Assessing nitrous oxide effect using electroencephalographically-based depth of anesthesia measures cortical state and cortical input

Abstract

Existing electroencephalography (EEG) based depth of anesthesia monitors cannot reliably track sedative or anesthetic states during n-methyl-d-aspartate (NMDA) receptor antagonist based anesthesia with ketamine or nitrous oxide (N₂O). Here, a physiologically-motivated depth of anesthesia monitoring algorithm based on autoregressive-moving-average (ARMA) modeling and derivative measures of interest, Cortical State (CS) and Cortical Input (CI), is retrospectively applied in an exploratory manner to the NMDA receptor antagonist N₂O, an adjuvant anesthetic gas used in clinical practice. Composite Cortical State (CCS) and Composite Cortical State distance (CCSd), two new modifications of CS, along with CS and CI were evaluated on electroencephalographic (EEG) data of healthy control individuals undergoing N₂O inhalation up to equilibrated peak gas concentrations of 20, 40 or 60% N₂O/O₂. In particular, CCSd has been devised to vary consistently for increasing levels of anesthetic concentration independent of the anesthetic’s microscopic mode of action for both N₂O and propofol. The strongest effects were observed for the 60% peak gas concentration group. For the 50–60% peak gas levels, individuals showed statistically significant reductions in responsiveness compared to rest, and across the group CS and CCS increased by 39 and 42%, respectively, while CCSd was found to decrease by 398%. On the other hand a clear conclusion regarding the changes in CI could not be reached. These results indicate that, contrary to previous depth of anesthesia monitoring measures, the CS, CCS, and especially CCSd measures derived from frontal EEG are potentially useful for differentiating gas concentration and responsiveness levels in people under N₂O. On the other hand, determining the utility of CI in this regard will require larger sample sizes and potentially higher gas concentrations. Future work will assess the sensitivity of CS-based and CI measures to other anesthetics and their utility in a clinical environment.

Appendix

As mentioned in the results, the analysis of frequency and damping of the poles focused on poles with frequencies in the band of 5–15 Hz because Yamamura et al. [39] observed strong effects of N₂O on damping in this frequency band. Here the effects of N₂O on the distributions of the poles are demonstrated for individuals for the highest N₂O gas levels.

1.1 Pole histograms of individuals in the 60% group

For the rest and gas conditions, some of the effects of N₂O on the EEG can be captured in the pole histograms of the ARMA model which reflect the normalized distributions of the dominant damped oscillations in the EEG. Figure 4a partly schematizes the meaning of these pole histograms which are plotted on the complex (number) plane because the poles, ${{\rho }_{k}}$, are represented by complex numbers. As described in the methods, the angle of a complex pole with respect to the origin of the complex plane corresponds to the oscillatory frequency of the pole. The radial distance of a pole from the origin, reflects the damping of the pole (the closer to the origin the more damped the oscillation). Figure 4b–e show the pole histograms for the rest (top row) and gas (middle row) data, and the difference between the rest and gas histograms (bottom row) for the four individuals in the 60% peak gas group. Figure 4b–e correspond to subjects 17, 18, 19 and 20, respectively. The gas data corresponds to the last 5 min of the equilibrated peak gas period, except for subjects 18 and 19 who, due to nausea and emesis, had their gas recordings shortened to less than 5 and 10 min in duration, respectively. For these two subjects the last 2 min of their gas recordings were used. In each subfigure the poles have been histogrammed by binning the complex plane. In the rest and gas data rows, only the portion of the complex plane corresponding to positive frequencies between 0 Hz (horizontal axis) and 20 Hz (vertical axis) is shown. In the row highlighting the differences in the rest and gas histograms the region of the complex plane corresponding to the significant alpha subband (8–12 Hz) changes is focused on. The rest data row indicates the presence of a strong alpha (8–12 Hz) peak during the resting eyes closed condition for all individuals. In the gas data row it can be seen that alpha poles within the same region have become damped. This is also reflected in the row showing the histogram differences where the radially inward blue to red shifts indicate an increase in damping when going from the rest to the gas cases.

Fig. 4

Normalised pole histograms for individuals in the 60% peak gas group. a Schematic illustrating the relationship between pole frequency and damping in the complex plane. An increase in pole angle coincides with an increase in pole frequency and thus the frequency of the dominant oscillations in the EEG. A decrease in pole radius, $\text{ }\!\!\gamma\!\!\text{ }$, means an increase in pole damping and thus a decrease in amplitude of the dominant oscillations of the EEG. b–e The pole histograms for the rest (top row) and gas (middle row) data, and the subtraction of the gas histogram from the rest histogram zoomed in at the alpha band region (bottom row). b–e correspond to the same participants in Fig. 2a–d, respectively. The color bars on the right of the gas histogram code the probability of a pole occurring at a particular point in the complex plane and apply to both the rest and gas histograms of the corresponding individual. The color bars to the right of the difference between the histograms codes the probability difference between the two normalized histograms. Blue, white and red indicate where the rest histogram is greater than, equal to, and less than the gas histogram, respectively