Multiple Physiological Signal Analysis For The Clinically Anesthetic Status Monitoring And Evaluation

Anesthesia is a significant compoent of modern medicine and plays a key role in assuring the sucessful implementaion of the different kinds of operations.To control the anesthesia procedure precisely and gurantee the patients' saftey and good experience of surgeries,the methodology of evaluating the anesthetic status becomes extremely important and the analysis methods has always been the hot topic for relevant medical reserchers.Given the fact that most anesthetists judge the patients' anesthetic conditions through vital signs and their clinical medicine experience and kowledge,this protocol is kind of subjective,which could result in over-anesthetic or under-anesthetic consequence like dyscongnition and uncomfort,expecially when it comes to special individual patients' physical condition and anesthetists' bias.Therefore,it is quite crucial and urgent to explore some new objective anesthetic evaluation methods to overcome these potentical side effects,which will reduece the psychological pressure of medical staff and increase the success in operations.In the thesis,large amounts of work have been conducted based on the practical requirement of anesthetic status monitoring.A multi-modal physiological signal acquisition system has been established.From the perspective of multi-kinds of physiolgical signals,extensive efforts are undertaken to investigate the signal feature difference and variance induced by anesthesia.Then,modern nonlinear and nonstationary signal analysis methods were employed to study the electrocardiograph(ECG)heart rate variability(HRV)similarity distribution patterns and electroencephalogram(EEG)time domain complexity characteristics,which is followed by index regression through machine learning to evaluate the anesthetic status accurately.Besides,Hilber Huang transform(HHT)is used to carry out the EEG time-frequency analysis and is compared with the Multitaper methods.Moreover,cross frequency coupling(CFC)is developed to study the functional connectivity between different specific areas based on ECoG sginal of macaque monkey.This is aimed at providing new insight into the anesthesia study,which may open new window for the investigating the depth of anesthesia.The details of contents are as follows:(1)Considering the disadvantages of standlone data collection systems,a new proposed WIFI wireless automatic data collection system is set up.This updated system is capable of collecting kinds of signals like EEG,ECG and blood pressure(BP),etc.It also consists of the physiologcial signal database.To sum up,it significantly improves the data collection efficiency and provides the huge amount of data source for future analysis.(2)Given the fact that HRV cannot evaluate the depth of anesthesia directly,the proposed methods proceed further based on HRV.After extracting the ECG R wave to R wave interval and quantitification of R interval histgram distribution pattern,a index named HRV distribution similarity index(SDI)has been calculated to distinguish the significant difference between different anesthetic states.Then,artificial neural network(ANN)is used to fit the index from data of 113 cases.The results proved the the SDI superiority in comparison with comercial index BIS.Our methods provides a more accurate method to montior the depth of anesthesia,which overcome the high cost of EEG derived devices and provides another potentially feasible solution to study the evaluation of depth of anesthesia.(3)A self-adaptive method called empirical mode decomposition(EMD)was used to decompose signal to sovle the contamination of EEG.Sample Entropy(SampEn)is then applied the filterred data to show different brain activity of brain in different anesthetic states.Compared with Permutation entropy(PeEn)and recurrence quantification analysis(RQA),statistical resutls verified the advantages of SampEn to correlate the anesthesia.Then,SampEn was as input of three regression models: ANN,support vector machine(SVM)and random forest(RF).By evaluating their performance,a proposed method based on EEG SampEn and RF has been determined to evaluate depth of anesthesia.(4)For the issues that mainstream methods cannot evaluate the degree of conciousness for all rutoine medicnes universally,HHT is employed to study the difference from new apsect of time-frequency patterns of EEG induced by propofol and desflurane.By comparing with the traditional Fourier based Multitaper methods used for time-frequency analysis,results favors the HHT from both simulations and real EEG data.Hence,HHT is used to investigate the time frequency patterns of EEG under different anesthetic states,thus providing new inspiration for us to use the drug specific pattern to analyze the depth of anesthesia.(5)To study the brain connectivity for different area under anesthesia,128 channel ECoG data of macaque monkey has been used as the data source.Spectral coherence was used to measuring the spectral correlation of four channels.Phase amplitude coupling(PAC),a kind of CFC was employed to calculate the coupling between phase of slow wave and amplitude of Beta wave for these four specific channels,which is used to be as a measure of functional connectivity between brain areas.This part proves the patterns of PAC are much relevant to anesthesia,which reflects the relationship between phase and amplitude of neural potential signal.This could be useful to explain the brain function mechanism under anesthesia,which will supply another implication for the anesthetic drugs function mechanism.