Application Of Chaos Theory On ECG Data Analysis

Study of Electrocardiogram (ECG) data can help us to understand the physiological phenomena and pathological mechanism of activity of disease. To attain the goal of clinical application, feasibility and suitability of chaos dynamics analysis method for ECG is discussed by analyzing of chaos characteristic of ECG and R-R intervals sequences. By this way, it is proposed to evaluate the cardiac activity and to find some new means and measures offering further earlier clinical diagnose.ECG data of healthy people and three kinds of heart disease patients (premature ventricular contraction, bundle branch block and supraventricular) which are analyzed in this thesis are derived from MIT-BIH ECG database. After summarizing and researching the principle of chaos dynamics, especially based on reckoning chaos dynamics parameters of time sequence, the following analysis technique such as power spectrum analysis, phase space reconstruction, correlation dimension, largest Lyapunov exponent and R-R intervals Poincare plot are used to research the chaos dynamics system and activity of ECG signals and R-R intervals sequences . Thereby it comes to a series of significant conclusions.1) The power spectrum of ECG signals and R-R intervals sequences are all continuous. This indicates that heart system signals are not simple stochastic signals, not period signals. 2) During the process of phase space reconstruction, delay times are reckoned with autocorrelation function method and C-C method. Comparing the result and validating them in two-dimensional phase space graphs, it concludes that C-C method is better than autocorrelation function method. And C-C method can reckon embedding dimension at the same time. 3) Correlation dimension of ECG signals and R-R intervals sequences are reckoned with GP method. The results are all non-integer. It means that heart system is chaotic. And the correlation dimension of R-R intervals sequences prefer to ECG signals can reflect pathological status of heart system accurately. 4) The largest Lyapunov exponents of ECG signals and R-R...