In vitro investigations on prosthetic valve thrombosis by ultrasound phonocardiography

OBJECTIVE: Prosthetic valve thrombosis is a serious complication associated with high morbidity and mortality; the early detection of thrombotic formations is crucial for prompt diagnosis and proper therapy. METHODS: Thrombotic deposits were simulated in vitro on commercial bileaflet mechanical valves. Valve closing sounds were acquired by phonocardiographic means in the frequency range from 6 to 55 KHz. The corresponding power spectra were calculated and analyzed by a specifically trained artificial neural network. RESULTS: Results demonstrate that the classifier is able to properly detect the simulated thrombotic formations; classification performances depend on the range of frequency considered: better performances (up to 100%) are achieved when the entire spectrum is investigated, rather than the audible (6-22 KHz) and ultrasonic parts (22-55 KHz) separately. CONCLUSIONS: In order to develop a diagnostic tool for the early detection of mechanical valve thrombosis, ultrasound phonocardiography combined with a properly trained classifier is suitable to detect thrombotic formations, discriminating different simulated functional conditions.