Nonlinear Analysis Of Diastolic Murmurs For Coronary Artery Stenosis

Coronary artery disease (CAD) has become one of the diseases which serious threat to healthof people. Due to its high incidence, high morbidity and high mortality, it not only seriously affectsthe health and quality of life of individuals, but also brings the country and society with heavyburden. Therefore, early detection, early diagnosis and early treatment of coronary heart disease isvery important. Coronary artery disease is caused by coronary artery stenosis which leads to cardiacinsufficiency, causing myocardial dysfunction and organic disease. Heart sound is one of the mostimportant physiological signals, which contains physiological and pathological information ofcardiovascular system. When coronary arteries become narrowed or blocked, the turbulence appearswhich is produced by blood moving across the stenotic arteries, and then diastolic murmurs appearsin the heart sound signals. Therefore, nonlinear analysis for diastolic heart sound is greatsignificance to achieve early detection, early diagnosis and early treatment of coronary heartdisease.This thesis mainly analyzes the diastolic heart sounds and the feature extraction algorithmbased on nonlinear analysis. Nonlinear analysis of diastolic heart sound signals is carried out inorder to extract effective pathological features of CAD. Firstly, S transform is used to analyze heartsound signal on time-frequency domain, and then the algorithm of nonlinear analysis based on Stransform and Renyi entropy is proposed. Secondly, removing trend term by EMD, and then thealgorithm of nonlinear analysis based on EMD and correlation dimension is proposed. Finally, thealgorithm performance has been proved by clinical heart sound signals.The main works of the paper are list as follow:(1) The algorithm of nonlinear analysis is proposed based on S transform and Renyi entropy.Firstly, the theories of S transform and Renyi entropy are studied, and then Renyi entropyparameters are selected. Secondly, the diastolic heart sound signals are analyzed by S transform,Renyi entropy is calculated based on time-frequency domain for diastolic heart sound signal. Finally,the analysis of clinical diastolic heart sounds based on Renyi entropy and S transform. The resultsshowed that Renyi entropy of two types signals has a significant difference, and Renyi entropy caneffectively reveal the increasing complexity of the heart sound signal which caused by coronaryartery stenosis.(2) The algorithm of nonlinear analysis is proposed based on EMD and correlation dimension.Firstly, the theories of EMD and correlation dimension are studied, and then the correlation dimension parameters are selected. Secondly, removing the trend in diastolic heart sound signal andthe new signal is reconstructed, and then the correlation dimension of the reconstructed signal iscalculated. Finally, the analysis of clinical diastolic heart sounds based on correlation dimensionand EMD. Experimental results show that correlation dimension of two types of heart sound signalhas significant differences, and the correlation dimension can be as an indicator which fornon-destructive diagnosis of coronary heart disease, and reveal the increasing complexity of theheart sound signal which caused by coronary artery stenosis.This paper researches the extraction of pathological information in diastolic heart sound signal.It provides an effective new way for non-invasive diagnosis for coronary artery disease andenhances the application potential of heart sound in clinical.