Detection of Mechanical Prosthetic Heart Valve Dysfunction Using Spectrum Estimation and Time-Scale Techniques

The diagnostic features in recorded heart sounds with mechanical prosthetic heart valves (either mitral or aortic heart valve) were investigated by using spectral estimation and discrete wavelet transform (DWT) techniques. Results showed that for patients with mechanical prosthetic heart valve, it is possible to identify the normal patients, the perivalvular leakage patients and the valve obstruction patients via combined spectral and time-scale features, and experiments also demonstrated that neither spectral feature nor time-scale feature separately could solve the identification problem.