Research On The Equivalentanalysis Model And Extracting Features Of The Heart Sound Signal

The process of heartbeat is very complex which is including electrical conductivity of biological, myocardial contractility and so on. The vibration signal generated by heart beat reaches to the body surface through the system of chest, and then forms the heart sound signal. The heart sound signal is one of the most important physiological signals and it is a typical tine-varying signal. The heart sound signal contains lots of information because of the time-varying characteristics. Noise and distortion of the heart sound cycle is an important basis for the diagnosis. The first heart sound and the second heart sound are considered to be the vibration signals which are generated by the collision in the process of value opening and closing. All of the vibration signals contain many kinds of transient non-steady ingredients. The changing of heart sound signal often means that different values or myocardial have pathological changes.In order to study the generation mechanism of the heart sound better, to grasp the reason of the nature of pathological signal, this thesis considers the vibration signal generated by the value opening and closing mainly. A spring-mass-damper system is used to simulate how the heart sound generates. The stretching of spring simulates the vibration of the values. The synthesis of each value vibration signals is called inner heart sound which has high similarity with the actual heart sound signal. But the system also needs further perfection.Modeling the heart sound signal accurately is the basis of heart sound encoding, compression and recognitions. Time-frequency analysis theory and method can analyze the signals'characteristics in time domain and frequency domain. They are a powerful tool for analyzing non-stationary signals and has become a hot research topic in the filed of signal processing in recent years. It also provides a new theory and methods for modeling heart sound. This paper uses sine dictionary and wavelet dictionary to model the heart sound based on the matching pursuit algorithm for the actual heart sound. Firstly, Using MP algorithm to decompose the heart sounds signal; then using the optimal atoms to compose the original signal. The experimental results show that using the wavelet dictionary to model the actual heart sound is better than using the sine dictionary.The key point of analyzing the heart sound is extracting the independent cardiac cycle, because abnormal signals are always present in each cardiac cycle. According to the characteristics of heart sound, the heart sound signal is considered as shock signal. The first heart sound and the second heart sound are the shock component. This thesis uses relative structural complexity to extract the heart sound based on the matching pursuit algorithm. Firstly, calculating the heart sound's complexity; then extracting the effective information from the curve of the complexity and the original signal. The experimental results show that this method can extract the active ingredients for the heart sound signal successfully and get some medical indicators, for example, heart rate, the first heart sound amplitude, the second heart sound amplitude, diastole, systole and so on.