| Cardiechema and respiratory murmur are important indexes which reflect physiology and pathology of heart and lung. Nowadays acouophonia of the cardiechema and respiratory murmur is still the main diagnosis method. However, due to traditional stethoscope have some difficulties in capturing faint physiological sound and the diagnosis results is liable to be influenced by the subjective experience of the auscultator, so the veracity of diagnosis is not unearthly perfect.Thus, in order to help the clinical doctor to judge the patients'condition more effectively according to the cardiechema diagnosis, it is necessary to establish a cardiechema analytical and discriminate system which can record, store, show, compare portrait and lateral powerful and accurately. It is hopeful to apply the simple auscultation skill to discover some disease at early state and the expense will be sharply declined if in-depth analysis can dig out the pathological changes concluded in the cardiechema, so the digital stethoscope is prospective.This diploma project proposes that to apply the digital signal processing algorithm such as Hilbert-Huang Transformation in analyzing and processing the acoustical signal, expressing the time-frequency characteristics of sound in graphic forms, transforming auscultation into visual examination, employing Empirical Mode Decomposition in order to extract a new kind of characteristic parameter to be applied by pattern recognition, utilizing neural network in order to realize preliminary automatic diagnosis.The paper introduces time-frequency analysis technology and algorithm of the neural network in the contemporary digital signal processing as well as the whole completion in the cardiechema and respiratory murmur such as the signal acquisition, digital filtering, extraction of signal envelop, automatic segmentation, time-frequency analysis, neurotic network pattern identification and so on. |
