How to Detect the Depth of Anesthesia in Human Brain Waves Utilizing Relative Phase Dynamics

1650 

Abstract Submission 

Youngjai Park^1,2, Joon-Young Moon^1,2

¹Center for Neuroscience Imaging Research, Institute for Basic Science (IBS), Suwon, Republic of Korea, ²Sungkyunkwan University, Suwon, Republic of Korea

Youngjai Park  
Center for Neuroscience Imaging Research, Institute for Basic Science (IBS)|Sungkyunkwan University
Suwon, Republic of Korea|Suwon, Republic of Korea

Joon-Young Moon  
Center for Neuroscience Imaging Research, Institute for Basic Science (IBS)|Sungkyunkwan University
Suwon, Republic of Korea|Suwon, Republic of Korea

The human brain shows emergent states such as consciousness and unconsciousness and dynamic transition between them. However, after many years of research, the search to understand the mechanism of conscious to unconscious transition and to correctly measure the level of consciousness still remains elusive. In recent years, researchers have found that the phase dynamics of low-frequency brain waves are related to the level of consciousness [1]. In this study, we further pursue to study the relationship between the phase dynamics of brain waves and the level of consciousness in anesthesia: we analyze the relative phase patterns of brain waves across the whole brain network and its relationship with conscious to unconscious state transitions using the electroencephalogram (EEG) recordings over anesthesia. We aim to understand how the phase dynamics of cortical brain waves differ between conscious and unconscious states, and what mechanism can account for such differences.

We analyze the EEG of 18 human participants covering the whole brain (128 channels) where 9 of them undergo general anesthesia and recover. We first band pass-filtered the brain wave signals from each channel and separate frequency bands (i.e., delta:1-4, theta:4-8, alpha:8-12, low-beta:12-18, and high-beta:18-30 Hz bands). We then define the relative phase of each channel at the specific frequency band by subtracting the global mean phase of the brain waves from the phase of the wave from each channel. By subtracting the global mean phase, we can gather the information on which part of the brain leads or lags behind the phase of the others.

We find that there are two robust and dominant modes of global phase patterns: parietal area phase-leading or lagging the frontal area. The brain dynamically transitions between these two dominant modes during various conscious states (e.g., resting states, and general anesthesia states). As the participants undergo general anesthesia, the dominance of these two modes begins to disappear and in deep anesthesia, these patterns themselves disappear: the global phase pattern becomes almost random (see Fig. 1). Front-to-back pattern, which is the frontal area leading the phase, is related to the depth of anesthesia in the K-mean cluster result. In the transition matrix in eyes open (conscious) and suppression (unconscious) states, the deeper the level of consciousness is, the more random the relative phase patterns become. Consistently, the distribution of the eigenvalues becomes homogeneous when the depth of anesthesia is deeper in the principal component analysis (PCA). As a result, the relative phase patterns can be a measurement to detect the level of consciousness.
We further study the functional connectivity from the brain wave dynamics, by constructing whole brain networks where each channel becomes a network node and the connection between nodes is given by the similarity between the waves from respective channels. The parietal area becomes hubs (i.e., the nodes with many connections). Therefore, two dominant modes of phase patterns (i.e., parietal phase-lead or -lag) can be identified as either the hub area phase-leading or -lagging.

Altogether, we find that the hub phase-leading and -lagging patterns are the robust and dominant modes in conscious states while the pattern becomes random in unconscious states. These patterns can be interpreted as information flow patterns (the hub being either the source or the sink) which dominantly exist during the conscious state [2], disappearing in the unconscious state. Therefore, the action of the anesthetics is to disturb the dominant information flow patterns existing in the conscious state.

EEG/MEG Modeling and Analysis ¹

Methods Development ²

EEG

Consciousness

Data analysis

Electroencephaolography (EEG)

Other - Relative phase