Optimized adaptive noise canceller for denoising cardiovascular signal using SOS algorithm

Abstract

Cardiovascular or electrocardiogram signals (ECG) are contaminated by various artefacts while recording which in turn degrades the quality of the vital information present in the signal. Hence, noise removal from ECG signal is an important matter of concern. In this paper, swarm intelligence techniques for the optimization purpose of adaptive filters/noise canceller (ANC) are utilized in the biomedical signal processing field. The analysis of results for de-noising ECG signals through adaptive filtration is presented using the symbiotic organisms search (SOS), particle swarm optimization (PSO) and harmony search (HS) algorithms which are applied to estimate and adjust the parameters of the adaptive filter in such a way to keep a close track of spontaneous variations of non–stationary signals in a faithful manner. From the simulation results, it has been observed that the ANC filter designed with the SOS technique achieves significant improvement in fidelity parameters, such as signal to noise ratio (SNR), mean square error (MSE), maximum error (ME), mean difference (MD), peak reconstruction error (PRE), normalized root mean squared error (NRMSE) and normalized root mean error (NRME) when compared with other reported techniques in the literature as well as the benchmark algorithms, namely PSO and HS. The SOS based ANC technique results in nearly 7 dB improvement in output SNR as compared to the recently reported ANC filter based on bacteria foraging optimization (BFO) algorithm and also provides a fair reduction of more than 90% in NRMSE and NRME when compared to other state-of-art denoising techniques. Furthermore, the ANC filter using the SOS technique enhances the correlation between the pure and reconstructed ECG signals.