Application of wavelet analysis to the phonocardiographic signal of mechanical heart valve closing sounds.

Heart valve disorders, caused by congenital defects, rheumatic fever, calcification, myocardial infarction and other cardiovascular diseases, often require native valves to be replaced by bio-prosthetic devices or mechanical heart valves (MHVs). Among MHVs, bileaflet valves are usually preferred for their hemodynamic features, similar to physiological ones, and their durability, but they are prone to complications due to thromboembolic events. Due to the asynchronous closure of the leaflets, bileaflet MHVs are also known to produce closing sounds typically characterized by the presence of two peaks in the time domain. The detection of this "double click" in the signal may be useful for the early diagnosis of bileaflet MHV malfunction. The closing sound is actually a non-stationary signal that can be properly explored by means of time-frequency analysis. This paper describes a preliminary approach to the investigation of bileaflet MHV closing sounds performed by Continuous Wavelet Transform (CWT) analysis. Signals were collected from 3 patients immediately after surgery by means of the Myotis 3C, which is a traditional phonocardiographic apparatus. Signals were analyzed by two algorithms: one embedded in the Myotis 3C, based on the Fast Fourier Transform (FFT); and one specifically created for the purposes of the present study, based on CWT. The performance of these algorithms was compared and the results showed that the proposed CWT technique correctly classifies as ''double'' a large number of clicks that are recognized as ''single'' by the Myotis 3C.