Nonlinear Analysis of Diastolic Heart Sounds Based on EMD and Correlation Dimension

Recent studies have applied nonlinear analysis methods for heart sounds to diagnose coronary artery disease (CAD). Coronary artery occlusion may cause diastolic heart murmurs, so analysis of diastolic heart murmurs has important significance to noninvasive diagnosis of CAD. Heart sound signal is typical nonlinear and non-stationary time series, nonlinear analysis method correlation dimension can effectively describe the nonlinear characteristics of heart sound signals, but the analysis of the correlation dimension shows that trend terms in the heart sound signals may lead to erroneous results. Empirical mode decomposition (EMD) is adaptive to remove trend for non-stationary signal, so a method combining EMD and correlation dimension was proposed for nonlinear analysis of diastolic heart sound signals. The EMD method was applied to reconstruct heart sound signals after removing trend, and the correlation dimension for reconstructed heart sound signals was used as characteristics to distinguish between normal heart sound signals and CAD heart sound signals. The diastolic heart sounds of 15 normal people and 15 patients with CAD were analyzed in the experiment, and the results showed that the proposed method can effectively distinguish between normal people and patients with CAD. Copyright © 2014 IFSA Publishing, S. L.