K-Nearest Network Analysis And Symbols Transfer Entropy Analysis Based On Template Method Of Electroencephalogram Rhythm Signal

The brain is a complex information system and cerebral cortex contains rich neurons.The brain electrical signal is a spontaneous electrical activity of the central nervous system which is generated on neural activity in the cerebral cortex and contains abundant physiological and pathological information of brain activity.It plays an important role in clinical brain disease diagnosis.In this thesis,the method of k-nearest neighbor network and the method of symbols transfer entropy based on template method are used to analyze electroencephalogram rhythm signal(EEG rhythm),besides,the thesis proposes multiscale analysis on symbols transfer entropy based on template method which provide basis for Clinical diagnosis of epilepsy and dynamic information capture of EEG.The main work of this paper is as follows:Firstly,epilepsy EEG rhythm analysis based on the k-nearest neighbor unoriented networkThis thesis applied k-nearest neighbor unoriented network to analyze epileptic EEG-δ rhythm.The method starts by reconstructing phase space from single lead epileptic EEG-δ rhythm and building the EEG-δ rhythm into a network with the method of k-nearest neighbor unoriented network,then converts it into time series and finally compares the power spectrum with the one of original EEG rhythm.The results show that the power spectrum after network transform is easier to distinguish between epileptic and normal person than the power spectrum of the original EEG rhythm.In addition,through the analysis of the concentration coefficient of the network,it can also distinguish between normal and epilepsy.The study can provide positive guidance for clinical diagnosis of epilepsy.Secondly,symbols transfer entropy analysis of EEG rhythm based on improved template methodThis thesis proposes a method of improved template to analyze adults and adolescents transfer entropy of EEG-δ rhythm,EEG-θ rhythm,EEG-α rhythm,and EEG-β rhythm on the basis of the conventional template method.Experimental results show that symbols transfer entropy based on improved template method of EEG-δ rhythm or EEG-θ rhythm is better than conventional method to distinguish between adolesents and adults EEG.Besides,entropy of adults are larger than adolescents,and the improved symbols entropy is bigger than conventional one,namely,symbols transfer entropy based on improved template is more able to reflect chaotic degree between the EEG rhythm.The research result is useful to the complex dynamic information on the brain.Thirdly,multiscale symbols transfer entropy analysis of EEG rhythm based on template methodThe thesis uses multiscale method to analyze symbols transfer entropy of adults and adolescents EEG-δ rhythm and EEG-θ rhythm on the basis of template method.The results show that the appropriate choice of the scale factor can improve the degree of differentiation of different physiological or pathological characteristic EEG rhythm,even provide the basis for clinical diagnosis.