Time Irreversibility Analysis Of EEG And Coupling Analysis Of Multivariate Bioelectricity Signal

In order to accuratly predict and diagnose health of body organs, effective way to collect andanalyze the biological electrical signals is needed. This paper put forward two new biologicalelectrical signal analysis methods for getting more effective analysis on the irreversibility andcoupling degree of biological electrical signals. What’s more, it was applied in practice. Thisthesis was mainly focused on the following three aspects:Firstly, time irreversibility of electroencephalagram signals were analyzed.In this thesis, we showed that the trajectories of electroencephalogram (EEG) possesses thecharacter of time reversal asymmetry, which could provide information about the entropyproduction of EEG. We developed a new method to estimate symbolic relative entropy whoseentropy production using forward and backward trajectories. Then we used this method to disposeand analyze the EEG of younger and elder subjects. It was turned out this method working and theaverage energy dissipation could be used as a parameter to detect nonequilibrium.Secondly, abnomal epilepsy electroencephalagram signals were analyzed based on symbolicrelative entropy.Symbolic relative entropy algorithm was used in the analysis of abnomal epilepsyelectroencephalogram. It showed that abnomal epilepsy electroencephalogram had different averageenergy dissipation from normal EEG’s. Therefore, the average energy dissipation could be used as aparameter to judge if the EEG signal was normal.Thirdly, the coupling of multivariate bioelectricity signals was analyzed.Symbolic partial mutual information was proposed in this paper, which was based on partialmutual information. This algorithm could be used to analyze the coupling between multivariate timeseries. We used this method to dispose and analyze the sleeping MBS (multivariate bioelectricitysignal) and waking one. it could be reached that the coupling of waking MBS was obviously largerthan sleeping one’s. Finally statistical hypothesis testing was done to prove this method workingand the average energy dissipation could be used as a parameter to detect nonequilibrium.