Abstract filter, LMS algorithm ?. INTRODUCTION ECG

Abstract -Electrocardiogram (ECG) is a method ofmeasuring the electrical activities of heart. Every portion of ECG is veryessential for the diagnosis of different cardiac problems. But the amplitudeand duration of ECG signal is usually corrupted by different noises. Removing motion artifactsfrom an electrocardiogram (ECG) is one of the important issues to be consideredduring real-time heart rate measurements in health care. It is essentialto reduce these disturbances and improve the accuracy as well as reliability.

The noises that commonly disturb the ECG signals are Random noise, Gaussianwhite noise, Power line interference, Baseline wander and Electromyography(EMG) noise .These noises can be classified according to their frequencycontent. The noise signals have been generated and added to the ECG signaltaken from MIT-BIH arrhythmia database.

Sometimes it is hard to do all the work on your own
Let us help you get a good grade on your paper. Get expert help in mere 10 minutes with:
  • Thesis Statement
  • Structure and Outline
  • Voice and Grammar
  • Conclusion
Get essay help
No paying upfront

In this paper we have done a broaderstudy for denoising every types of noise involved with real ECG signal. Oneadaptive filter, least-mean-square (LMS) is applied to remove the noises. PSNR and MSE performance parameterare  estimated. Keywords – ECG signal,artifacts, adaptive filter, LMS algorithm ?.  INTRODUCTIONECGis generated by the heart muscle and measured on the skin surface of the bodyThis signal is ranging from 10microvolts to 5 mille-volts, frequency from 0.05Hz to 100 Hz.ECG is helpful to  detectchanges in cardiac muscles like myocardial infarction, conduction defects andarrhythmia .

When the electrical abnormalities of the heart occur, the heartcannot pump and supply enough blood to the body and brain. As ECG is agraphical recording of electrical impulses generated by heart, it is needed tobe done when chest pain occurred such as heart attack, shortness of breath,faster heartbeats, high blood pressure, high cholesterol and to check theheart’s electrical activity. It recognises the variability’s of heart activity,so it is very important to get the ECG signal  free from noise.   Basically ECG signal is characterized by five  peak points – P, Q, R, S, and T. The waveform which is repetitive and have various bumps andparts of the waveform are designated as the P-wave, QRS complex and T-wave,PR-segment, ST-segment, PR-interval and QT-interval as given in Figure 1. Theorigins of these waves are:i.P wave: sequential activation (depolarization) of the right and left atriaii.QRS complex: right and left ventricular depolarizationiii.

T wave: ventricular repolarisationiv.U wave: repolarisation of the papillary muscles, rarely seen.   Fig.1:ECG waveform II.

  NOISES IN ECG SIGNAL ECGSignals generated from human body are often very weak so as to be easilycovered by background noise. The noise in the ECG signals occur due to variousreasons like electromagnetic interference due to ubiquitous supply lines andplugs, movement of patient, signals generated by other organs and impedance mismatchingbetween electrodes. Hence the ECG signals can be corrupted by various types of noisessuch as Power line interference, Electrode contact noise, Motion artifact, Musclecontraction, Base line drift, Instrumental noise generated by electronicdevices . Power line interference noise is electromagnetic field from thepowerline which causes 50/60Hz sinusoidal interference.

This noise causesproblem in interpreting low amplitude waveform like ECG. Various methods havebeen employed for the removal of artifacts from ECG signals. Adaptive filteringis one of the efficient method in the removal of noises in the ECG signals. III.GENERATIONOF NOISES The low frequency noise (base linewander, i.

e. electrode contact noise and motion artifact) has frequency lessthan 1Hz, high frequency noise (EMG noise) whose frequency is more than 100Hzand power line interference of frequency 50 Hz or 60Hz (depending on thesupply) can be generated as follows. These noises are generated in MATLAB basedon their frequency content, which is then added with the ECG signal to get thenoisy ECG. A .Generation ofRandom noise Highfrequency random noise signal is generated. The generated high frequency noiseis shown in Figure 2.

