Estimation of Splitting Interval in Second Heart Sound by Optimizing a Demixing Vector

A non-invasive method to estimate the splitting interval (SI) of the second heart sound (S2) was developed and evaluated. Recently, there has been a growing need for a non-invasive SI estimation for the early detection of cardiovascular diseases. Therefore, we proposed an SI estimation algorithm by separating the aortic and pulmonary components (A2 and P2, respectively) from S2 with an optimized demixing vector and mathematical model of S2. Herein, we evaluated the SI estimation performance of our algorithm. As a result, our algorithm could determine a simulated S2's SI with an error of
$1.3+1.0$
ms. A2 and P2 were separated from recorded human S2 signals. The resulting SI estimation of the recorded data showed a correlation coefficient with S2-S2 intervals of
$r=- 0.67$
. This negative correlation is consistent with reported physiological features of the heart.