Font Size: a A A

The Study Of Heart Sound Automatic Analysis And Recognition Algorithm

Posted on:2004-06-18Degree:MasterType:Thesis
Country:ChinaCandidate:J ChenFull Text:PDF
GTID:2144360095956863Subject:Biomedical engineering
Abstract/Summary:PDF Full Text Request
Cardiac auscultation is widely used to evaluate cardiac function in patients. Since heart sounds are generally of low frequency component and very faint, this method depending on doctor's knowledge and experience is subjective, and sometimes is unreliable and ineffective. Recently it has been shown that the phonocariogram can be used for cardiac performance assessment. Specifically, it can be demonstrated that changes in the amplitude of the first heart sound correlate with changes in the maximum rate of rise of left ventricular pressure, a common measure of cardiac contractility. Thus the cardiac contractility can be measured by heart sound. But general phonocardiogram analysis instrument only can monitor the static patient's short-time heart sound, cannot monitor the patient's long-time heart sound in various states. Therefore, development of phonocardiogram analysis instrument is very important to study the regularity of real-time heart sound and analyze the cardiac contractility variability in various states.This thesis put forward a heart sound detection scheme based on bluetooh technique. Using its characteristic of short-range radio link, the data of heart sound was dynamically transmitted to computer and was processed in real-time. Before any automatic analysis can be done, the heart sound needs to be recognized and various components need to be localized. For heart sound signal, one of the major problems is noise corruption. Using the common method to recognize, the accuracy is not enough. This study put forward a heart sound envelope detection method based on the mathematical morphology. Then on the heart sound envelope, S1 and S2 were recognized. The test results proved this method effectively restrained the interference of noise and enhanced the accuracy of recognition. At last, the extracted relative S1 amplitude was used to evaluate the cardiac contractility variability.From various sources, eighty heart sound samples collected were tested using the algorithm. The accuracy of recognition was up to 86%, specially, for the normal heart sound, the accuracy reached 100%. The result shows that the algorithm proposed in this paper has high performance. The foundation for farther analysis of cardiac contractility variability was established.
Keywords/Search Tags:Heart Sound, Bluetooth Technique, Heart Sound Recognition, Mathematical Morphology, Cardiac Contractility Variability
PDF Full Text Request
Related items