Analysis Of Physiological Electrical Signals Symbol Based On Multivariate Transfer Entropy

Physiological signals as important indicators of various physiological parameters in the human body, it plays an important role in the clinical diagnosis and treatment. Use of complexity theory to study the physiological signals has become a hot spot. The symbolic transfer entropy is a nonlinear system indicator to reflect the degree of chaos, which can be used as a characteristic of physiological signals. It is playing an increasingly important role in the study of physiological electrical signal feature extraction. But symbolic transfer entropy is generally used to measure the dynamic characteristics and directional information between one or two variables and ignores the interaction between multivariate.In this thesis, the main work and contributions include:Firstly, this paper adopts multivariable symbols transition entropy based on the multivariable transfer entropy. Because of the symbolic dynamics using traditional static partitioning method while retaining the dynamic characteristics but the non-stationary time series or the results have been seriously affected. So the improved traditional time series static partition method adopts dynamic adaptive segmentation.Secondly, this paper use the algorithm of multivariable symbols transfer entropy to analyze epileptic EEG and attention EEG that the state are close eyes and counting. The final multivariate symbolic transfer entropy value by choosing the best lead and the best data length proved that this algorithm can significant distinction between normal and epileptic patients, and also can distinguish EEG counting state and the eyes closed. While superimposed Gaussian noise in the original EEG signal time sequences verified the robustness of this algorithm.Thirdly, using the modified multivariable symbols transfer entropy to analysis the ECG of people both normal and patients and coronary heart disease. In the experiment by selecting the best lead on to determine the improved algorithm can significantly distinguish between normal subjects and patients with coronary heart disease. After the original sequence is superimposed on the gauss noise result that the algorithm is still reliable and effective.