Phonocardiogram: A robust algorithm for generating synthetic signals and comparison with real life ones. (July 2020)

• Record Type: Journal Article
• Authors: Tiam Kapen, Pascalin; Youssoufa, Mohamadou; Kouam Kouam, Serge Urbain; Foutse, Momo; Tchamda, André Rodrigue; Tchuen, Ghislain
• Is Part Of: Biomedical signal processing and control. Volume 60(2020)
• Publication Date: 2020-07
• Subjects: Phonocardiogram -- Synthetic signals -- Mathematical modeling -- Real life PCG signals -- Robust algorithm; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.101983
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: A robust algorithm for generating synthetic PCG signals close to real life ones. The amplitude, S1S2 duration and heart rate of the real PCG used in the synthesis. The signals preprocessed with a second order Butterworth low pass filter of cutoff frequency of 500 Hz. Time domain correlations between the two sets of signals showed a good agreement. Percentage difference: 1.61% (heart rate), 2.27% (amplitude), 4.51% (S1S2 duration). Abstract: The work presented in this paper aims to develop and implement a robust algorithm for generating synthetic phonocardiogram (PCG) signals based on characteristics obtained from real life signals. The purpose of this work is to identify a few set of parameters that can be linked to the physiological state of the heart. The mathematical model used in the synthesis of the signal was based on three coupled ordinary differential equations (ODE) proposed by Jabloun et al. (2013). The amplitude, S2S1 duration, and heart rate of the real life PCG signals were extracted and used in the synthesis of the signals. The real life PCG signals were obtained from the MITHSDB and AADHSDB databases from Massachusetts Institute of Technology and Aalborg University respectively. The signals were preprocessed using a second order Butterworth low pass filter of cutoff frequency of 500 Hz to obtain cleaner versions. The system of equations was numerically solved with the fourth-order Runge-Kutta method. These synthetic PCG signals were then compared with the real life PCG signals. The time domain correlations between the two sets of signals showed a good agreement between the two with values above 0.8 for all tested cases. The percentage differences in heart rate (beats per minutes), amplitude and S2S1 durations between the real life and synthesized PCG signals also showed a close match with percentage values as small as 1.61%, 2.27% and 4.51% for heart rate, amplitude and S2S1 duration respectively.