Denoising Phonocardiogram signals with Non-negative Matrix Factorization informed by synchronous Electrocardiogram

摘要

The phonocardiographic signals (PCG) are of interest for the analysis of the cardiac mechanical function. However, they are not always directly exploitable because of ambient interference (gastric noises, breathing noises, etc.). We aim to denoise PCG signals using another cardiac modality, the electrocardiographic (ECG) signal. In this paper, we investigate an informed non-negative matrix factorization to extract signal components out of the noisy PCG signal, considering synchronous ECG information. Our approach is applied and evaluated on a database consisting of real and artificially noisy PCG signals.