Time-frequency cardiac passive acoustic localization

Abstract

Abstract-In a previous work, the authors have used a novel, radically different approach based on concurrent multi-sensor array measurements and "super-resolution" array processing scheme for phonocardiographic signals. Using a specially designed passive acoustic array, acoustic "images" corresponding to the various distinct phases of the heartbeat were obtained. In this paper, this approach is extended using a novel framework, called TimeFrequency Cardiac Passive Acoustic Localization (TF-CARDIOPAL). The Choi-Williams distribution is utilized to calculate a spatial time-frequency matrix to characterize the spatial time-frequency distribution for the nonstationary array signals. Distinctive components of this matrix on the time-frequency plane yield important clues on possible source locations in the heart. This finding has been used in extracting localization information from the first heart sound signals from a healthy human subject. Meaningful source localization results that are well correlated with the mechanical activation of the heart are obtained.