Study Of Adaptive Methods Used For ECG Denoising

Electrocardiograph (ECG) plays an irreplaceable role in diagnosing heart diseases. ECGwas poluted by a variety of noises during acquisition process, including powerlineinterference caused by the power system, baseline drift caused by human respiratory, EMGcaused by muscle tremors, electrode contact noise caused by electrode loss, motion artifactsand so on. Because bioelectricity is very weak, noises usually have great impact on ECGanalysis, so the primery task after collecting ECG is denoising.In this paper we used the advantage of adptive filter in dealing with non-stationarysignals, to study the problem of power line interference and baseline wander, and focused onsolving the power line fluctuation and frequency overlapping problem of baseline wander, asfollows:Powerline interference makes the signal to noise ratio (SNR) of ECG reduced greatly.There is±3%fluctuation around the main frequency of powerline interference. Classicalmethods to filter the powerline interference have no effect on powerline fluctuation. In thearticle we come up with a frequency-tracking adaptive filter to track frequency change ofpowerline interference. We got the instantaneous frequency of the adaptive filter using theweight of adptive filter during adaptive filtering process. We adjusted the frequency of thereference signal according the relationship between the instantaneous frequency and powerline interference. At last we achieved the purpose to track the frequency change of powerfrequency. In the paper, both effectiveness and stability of the frequency-tracking adaptivefilter were evaluated. Experimental results show that the frequency tracking adaptivefrequency filter could track the frequency of powerline fluctuctions effectively and has simplestructure and better robustness.ST-segment of ECG signal is an important basis to diagnose heart diseases such asmyocardial ischemia. Baseline wander will influence the diagnosing of those diseases for itsfrequency is overlapped with ECG ST-segment. Morphological filter can solve this problem.We used this advantage combined morphological filtering with adaptive filtering. Used thebaseline drift extracted by morphological filtering as reference signal of adaptive filter.Morphology-adaptive filtering system for baseline wander of ECG was verified from thelongitudinal and transverse. Compared with the morphological filter, this filter could suppressthe elevation phenomenon of ST-segment induced by morphological filter effectively. Thebaseline wander extracted from morphological filter have the best filtering result as referencesignal of adaptive filter on the SNR, MSE, StSR, StVR, compared to the straight line, slowsinusoidal curve, the baseline extracted using wavelet transform.