Research On The Anesthesia Depth Monitoring Algorithm Based On The Analysis Of EEG Signal

Anesthesia is an important part in modern surgery.According to the main parts of the anesthesia,the clinical anesthesia can be mainly divided into the local anesthesia and the general anesthesia.This thesis mainly discusses the method of monitoring depth of anesthesia in general anesthesia.It is generally believed that general anesthesia mainly affects the activity of Central Nervous System by using a variety of anesthetic drugs to finally reach a sedative,unconsciousness and forgetting state,which is suitable for surgical procedures.In the process of clinical anesthesia,too high doses of anesthesia drugs may lead to the patients difficult to wake up,and also cause the waste of anesthesia drugs.On the other hand too low doses of anesthesia drugs may lead to the patients feel pain and even wake up during the operation.Inadequate anesthesia depth will bring patients with varying degrees of physical or psychological damage.So to control and maintain a proper anesthesia depth is the main task of the clinical anesthesia doctor.Since,the anesthesia drugs acts mainly on the Central Nervous System,the EEG signal processing during anesthesia is useful to monitor the patient's depth of anesthesia.The research of the method of monitoring the depth of the anesthesia based on the EEG signal analysis has a long history.The method of EEG signal analysis has developed form the time-frequency domain analysis method to the nonlinear dynamic analysis method.And then,the nonlinear dynamic analysis method has attracted many researchers' attention in the field of the research of monitoring the depth of the anesthesia.In this dissertation,we have collected 30 cases EEGs recorded during general anesthetic surgery,the collected EEG signals ware divided into three parts in this study due to different anesthesia state.In this study,firstly,the noise must be removed from the EEG signal for the analysis in next step before feature detection.And then,all nonlinear complexity analyzing methods and their applications in EEG signal processing were summarized.We selected four complexity measures—approximate entropy,sample entropy,permutation entropy and wavelet entropy to analysis 30 cases EEG signal.The results showed that the approximate entropy,sample entropy,permutation entropy and wavelet entropy of the EEGs obtained from wake state,light anesthesia state and moderate anesthesia state respectively all exist obvious difference,and the deeper the depth of the anesthesia,the smaller the index of the complexity measures.And then,we made a comparison in these four complexity measures for monitoring the depth of anesthesia using a BP neural network.At the same time,to explore the best network model for monitoring depth of anesthesia we made a comparison in the network which using single EEG feature and the network which using multi-features.The research showed that the BP neural network which combined the sample entropy feature with the wavelet entropy feature worked best in monitoring the depth of anesthesia,the best judgment accuracy of the network can be as high as 99.98%.Our research had got a new idea to monitor the depth of anesthesia accurately.At the end of this paper,further research interests in the future were prospected through the results of this research.