Comparison of two spike areas in bileaflet mechanical valve closing sound

Abstract

We previously reported our development of a wavelet analysis system which demonstrates that in vivo bileaflet mechanical valve sound splits into two spikes at higher frequency levels and, based on this system, proposed criteria for detecting malfunctioning bileaflet valves (MBVs). However, the results of that study were only tentative due to the small number of patients with MBVs enrolled in the study. Here, we discuss the possibility of new criteria based on the scalographic properties of two spikes of bileaflet valve sound. The study cohort comprised 12 patients who each received a Carbomedics valve. Based on cinefluoroscopy findings, seven valves were classified into a "normal" group, and the other five were classified into a "malfunction" group. Five consecutive valve sounds for each valve were collected for the wavelet analysis in order to re-evaluate the previously proposed criteria and to measure both anterior spike area (Aa) and posterior spike area (Pa) for calculating the spike area ratio (Aa/Pa). The proposed criteria, namely, a single spike or coefficient of variation of <0.1120 detected only two of the five malfunctioning valves, as well as one normal valve to be malfunctioning. The mean Aa/Pa of all malfunctioning valves [2.45±0.63; 95% confidence interval (CI) ±1.01, 95% confidence limits (CL) 1.44-3.46] was significantly higher than that of all normal valves (1.17±0.27; 95% CI ±0.25, 95% CL 0.92-1.42). Based on this result, we determined the cutoff value of Aa/Pa to be 1.4. The combination of a single spike on the scalogram and an Aa/Pa of >1.4 detected more MBVs than previously proposed criteria. This combination may represent new criteria for detecting MBVs.