Permutation entropy: A reliable measure for automatic monitoring of anesthetic depth during surgery?

Permutation Entropy (PE) has recently been applied to characterize anesthetic-induced changes in the frontal electrical brain activity (EEG) during anesthesia. In this work we investigate the stability of PE as a means of identifying between the awake and anesthetized EEG over the entire duration of surgery under different anesthetic regimes and using a full set of EEG sensors. Average classification rates from 22 patients range between 98-99% (specificity, sensitivity and accuracy), when using information from whole-head EEG. The findings support the robustness of PE for discriminating 'awake' and 'anesthesia' throughout the entire surgery, independently of the anesthetic regime followed.