Extraction of the aortic and pulmonary components of the second heart sound using a nonlinear transient chirp signal model

Abstract
The objective of this paper is to adapt and validate a nonlinear transient chirp signal modeling approach for the analysis and synthesis of overlapping aortic (A/sub 2/) and pulmonary (P/sub 2/) components of the second heart sound (S/sub 2/) The approach is based on the time-frequency representation of multicomponent signals for estimating and reconstructing the instantaneous phase and amplitude functions of each component. To evaluate the accuracy of the approach, a simulated S/sub 2/ with A/sub 2/ and P/sub 2/ components having different overlapping intervals (5-30 ms) was synthesized. The simulation results show that the technique is very effective for extracting the two components, even in the presence of noise (-15 dB). The normalized root-mean-squared error between the original A/sub 2/ and P/sub 2/ components and their reconstructed versions varied between 1% and 6%, proportionally to the duration of the overlapping interval, and it increased by less than 2% in the presence of noise. The validated technique was then applied to S/sub 2/ components recorded in pigs under normal or high pulmonary artery pressures. The results show that this approach can successfully isolate and extract overlapping A/sub 2/ and P/sub 2/ components from successive S/sub 2/ recordings obtained from different heartbeats of the same animal as well from different animals.