Monitoring Depth of Anesthesia by Nonlinear Correlation Measures

Background: Monitoring the depth of anesthesia (DOA) takes an important role for anesthetists in order avoiding undesirable reactions such as intraoperative awareness, prolonged recovery and increased risk of postoperative complications.The Central Nervous System (CNS) is the main target of anesthetic drugs, hence EEG signal processing during anesthesia is helpful for monitoring DOA. In order to study and interpret charactheristics of signals, assessment of orrelation between neurophysiological signals is an usful tool. Methods: This study applies nonlinear interdependence and Cross recurrence analysis measures to analyze EEG signals recorded from eight human volunteers under a brief Propofol anesthesia and compared with Cross correlation measure. The prediction probability (Pk) was used to assess the performance of measures for predictiong BIS. In order to evaluate the efficiency of all measures in distinguishing different anesthetic states, box plots of EEG-derived measures were compared with Kruskal–Wallis test. Moreover, the coefficient of variation (mean±SD) was employed to describe the index stability during different states. Correlation coefficient (R) between each index and bispectral index (BIS) is measured to investigate their performance. Results: Nonlinear measures demonstrates the less variability at the loss of consciousness in comparison with BIS (standard deviation of 0.09) and show that P values less than 0.05 and the highest Pk and higher correlation coefficient with BIS index. Conclusion: Nonlinear measures demonstrates the less variability at the loss of consciousness in comparison with BIS (standard deviation of 0.09). Furthermore, these measures demonstrate good performance to differentiate anesthetic states and had higher correlation with BIS and better robustness to noise in comparison with linear measure.