The Research Of Diagnosis Of Heart Disease Method Of Heart Sound Signal

Heart sounds signal is one important physiological signal of human body. Itcontains physiological pathology information of every part of the heart. To improvecardiovascular disease diagnosis ability and diagnose rate are a very importantsignificance for heart sounds signal collection and analysis research. Heart soundssignal is a time-varying, non-stationary, weak, complex signal, so heart sounds signalfeature extraction, classification and identification is the problem of the study.This paper can be divided four aspects, the heart sounds signal collection andheart sounds denoising, composition analysis, feature extraction, features divisibilityof heart sounds signal is analyzed.Combine with LabVIEW8.6virtual instrument development platform to realizecollection procedures of heart sounds signal acquisition. In this acquisition process,we can realize to play the real-time waveform signal heart sounds to improveacquisition efficiency; Use wavelet transform method to eliminate the noiseinterference of the collection process and a good de-noising result was obtainedDesign the automatic segmentation algorithm of heart sounds through the heartsounds envelope extraction, envelope normalization, envelope difference can get thestarting point of heart sound signals major components based on MATLAB. The keyof Heart sounds automatic segmentation is to identify the starting point of the firstheart sounds, the first heart sounds will be as reference, to locate the first heart soundsand the second heart sounds quickly according to heart sounds signal characteristics:systolic is shorter than diastolic. According to the need of study, we also can get heartcharacteristic parameters like systole time, diastole time, cardiac cycle time and theheart rate and so on.A variety of time-frequency analysis methods are compared and analyzed in thispaper, short-time Fourier transform time-frequency resolution is low andtime-frequency distribution clustering is poor, cannot fully reflect the murmurs ofpathology; Although Wigner-Ville distribution has high time-frequency resolutionand better time-frequency distribution clustering, but it contains cross interferenceitem, it seriously influences heart sounds signal analysis. S transform givesconsideration to the advantages of wavelet transformation and Fourier transform, itnot only has good time-frequency resolution and clustering, but also does not containthe cross-terms. The application of S transformation in heart sounds signalcharacteristics analysis, not only can be fully reflected the main frequencycharacteristics of the signal, but also show all kinds of pathological murmurs, is good for the heart disease diagnosis and analysis. Finally, make use of the distance withinclass and class separation distance theory to evaluate the feature extraction algorithm,we can draw the conclusion: the extraction feature has the separability, and can beused for automatic analysis of disease diagnosis effective approach.