Nonlinear Analysis Of Dynamic Characteristic Of Neural Mass Model

Electroencephalogram(EEG)signal is a typical nonlinear signal with high instability and complexity,which contains a lot of the brain’s physiological and pathological information.In order to fully understand the cognitive function and effectively detect abnormal brain electrical activity,various nonlinear dynamic analysis methods are applied to the study of EEG signals.Compared to tradition linear analysis methods,the nonlinear dynamic analysis methods can extract more effective feature of EEG signals and provide help for the diagnosis of brain disease and identification of different brain activity states.In this study,a variety of nonlinear dynamic analysis methods are applied to research the different kinds of EEG signals produced from neural mass model to verify the validity of the analysis methods and provide theoretical basis for the diagnosis of brain disease and identification of different brain activity state.The specific work is as follows:Firstly,the modified permutation-entropy algorithm is applied to extract and analyze quantitatively the characteristics of complexity of the normal EEG signal,sporadic epileptic spikes and sustained discharges of spikes produced from single neural mass model and compared with original permutation-entropy and Rényi permutation-entropy to verify the effectiveness of modified permutation-entropy in the analysis of EEG signals.Secondly,neural network model with eight nodes is constructed based on the concept of graph theory and complex network.The characteristics of complexity of eight channels EEG signals are extracted by using the modified permutation-entropy algorithm and analyzed to study the influence of different network parameters for dynamic characteristic.Finally,the analysis method of Lorenz plot is applied to analyze the EEG signals with different rhythm produced from multi-kinetics neural mass model and provide the basis for identification of different brain activity state.