    Fig. 2: Randomnoise signal B.Generation of  Gaussian White noise                                Generated Gaussian white noise isshown in Figure 3.   Fig. 3: Gaussianwhite noise signal C. Generation of Baseline wander noise Generated the baseline drift by  which is shown in Figure 4.   Fig. 4: Baselinewander noise signal D.

Generation ofPower line interference noise  We haveconsidered the 50 Hz power supply. So, we have taken a sine wave of 50 Hzamplitude to represent the power line interference. The resulted power lineinterference is shown in Figure 5.

   Fig.5: Powerline interference noise signal    IV. ADAPTIVEFILTERING ECGsignal has been a major diagnostic tool for the cardiologists and ECG signalprovides almost all the information about electrical activity of the heart. Socare should be taken while doing the ECG filtering, such that the desiredinformation is not distorted or altered in any way. The original ECG signal is taken from the MIT-BIH arrhythmia database. The different types of noise signal are generated by using MATLAB. The noise signal is then added with the real ECG signal. To remove the different types of noises, the noisy ECG signal is then pass through adaptive filteralgorithms (e.

g., LMS). However, the basic block diagram  of adaptive filtering is shown in figure 6.  Fig.6: Adaptive filter Inthis paper, we have designed a multistage filter for cancellation of theseartifacts.

The filter consists of two stages(1st stage and 2nd stage) as shownin Fig 7.  Fig.7: LMS two stage filtering Least mean squares (LMS) algorithms are a class ofadaptive filter used to mimic a desired filter by finding the filtercoefficients that relate to producing the least mean squares of the errorsignal (difference between the desired and the actual signal).

It is astochastic gradient descent method in that the filter is only adapted based onthe error at the current time. The LMS Algorithm consists of two basicprocesses 1. Filtering process -Calculate the output of FIRfilter by convolving input and taps. Calculate estimation error by comparingthe output to desired signal. 2.

Adaptation process:-Adjust tap weights based on the estimation error.Considera length L LMS based adaptive filter, that takes an input sequence x(n)and updates the weights as w(n + 1) = w(n) + ? x(n) e(n) wherew(n) = w0(n) w1(n)….wL?1(n)t isthe tap weight vector at the nthindex, x(n) = x(n) x(n?1)….x(n?L+1)t e(n) = d(n)?wt(n) x(n) withd(n) being the so-called desired response available duringinitial training period and ? denoting so-called step-size parameter.Inorder to remove the noise from the ECG signal, the ECG signal s1(n) withadditive noise p1(n) isapplied as the desired response d(n) for the adaptive filter. Ifthe noise signal p2(n),possibly recorded from another generator of noise that is correlated in someway with p1(n) isapplied at the input of the filter, i.e.

,  x(n) = p2(n)  The filter error becomes, e(n) = s1(n) + p1(n) ? y(n)  The filter output y(n) isgiven by, y(n) = wt(n)x(n) Since the signal and noise areuncorrelated, the mean-squared error (MSE) is, Ee2(n) = E{s1(n) ? y(n)2} + Ep21(n) Minimizing theMSE results in a filter output that is the best least-squares estimate of thesignal s1(n).   V.SIMULATION RESULTSToshow that LMS algorithm is really effective in clinical situations, the methodhas been validated using several ECG recordings with a wide variety of wavemorphologies from MIT-BIH arrhythmia database. The arrhythmia data baseconsists of 48 half hour sets of two channel ambulatory ECG recordings, whichwere obtained from 47 subjects including 25 men aged 32-89 years and women aged23-89 years. The recordings were digitized at 360 samples per second perchannel with 11-bit resolution over a 10mV range. The generated noises inMATLAB based on their frequency content, which is then added with the ECGsignal to get the noisy ECG. The output of the first stage is coupled to thesecond stage where the noise free ecg signal can be obtained.

We have consideredfour different types of noises to corrupt our signal namely Power line Interference,Baseline Drift, guassian noise and white noise. Results of LMS along with meansquare error (MSE) and peak signal –to-noise ratio are also shown. Six ECGsignal were obtained from this database to validate the results. The followingfigures are the ECG signals corrupted by various noises.   Fig.8: Pure ECG signal   Fig. 9: Noisyecg signal(random noise)    Fig.

