| Heart sound, as a vital physiological signal of human body, canreflect the mechanical movement of the heart and great vessels, and it is the basicmethod for clinical assessment of the heart function condition with the advantage ofnoninvasiveness and of convenience. Considering the life is the most complicatednonlinear dynamic system, with the heart as the core of the life cycle, which determinesthe nonlinear and complex character of the heart sound generated by the heart vibration.For a long time, researchers have simplified and abstracted the complex cardiac systemin order to build an ideal linear model, and analyzed it in terms of time domain,frequency domain, and time-frequency combination, etc. However, for half a century,the related studies with linear methods haven’t been effective enough to analyze theessentially nonlinear life activities. Since chaos, as one distinctly important movementpattern of the nonlinear systems, can well reveal the special regularity of the intrinsicrandomness of the nonlinear process, it is used to analyze the heart sound in this paper,so as to understand the inherent characteristics of heart sound more deeply from essenceand realize the computer aided diagnosis of cardiac disease based on heart sound signalwith a view from a new angle.In order to improve the recognition precision and classification accuracy, themethod combining the wavelet packet analysis and chaos theory is proposed to processthe hear sound, including feature extraction, classification and recognition. Comparedwith wavelet transform, wavelet packet has stronger time-frequency resolution whichcan extract more detail time-frequency information from the original signal. Therefore,on the one hand, the heart sound signal is analyzed with wavelet packet from the aspectof time-frequency domain, through which the heart sound signal is decomposed intovarious frequency bands and then the energy features are extracted. On the other hand,the characteristic components of the heart sound signal are separated after the waveletpacket decomposition, and then analyzed with the method of chaos qualitatively andquantitatively. And the qualitative analysis includes phase plot and recurrence plot,while the quantificational analysis involves the computation of the chaotic characteristicparameters like the correlation dimension and largest Lyapunov exponent. Then, theenergy features and the chaotic characteristic parameters are combined as thecharacteristic vector of the heart sound, and the optical ones are selected with genetic algorithm. Finally, the characteristic vector is input into support vector machine (SVM)realizing the automatic recognition of the heart sound signals.The heart sound signals are acquired through the designed signal collecting system,including the normal heart sound and abnormal kinds such as, heart sound witharrhythmia and premature beat, mitral stenosis, splitting first heart sounds, aorticinsufficiency and ventricular septal defect, and all of those heart sound signals are testedwith the proposed method in the paper. The results showed that both the qualitative andquantitative chaotic characteristics of the heart sound had significant differences, andthe correlation dimension and largest Laypunov exponent of the abnormal heart soundswere larger than those of the normal heart sounds which indicated that the abnormalones were more complicated than the normal ones. Moreover, the features of heartsound combining with the wavelet packet energy and the chaotic characteristics couldget a higher recognition rate, and this demonstrated that the chaotic characteristics had agreat significance for revealing the nonlinear character of the heart sound, which couldlay a foundation for the future research of the essential nonlinearity of the heart soundand the cardiac disease diagnosis. |
PDF Full Text Request