Time-varying filtering of the first and second heart sounds

Ensemble averaging has been widely used as a solution to wide-band noise problem for the first and second heart sounds. The need for a large number of samples of similar morphology, makes the ensemble averaging an inefficient method for this application. In the method presented here, an ensemble of the first and second heart sound is constructed from a single recording so that its statistical properties satisfy the basic assumptions involved in the time-varying Wiener filtering strategy. The time-varying system function required for such a filtering technique, is obtained by using a bank of octave filters which matches the time-varying structure of heart sounds. A simulation study indicates that, compared to conventional ensemble averaging, this method achieves the same improvement in signal-to-noise ratio with a reduction of between 6 and 8 in the number of required samples