A novel methodology for evaluation of S₂ wide split via estimated parameters

Aimed at the shortcomings of using time interval ( [Formula: see text] ) between the sounds produced by the aortic valve closure (A₂) and the pulmonary valve closure (P₂) to detect the wide splitting of the second heart sound (S₂), which are the [Formula: see text] easily influenced by the heartbeat and not easily distinguished from the fixed splitting of S₂ without considering the entire respiratory phase, and from the third heart sound (S₃), this study proposes a novel methodology to detect the wide splitting of S₂ using an estimated split coefficient of S₂ ( [Formula: see text] ) combined with an adaptive number (N_Ada) of S₂.

The methodology is orderly summarized as follows: Stage 1 describes the segmentation-based S₂ automatic location and extraction. A Gaussian mixture model (GMM)-based regression model for S₂ is proposed to estimate the positions of A₂ and P₂, then an overlapping rate (OLR)-based [Formula: see text] and the [Formula: see text] are estimated, and finally, a N_Ada-S₂ is automatically determined to calculate the statistics of [Formula: see text] and [Formula: see text] . In stage 3, based on the combination of estimated features, the detection of wide splitting of S₂ is determined.

The performance is evaluated using a total of 3350-period heart sounds from 72 patients, with an overall accuracy of 100%, F₁=1 and a Cohen's kappa value (κ) of 1.

The significant contributions are highlighted: A novel GMM-based efficient methodology is proposed for estimating the characteristics of A₂ and P₂. A novel OLR-based [Formula: see text] is defined to replace the current state-of-the-art criterion for evaluating the split degree of S₂. Considering respiration phases combined with CR are proposed for the high-precision diagnosis of S₂ wide split.