Research Of Chd Heart Sound Recognition And Quantization

Heart sound signal is a compound sound of heart valves switch, tendon and muscle relaxation shrinkage, blood flow impact and cardiovascular wall vibration. The murmurs included in heart sound are useful information for diagnosis, especially for the primary diagnosis of CHD.As heart sound signal include complex and changeful cardiac physiology and pathology information, the majority of heart sound analysis and researches are limited to qualitative description at present, and quantitative analysis will need great efforts in the future. This study is named as " Research of CHD Heart Sound Recognition and Q uantization". this dissertation mainly studies on recognition and quantitative analysis of clinical CHD heart sounds. Main works are introduced in detail from the following aspects:(1) Be familiar with basic information of heart sound auscultation, grasp the theory of heart sound signal generation and the relationship between heart murmur and heart diseases; summarize the researches on heart sound in home and overseas, understand development of heart sound research, and find out main problems in this field.(2) Heart sound signal acquisition and pre-processing: The digital stethoscope and heart sound acquisition system based on BIOPAC that used in HS data acquisition is introduced. And the functions and main parameters of the system are described in detail. The HS data is pre-processed in appropriate way, for later feature extracting and analysis.(3) Heart sound recognition system research: Mainly introduce a heart sound recognition system based on MFCC parameters and LPMCC as heart sound identification parameters, and using VQ and DTW algorithm as the identification model, the system has realized a well recognition on normal heart sounds with common CHD heart sounds.(4) Heart sound quantitative system research: Aiming at common clinical C HD. This study proposes a clinical CHD defect aperture quantitative method, achieve t he goal of the clinical heart sound quantitative analysis preliminarily.This specific following contains:1. Pathology information of several common clinical C HD are introduced, and according to the actual auscultation knowledge, heart sound signal from suitable collection position is chosen for the analysis in the next step.2. Heart sound preprocessing. Based on the time continuity and short-time stationary characteristics of HS, in this thesis the heart sound need a series of processing include pre-emphasis and windowing after wavelet threshold de-noising.3. The system of CHD heart sound recognition. The system of CHD heart sound recognition include feature extraction and heart sound recognition, on the stage of feature extraction, the thesis introduced the voice recognition parameters: MFCC and improved LPCC: LPMCC parameters, on the stage of heart sound recognition, a recognition model based on the mixture of DTW and VQ was proposed, and achieves a good recognition effect.4. The CHD defect aperture size quantitative analysis based on wavelet spectrum entropy. In this thesis, on the basis of wavelet decomposition reconstruction, the wavelet energy spectrum entropy under different frequency bands was extracted, then analysis and compare the correlation with the defect aperture size, which can extract the quantitative evaluation method of common CHD heart sounds defect aperture size.In conclusion: This dissertation researches on clinical C HD heart sound recognition and quantitative system. And validated by clinical heart sound data, experiment show that this recognition system can recognize well on normal heart sounds and common CHD heart sounds, The quantitative method based on wavelet spectrum entropy can represent the CHD defect aperture size as well.