Separation of Cardiopulmonary Acoustic Indicators

These studies demonstrate a combination of tested methods to separate lung and heart sound signals' (LS-HS) principal components recorded using a digital dual stethoscope. Initially, when these raw signals are acquired, their principal components intertwine in the characteristics of the frequency and time domains. Applying Independent Component Analysis (ICA) combined with Empirical Mode Decomposition allows separation of their origin. Further, Gaussian Mixture Models (GMM) were utilized to determine to which specific class of components they belong. Each set of recorded signals was evaluated and the outcomes were compared against the verified reference group. As cardiopulmonary diseases represent a major global health concern, this can be helpful in an early screening for distinct cardiopulmonary problems manifested by peculiar acoustic indicators.