [4][5][6][7][8][9][10][11][12][13] or anesthesia [14][15][16][17][18][19][20]. Generally, the complexity of a signal refers to its diversity of patterns, which can be estimated from different complexty or information metricsas seen from the increasing number of publications over the last decade [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20].…”

“…Accordingly, the complexity of electroencephalographic (EEG) signals is observed to decrease during unconscious states, such as during sleep [4][5][6][7][8][9][10][11][12][13] or anesthesia [14][15][16][17][18][19][20]. Generally, the complexity of a signal refers to its diversity of patterns, which can be estimated from different complexty or information metricsas seen from the increasing number of publications over the last decade [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]. But in spite of the widely observed decrease in complexity of field recordings during unconscious states, we lack a description to the neural mechanism causing it; with some reports even challenging its relevance for consciousness [21].…”

Network Analysis of Time Series: Novel Approaches to Network Neuroscience

“…While the nervous system can have functionally infinite degrees of freedom (corresponding to the number of neurons, or voxels in a dataset), correlations between elements can constrain the evolution of the whole system to a subspace of the possible global state-space (the manifold). These low-dimensional neural manifolds have been found at every level of neural function: ensembles of spiking neurons (Gallego et al, 2017(Gallego et al, , 2018(Gallego et al, , 2020Chaudhuri et al, 2019), electrophysiological recordings (Kuo et al, 2018;Varley et al, 2021b) whole-brain fMRI studies (Shine et al, 2018(Shine et al, , 2019a, and have been conjectured to relate to differences associated with mental illness and psychopathology (Carhart-Harris and Friston, 2019;McIntosh and Jirsa, 2019). Conceptually, the idea of a neural manifold can be understood as a "space", every point of which corresponds to a possible "state" a system can be in.…”