Processing of prosthetic heart valve sounds for single leg separation classification

Area of Science:

• Biomedical Engineering
• Acoustical Signal Processing
• Mechanical Heart Valve Diagnostics

Background:

• Mechanical prosthetic heart valves, while life-extending, are susceptible to fatigue and structural failure over time.
• Noninvasive monitoring of heart valve sounds in noisy environments presents significant signal processing challenges.
• Extracting diagnostic information from low-amplitude, nonstationary transient sounds like heart valve opening is crucial.

Purpose of the Study:

• To discuss acoustical signal processing techniques for analyzing noisy heart valve sounds, specifically focusing on the opening cycle.
• To extract critical information about the outlet strut condition, a component implicated in valve failure.
• To develop preclassification signal processing methods to enhance the extraction of desirable acoustic signals from prosthetic heart valves.

Main Methods:

• Utilized sensitive surface contact microphones to measure heart valve sounds in a noninvasive manner.
• Developed sophisticated signal processing algorithms to extract weak opening sounds from background noise.
• Employed a parametric processing approach based on an autoregressive model to characterize sounds from the Bjork-Shiley convexo-concave (BSCC) valve opening cycle.

Main Results:

• Successfully developed preclassification signal processing techniques to improve signal-to-noise ratios for transient heart valve sounds.
• Created a beat monitor that filters out beats failing spectral acceptance criteria, ensuring data quality.
• Demonstrated the potential of analyzing opening sounds for detecting outlet strut condition, a key indicator of valve integrity.

Conclusions:

• Advanced acoustical signal processing is vital for noninvasively assessing the health of mechanical prosthetic heart valves.
• Focusing on the opening cycle sounds provides direct insights into outlet strut condition, aiding in early failure detection.
• Parametric modeling, particularly autoregressive approaches, shows promise for characterizing and analyzing these complex valve sounds.