Digital multiband filter design with power spectrum analysis for Electrocardiogram signals

In this paper, digital multiband filter design is proposed with power spectrum analysis for de-noising of an Electrocardiogram (ECG) signal containing various types of noises in a different frequency band. A single digital multiband (five band) filter is designed to remove such noises instead of using separate High Pass, Low Pass and Band Pass filter. The multiband filter coefficients are obtained using the least mean square based optimization technique. The power spectrum of filtered ECG signals is analyzed for quantitative measurement of designed filter, in terms of required attenuation level in the presence of power-line interference (PLI), baseline wandering (BW) and motion artifact (MA) noises. The experimental investigation is carried out using Physionet ECG ID database which contains records of real time ECG signals with all such noises inherently present in it.