A novel cardiac spectral envelope extraction algorithm using a single-degree-of-freedom vibration model

A novel cardiac spectral envelope extraction method was developed to acquire the optimal envelope information in the frequency domain to differentiate normal heart sounds and valvular heart disorders for further analysis. This automatic algorithm was based on the vibration model of a mass-spring-damper system. To acquire the best spectral envelope information from both a normal heart sound signal and an aortic regurgitation murmur signal, the morphological influence of two analytical parameters, the natural frequency p (range: 1-100 Hz) and the damping factor ζ (range: 10-500%), on the single-degree-of-freedom model was evaluated using the root mean square error. The cardiac spectral envelope curve could be optimized at the natural frequency of 35 ± 5 Hz and the damping factor of 70 ± 8%. The spectral envelope extracted by the single-degree-of-freedom model showed more informative and efficient morphological characteristics than that extracted by an autoregressive power spectrum density model reported previously.