Research Of Cardiac Reserve Parameters Extraction Algorithms Based On Heart Sound

Cardiac reserve (CR) refers to one’s ability to improve the heart function, CR isimportant to measure the heart function. Through detection, evaluation and research thecardiac reserve, the doctors and patients can be aware of the impact of heart disease to one’sheart, you can analyze the cardiac reserve differences between the various groups or the stateof cardiac function in stress situations. At present, the cardiac reserve detection techniquesmainly include echocardiographidetection, electrocardiogram (ECG) detection, heart sounddetection, etc. The method of echocardiography is not sensitive to detect some cardiacfunction indicators, so this method is not convenient to be used. Now the method of ECGanalysis is the best way to detect the heart chronotropic function and conductive deficiency,however it cannot be used to detect the heart inotropic. Heart sound (HS) signals containabundant and important physiological and pathological information of human body. It hasproved that the amplitude of the first heart sound in the cardiac cycle is the standard measureof myocardial contraction ability. Because of this advantage of heart sound signals, they canbe taken to detect the heart inotropic. In time and frequency domain, the heart sound signalhas its own certain characteristics. However, due to the influence of physiological factor, thenoise outside and the acquisition environment, the clinical heart sound signals are extremelycomplicated, which is difficult to be extracted the cardiac reserve parameters. So it isnecessary to further analyze the heart sound signals by using the effective algorithms, so as tointelligently, conveniently, noninvasively and quantitatively extract the cardiac reserveparameters.The title of this paper is the study of algorithms using for cardiac reserve characteristicbased on the heart sound signals. The focus of this paper is to extract the heart soundcharacteristic parameters using the scientific, accurate and effective heart sound analysisalgorithms, and then to detect and evaluate the heart reserve, which is of great significance forthe cardiac reserve noninvasive detecting promotion and popularization. In order to completethe above goals, this paper mainly did the research from the following several aspects.(1) Heart sound signal acquisitionThe heart sound signals acquisition devices are made up by a stethoscope developed byour laboratory and Yamaguchi University in Japan, a sixteen channels heart sound recorder,an amplifier and a laptop computer. This acquisition system can collect and show thereal-time and high quality of heart sound signal. Also it can prevent heart sound signals fromoutside noise. In our laboratory, there were more than800cases of heart sound signalscollected by this acquisition system. Among these cases,200cases were collected from100volunteers whose health was good,600cases were collected from the patients with the heart disease,30cases were collected from the15high-level athletes in Xihua University. In thispaper, the cases of heart sound signals were used to analyze and to test the effectiveness of thealgorithms proposed in this paper.(2) Heart sound signal pre-processingThe heart sound signal is inevitably mixed with outside noise during being acquired,which brings lots of trouble to analyze the heart sound. According to the source of noise andthe noise own characteristics, the noise is separately named as the random noise,50Hzpower-frequency noise and the instrument noise. This paper respectively chose the bestmethods for different noise. A high-pass filter, a trap power frequency filter and a waveletthreshold de-noising filter were separately designed to improve the the ratio of heart soundsignal and noise (SNR). It achieved the ideal effective result to de-noise the noise mixed inthe heart sound by testing the standard heart sound data and clinical heart sound data.(3) The research of envelope extraction algorithmThe original heart sound signal collected often contains abundant of information.However part of those makes no sense to us. The method of extracting the envelope is toeffectively reduce the dimension of heart sound signal, and to extract the representativecharacteristic paramerts from the heart sound signal. In this paper, several common-usedenvelope extraction methods were analyzed and contrasted, they were respectivelyHilbert-Huang transform (HHT), normalized shannon energy method, the cardiac soundcharacteristic waveform algorithm (CSCW) and homomorphic filtering envelope extractionmethod. The above several methods were used to extract the envelope of heart sound signal.At last, it could obtain the advantages and disadvantages of each envelope extractionalgorithm by testing the clinical heart sound signals. In practice, in order to achieve theadaptiveness, one can choose different envelop extraction algorithms according to thedifferent heart sound signals.(4) The selection of threshold line and characteristic parameters extractionHeart sound signals characteristic parameters especially time-domain characteristicparameters such as the duration time of the first heart sound and second heart sound time aredetermined by envelop threshold line. Different threshold lines can obtain different groupsof heart sound signals characteristic parameters. So the next problem is to select whichthreshold line for obtaining the representative heart sound signals characteristic parameters. Itis sensitive to select the heart sound threshold line, because it varies with individualdifferences and the different conditions of the difference of heart disease. At previous, thethreshold line was selected manually according to people experience, which broughtinconvenience to the inexperienced. This paper adopted the fuzzy c-means clustering (FCM)algorithm which was in field of the fuzzy mathematics to select the threshold line and extract the effective characteristic parameters. The characteristic parameters extracted by FCMalgorithm were used to identify the normal heart sound signals and abnormal heart soundsignals. And the rate of identification was higher than that obtained through the characteristicparameters extracted by manully selecting the threshold line, and it also proved theeffectiveness of FCM to select the threshold line and extract the characteristic parameters.(5) Evaluation the cardiac reserve parametersAt the end of this paper, the cardiac reserve evaluation system based on heart soundsignals was designed by MATLAB GUI. The system was visual and could avoid the tediouscode. And it could realize the intelligent analysis of heart sound signals and then extractedand evaluated the cardiac reserve parameters by simple manual operation. In this paper, thecardiac reserve indicators which have been proved that they can be used to evaluate thecardiac function are heart rate(HR), the rate of first heart sound amplitude and the secondheart sound amplitude(S1/S2), the rate of diastolic and systolic limit times(D/S). It couldobtained which scope the three indicators belonged to according the medical recongnizedthree indicators scopes. The clinical heart sound signals evaluation result conformed to thegatherers’ body quality and heart sound fuction.The analysis method of heart sound was verified effective by testing the clinical heartsound, it obtained more satisfactory results. But in this paper the number of heart soundapplied to analyze was relatively smaller. So, the future work is mainly focused on collectingmore types of clinical heart sound, optimizing the heart sound analysis methods, strugglingfor extracting more accurate parameters using for detecting the heart function.