Lung-Heart Sound Separation Using Noise Assisted Multivariate Empirical Mode Decomposition

Separating lung sound (LS) from breath sound (BS) recording has been of interest to doctors and researchers in the last two decades. Many algorithms have been developed to solve this question, one of them is based on the empirical mode decomposition (EMD). Due to the notorious mode mixing issue in the standard EMD, this paper surveys LS extraction based on EMD extensions, including ensemble EMD (EEMD), multivariate EMD (M-EMD), and noise assisted M-EMD (NAM-EMD). In this study, the algorithm for LS extraction is composed of heart sound (HS) segmentation, LS separation, and segments reconstruction. The performance evaluation by auditory and numerical analyses reveals that NAM-EMD based LS extraction is superior to the standard EMD and its extensions.