Statistical segmentation and correlation analysis of the EEG between the different phases of general anesthesia

Electroencephalography (EEG) is routinely used to monitor general anesthesia (GA). Unanswered questions concern the possibility of using the EEG signal to classify patients as more or less sensitive to anesthetic drugs from the onset of anesthesia. We investigate here possible statistical correlation between different phases of general anesthesia. We test whether it could be possible to predict the speed of return to consciousness from the induction or the maintenance phases. For that goal, we tracked the maximum power of the -band and follow its time course. Using an optimization procedure, we quantify the characteristic shift of the -band during recovery and the associated duration. Interestingly, we found no correlation (Pearson coefficient) between these shifts and the amount of -band or iso-electric suppressions (flat EEG epochs) present in the maintenance phase or induced by a propofol injection during induction. To quantify the instability of the -band, we introduce the total variation the -band that accounts for all possible deviation from a flat line. To conclude, the present analysis shows that it would not be possible to anticipate the duration of recovery of consciousness from previous phases of general anesthesia in children and adults. Possibly the involved neuronal mechanisms during the different phases are not comparable.