A novel approach for decomposition of biomedical signals in different applications based on data-adaptive Gaussian average filtering

Abstract

The analysis of biomedical signals plays a crucial role in modern medicine and physiological research. Due to exogenous or endogenous interferences and complex interaction among physiological systems, biomedical signals usually possess nonstationary characteristics. To extract the features buried in such kind of signals, the signal decomposition algorithm that is data-adaptive and stable is highly demanded. This paper introduces a novel decomposition algorithm termed as data-adaptive Gaussian average filtering (DAGAF) that is new and potential in biomedical applications. Six biomedical scenarios, including finger photoplethysmography (PPG) signal with obscure respiratory-induced intensity variation (RIIV) component, wrist PPG signal with apparent RIIV component, seismocardiography (SCG) signal with implicit respiration component, electrocardiogram (ECG) with baseline wander (BW), ECG with power-line interference (PLI), and R-R intervals (RRI) sequence with implicit low-frequency trend wave, are adopted as examples for computer experiments. The results of computer experiments verify that the DAGAF algorithm can satisfy the specified requirement in different biomedical scenarios. DAGAF algorithm possesses the advantages of mathematical formulation and computational efficiency. It can be an alternative choice besides the commonly used empirical mode decomposition (EMD) and ensemble empirical mode decomposition (EEMD).