Research Of Methods For R Wave Detection Of ECG And Heart Rate Variability (HRV) Based On Multi-scale Time-Frequency Analysis

Heart Rate Variability (HRV) based on RR interval is a nontraumatic diagnosis method developed in the recent 20 years for assessing the function of autonomic nervous system of heart, which has found application in scientific research and clinic. The first phase for the analysis of HRV is to check and locate the R wave precisely. The wavelet analysis has been applied in R wave detection effectively to display the advances of multi-scale analysis to check such non-stationary signals as ECG. In Chapter Three, the principle to detect odd signals by wavelet is showed, in which the position of odd point in a signal is determined by corresponding relation between those extremums of its wavelet transform and odd points. And the relative flow of methods for checking R wave by first rank differential wavelet and second rank differential wavelet is given, and developing orientation is discussed.In Chapter Four the further principle of Experimental Mode Decomposition (EMD) is illustrated, which is characterized by multi-scale analysis the same as wavelet and by self-adaptation better than wavelet. EMD has been applied in earthquake, water wave and mechanical fault detection with good results, but no much report about its application in R wave detection is found. And the possibility and efficiency of a way by EMD to check the position of R wave in an ECG is discussed. According to the analysis of the figures, the way is effective to depress and relieve the noises from the power, muscle EM and respiration, and separate the R wave from the ECG with aberrances by diseases. The EMD is applied to detect R wave in ECG signals from MIT/HIB databases and the accurate rate is relatively high.With accurate R wave position, analysis of HRV based on RR intervals is possible. HRV could reveal and assess the activity of the autonomic nervous system of heart and its balance, which is a worthy index to predicate or judge the heart diseases. It also has been used to assess the public health and monitor the condition of athletes, pilot and astronauts. Chapter 5 indicates that multi-scale time-frequency analysis such as wavelet and EMD could offer more information of HRV than traditional linear analysis ways in time domain and frequency domain, which is a research trend with those nonlinear ways. Especially the Hilbert spectrum and partial Hilbert spectrum from EMD demonstrate better resolution than common time-frequency analysis tools, and are proper to analyze the HRV.