Investigation Of Heart Rate Variability Based On Time-frequency Analysis

The analysis of heart rate variability (HRV) has become an important tool for non-invasively detecting the cardiovascular modulation of autonomic nervous system. HRV refers to the tiny fluctuations of instantaneous heart rate between consecutive beats or the minor differences of the successive cardiac cycles. It contains a lot of information concerning the cardiovascular regulation, while extracting and analyzing of the information of HRV can offer a quantitative evaluation of the tenseness and the balance of the sympathetic and the vagus nerve activities and their effect on cardiovascular motion. Clinical medicine has given the evidence that investigating HRV has an important meaning not only for the early diagnosing cardiovascular diseases, monitoring during convalescence and evaluating after recuperation, but also for the quantitatively valuing the cardiac functions during physical exercises or in other special environments.The reasons of fluctuations of heart rate are complicated. The influence of sympathetic and parasympathetic nerve caused by various factors can all lead to its oscillations. Hence, the frequency components contained in HRV are never invariable, but time-variant. In other words, the signal of HRV is a nonstationary signal, which offers the possibility to apply the theory of joint time-frequency analysis to deal with HRV, extract its time-frequency characteristic parameters and explore the relation between these indices and various cardiovascular diseases.After reviewing the current analysis methods of HRV in home and abroad, the joint time-frequency methods which are suitable to study HRV are discussed, investigated and developed in the thesis, and their theoretical background is joint time-frequency analysis. The main works and conclusions of the thesis are as follows.(1) Investigation and comparison of the traditional methods of HRV in time-domain and frequency-domain analysis are made, and their corresponding quantitative parameters and psychological meanings are also discussed. A deeper understanding of the nature of the methods has shown that the nonstationarity of HRV is not taken into account in them, and furthermore the parameters can only describe the total properties of HRV throughout the whole recording, but not reflect the local and transient episodes due to the changes of HRV caused by various factors during the recording.(2) Detecting the heart rate and its fluctuations is carried out before analysis. Thebasic procedure is described below. First an electrocardiogram (ECG) is measured and filtered, and then the RR interval time series are exported by detecting the QRS complex and eliminating the non-sinus rhythm. The preprocessing of the raw data is the basic, even the key point of signal analyzing, because minor differences in this procedure can lead to disparate results. During ECG noise reducing, the technique of adaptive signal processing is employed to remove the power interference and base wander. The algorithms of QRS complex detecting is based on amplitude and first derivative of ECG The RR interval series is interpolated by cubic spline before being evenly resampled.(3) The time-frequency methodological approaches to deal with the uniformly resampled HRV signal, mainly based on wavelet transform and spectrogram, are presented. The innovative point of the method based on wavelet transform not only can obtain the traditional quantitative parameters, but also can calculate their dynamic value varying with time, called short-time power and short-time LF/HF ratio, which can trace the sympatho-vagal balance and their modulation of heart rate during the drag injection. Decomposing HRV based on wavelet transform and independent component analysis is also the brilliant aspect in the context. The psychological background of this decomposition is that HRV is modulated by the factors of nervous system (sympathetic and parasympathetic parts) and non-nervous system. Finally, 120 HRV data (including 20 normal subjects, 40 patients of coronary heart disease and 60 hypertensive pa...