10: Noisy ecg signal (Gaussian white noise)    Fig. 11: Noisy ecgsignal(baseline wander noise)   Fig.12: Noisy ecg signal (powerline interference noise)                 The two stageadaptive filtering of the various noises is obtained as follows.   Fig.13: LMS Result(of random noise)   Fig.

14: LMS Result(of Gaussian white noise)   Fig.15: LMS Result(of Baseline wander noise  Fig.16: LMS Result(of Powerline interfernce noise) TABLE I. VALUES OF PERFORMANCE PARAMETERSOF TWO STAGE ADAPTIVE FILTER FOR DIFFERENT TYPES OF NOISE          NOISES   RECONSTRUCTED SIGNALS   MSE             PSNR   1ST STAGE   2ND STAGE   1ST  STAGE STAGE S   2ND STAGE   Random   0.0154   6.09e-10   16.1990   90.

2149 Gaussian white     0.0148   1.33e-08   16.

3748   76.7935 Baseline wander     0.0247   1.35e-08   14.1409   76.7579 Powerline interference     0.

0038   9.06e-07   22.2214   58.4863    VI. CONCLUSION In this paper, the problem of noisecancellation from ECG signal using adaptive filters are proposed and tested onreal signals with different artifacts obtained from the MIT-BIH database. Forthis, the input and the desired response signals are properly chosen in such away that the filter output is the best least squared estimate of the originalECG signal.LMS algorithm work effectively in removing the noises from the ECGsignal.     REFERENCES 1R.

Panda and U. C. Pati, “Removal of artifacts from   electrocardiogram using digital filter,” inElectrical, Electronics and Computer Science (SCEECS), 2012 IEEE Students’Conference on IEEE, 2012, pp. 1–4 2S.Mahmoodabadi, A. Ahmadian, M. Abolhasani, M.

Eslami, and J. Bidgoli, “Ecgfeature extraction based on multiresolution wavelet transform,” in Engineeringin Medicine and Biology Society, 2005.IEEE-EMBS 2005. 27th Annual InternationalConference of the. IEEE,2006, pp.

3902–3905. 3R.Sameni, M. B. Shamsollahi, C. Jutten, and G.

D. Clifford, “A non linearbayesian filtering framework for ecg denoising,”IEEE Transactions on BiomedicalEngineering, vol. 54, no. 12, pp. 2172–2185, 20074C.H.Chang, H.J.

Ko, and K.-M. Chang, “Cancellation of Artifacts in ECG Signalsusing Signbased Normalized Adaptive Filtering Technique,” in Biomedical Engineeringand Informatics (BMEI), 2010 3rd International Conference on, vol. 2. IEEE,2010, pp.

787–790. 5 S. Singh, “Performanceevaluation of different adaptive filters for ecg signal processing 1,” 2010.

 6M. G. Khan, Rapid ECG interpretation. Springer Science & Business Media,2008.

 7M. Z. U. Rahman, R. A.

Shaik, and D. R. K.

Reddy, “Efficient sign basednormalized adaptive filtering techniques for cancelation of artifacts in ecgsignals: Application to wireless biotelemetry,” Signal Processing, vol. 91, no.2, pp. 225–239, 2011. 8P. E. McSharry, G. D.

Clifford, L. Tarassenko, and L. A.

Smith, “A dynamicalmodel for generating synthetic electrocardiogram signals,” IEEE Transactions onBiomedical Engineering, vol. 50, no. 3, pp.

289– 294, 2003. 9V. Viknesh and P. R. Prashanth, “Matlab implementation of ecg signalprocessing,” IOSR Journal of VLSI and Signal Processing, vol. 3, no.

1, 2013. 10M. Alfaouri and K. Daqrouq, “Ecg signal denoising by wavelet trans-formthresholding,” American Journal of applied sciences, vol. 5, no.

3, pp.276–281, 2008. 11S. Khan, S.

M. Anwar, W. Abbas, and R. Qureshi, “A novel adaptive algorithm forremoval of power line interference from ecg signal,” Science International,vol.

28, no. 1, 2016.  


I'm Gerard!

Would you like to get a custom essay? How about receiving a customized one?

Check it out