Detection And Classification Of Heart Rate Variability Signals

The acquisition of heart rate variability (HRV) signals is very important to clinical diagnosis and physiological research. The obtainment of HRV signals is a pivotal problem. It needs to think about the methods of getting HRV signals carefully to ensure the accuracy of the HRV analysis. If the choice is unsuitable, the basisi of the analysis is not steady and the veracity of analysis has no way to talk about. By the definement of HRV, the practical HRV signals are the R-R interval series by dealing with the ECG signals. So the sticking point of analying HRV signals is the acquirement of the R waves.The ECG signals which getting from clinic is one kind of the non-stationary signals. The time-frequency analysis, as a basic method, alaways plays a great role in analyzing the non-stationary signals, which involved the signals analysis, modeling, compounding, reconstructing, filtering, pattern recognition and detecting.The wavelet transform (WT) and the neural network (NN) have made a great progress as a new tool to analyze the signals.The WT has the characteristics of Multi-resolution and the auto-focusing.So the WT was named as"microcope" of the informations.The NN has many adventages, for example, the non-linearity, high degree of accuracy, study by itself.etc. So the WT and the NN have a broad perspective used in biomedicine.In this paper, after studying the WT and the NN, the WT was used in the pretreatment of the ECG signals.The method of unfixed threshold combined with a'trous algorithm of stationary wavelet transform can keep the most of the QRS's characteric points. By the detection method of changing terms combined the characteristic of the value of R wave relative amplitude is bigger than others to deal with the HRV signals and find the situations of R wave. Then work out the series of R-R intervals and use the form of wave to reflect the changing of the HRV intervals. The neural network as a new tool was used widely in the...