Analysis Of Intraoperative Nociceptive Responsiveness Based On EEG Signal Coupling Feature

Moderate anesthesia makes patients safe and painless during the operation,and provide a stable and good operation environment for surgeons;However,the existing clinical anesthesia monitoring methods are not comprehensive and far from the demand for precise anesthesia.At present,there are two kinds of methods for monitoring the responsiveness of noxious stimulation.One is based on the autonomic nerve activity to monitor the physiological characteristics such as electrocardiogram,respiration,and pulse that are affected by the autonomic nerve.Such methods are susceptible to physiological conditions such as blood circulation system and respiration.The other is based on electroencephalography(EEG),which monitors from the perspectives of spectrum,nonlinearity and evoked EEG.The EEG-based methods can directly measure the responsiveness of noxious stimulation in the cerebral cortex.At present,the assessment methods of responsiveness of noxious stimulation which are based on EEG spectrum characteristics has been recognized in clinical practice,but there are still some shortcomings,such as being vulnerable to interference and low detection accuracy.The study of anesthesia related mechanism has proved that change of EEG coupling is one of the main effects of anesthetics on brain.However,so far,few studies have explored responsiveness of noxious stimuli from the perspective of EEG coupling.In this paper,EEG coupling relationship was introduced to evaluate responsiveness of noxious stimuli,and the features of EEG coupling were combined with the frequency domain and nonlinear features of EEG to develop methods for intraoperative analgesia monitoring.A clinical trial using sevoflurane as induction and maintenance drug,meanwhile intubation and tetanization as standard noxious stimulation was designed,and multi-channel EEG data were collected from 38 patients undergoing lower extremity surgery.FIR filter,independent analysis,wavelet threshold denoising and quadratic variation reduction were used to remove the possible power frequency,electrocardiogram(ECG)and electrooculogram(EOG)interference,electromyography(EMG)artifacts and baseline drift in the collected data.Modulation index(MI)was introduced to characterize the phase-amplitude coupling patterns in and between EEG channels.In addition,this paper analyzes the power ratio of EEG frequency bands,sample entropy and the relationship between responsiveness of noxious stimuli,and confirms that the α-band of central parietal lobe was weakened and the γ-band of frontal lobe was enhanced when the somatomotor response to noxious stimulation,and sample entropy shows a consistent increasing trend in all EEG channels.Finally,the random forest algorithm is used to analyze the relevant feature weights,and the features are selected and combined to generate the discrimination model and prediction model of responsiveness to noxious stimulus.Through clinical data verification,the accuracy of the discriminant model reached92.22%,while the accuracy of the prediction model was 78.89%.In the evaluation of the influence of different feature combinations on the model,it was found that the discrimination of noxious stimulus responsiveness was mainly affected by the phase amplitude coupling mode in the prefrontal channels.The prediction of responsiveness was mainly influenced by the phase amplitude modulations within and between the frontal and parietal channels.Moreover,the addition of EEG phase amplitude coupling features improves the accuracy of the model in the discrimination and prediction of noxious stimulus.