Research On Feature Extraction And Classification Of Heart Sound Signal

Valvular heart disease has become one of harmful heart disease to human health.Detection of heart sounds is an important method with judging the state of heart in terms of heart disease diagnosis。 Heart sound is a common measure, it is as important as ECG. It reflects the mechanical activity.Auscultation of the heart sounds mainly rely on the accuracy of the equipment and the doctor’s subjective judgment.(1)Signal pretreatment:Aiming at the the characteristics of the heart sound, Adaptive decomposition method is applied to analyze the heart sound signals. Using the EEMD and LMD method respectively to decompose three types of state of heart sound, To obtain a series of IMFs and PFs. By the correlation coefficient and the kurtosis value calculation to screening the number of components.(2)Feature extraction:We extract the heart sound signal feature based on SVD and Energy Entropy. Singular value can reflect the algebra character of the heart sound effectively, By SVD of the IMF components to form singular value feature vector. The energy entropy is a measure of the nonlinear characteristics complexity, By energy entropy of PF components calculation to form energy value feature vector.(3)Classification method:Two classification methods are brought forward according to the two different feature extraction methods.The minimum distance method is used to classification,energy value feature vectors is used to classification by Least square support vector machine...