Non-invasive determination of mitral valve acoustic properties: a proposed method to determine tissue alterations due to disease

The purpose of this study was to determine the time-frequency content of the heart sounds, in both animal and computer models. In normal, unoperated sheep (N=8), the first heart sounds (S1) were recorded with an acoustic sensor. Data analysis consisted of cross-correlation of successive heartbeats to determine beat-to-beat regularity, as well as a Wigner time-frequency distribution analysis, to generate time-frequency "signatures" of each S1. In addition, a dynamic, fluid-structure coupled finite element model (FEM) of the mitral valve was constructed, and an identical time-frequency analysis of the normal velocity vector of the valve elements was performed. The average cross-correlation coefficient for the eight consecutive beats of all eight sheep was 0.90. The time-frequency signatures were generally the shape of a descending crescent centered on 65 Hz. Time frequency analysis of the normal velocity vector of the FEM valve elements was a descending crescent centered on 48 Hz. We conclude that time-frequency analysis can be used to obtain a "signature" of the radiated vibrations of the mitral valve, which correlates with numerical models.