Noise Removal from ECG Signal and Performance Analysis Using Different Filter

Transcription

1 International Journal o Innovative Research in Electronics and Communication (IJIREC) Volume. 1, Issue 2, May 214, PP ISSN (Print) & ISSN (Online) Noise Removal rom ECG Signal and Perormance Analysis Using Dierent Filter Manoj Sharma Electronics and communication CBS Group o Institutions Jhajjar,India manojsharma88@gmail.com Hemant dalal Electronics and communication CBS Group o Institutions Jhajjar,India hemantdalal87@gmail.com Abstract: This paper presents removal o noise rom the ECG signal by using Digital ilters designed with FIR and IIR technique. The analysis o ECG signal has great importance in the detection o cardiac abnormalities. The ECG signal is preserve o the electrical perormance o heart versus time. ECG signal o a normal heart beat consists o a three parts P wave, QRS complex and T wave. The P wave relect the activation o the right and let atria. The QRS complex shows depolarization o the activation o right and let ventricles. Results are obtained or the given order o the ilter using windowing technique or the FIR ilter. The wavelet transorm is used to reduce the eect o noise to get reined signal. The power spectral density and average power, beore and ater iltration using dierent window techniques and wavelet utilization at 4 and 6 db are compared. Order o the ilter is also dierent. Filter with the Kaiser window shows the best result. Keywords: ECG, FIR Filter, Windowing Technique, Wavelet Transorm, power spectral density and average power 1. INTRODUCTION Intererence occurs in ECG signal is very common and serious problems. Digital ilter are designed to remove this limitation. FIR with dierent windowing method is used. The results are obtained at low order. The input signals are taen rom ECG database which includes the normal and abnormal waveorms. FDA tool is used in MATLAB to design these ilters [1]. Many times when ECG signal is recorded rom surace electrode that are connected to the chest o patient, the surace electrode are not tightly in contact with the sin as the patient breath the chest expand and contract producing a relative motion between sin and electrode. This results in shiting o baseline which is also nown as low requency baseline wander. The undamental requency o baseline wander is same as that o respiration requency. It is required that baseline wander is removed rom the ECG beore extraction o any meaningul eature. 4 ECG DATA ONE Fig1. ECG data with 8 samples on the conerence website. Baseline wander maes it diicult to analyze ECG, especially in the detection o ST-segment deviations. 2. FILTER DESIGN METHODS 2.1. Window Use in Designing ARC Page 32

2 Noise Removal rom ECG Signal and Perormance Analysis Using Dierent Filter FIR ilters can also be designed using the windowing method. The ideal ilter have ininite number o samples in time domain given in equation 3. Windows are perormed in order to have inite number o samples in time domain or reliable ilter design. Fig4. Magnitude response o an ideal window. A window unction rom wc to wc is employed to show the windowing eect [15]. There are dierent windowing unctions. The important window unctions are rectangular window, Hamming, Hanning, Blacman windows[15] Rectangular Window The ilter is required to have inite number o values within a certain interval, rom -M to M. This is equivalent to multiplying d () by a rectangular unction given by 1, i n M (4), otherwise Hamming Window Discontinuties in the time unction cause ringing in the requency domain. The rectangular window is replaced by a window unction ending smoothly at both ends which will cause reduction in ripples. The hamming window is an important window unction. The hamming window is deined as: cos 2 N n 1 n 1,2,3,4... N 1 (5), Where N is the order o the ilter and M is the window length. This equation deines the window samples as already shited (indices rom to N-1 ). So the impulse response o the FIR low pass ilter designed using the hamming window is[15]: h (. d( n M ) h( cos 2 n N 1 sin( M ). w M ). The ripples that occur in rectangular windowing in both the pass band and the stop band are virtually eliminated. Thus, the iltered data will have a wider transition width. The Hamming window is deined mathematically as: 2 n.5.5cos n,1,2,3,4... N 1 (7) N 1 International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 33 (6)

3 Manoj Sharma & Hemant dalal The dierence o Hamming window is perormed window unction. This unction is quite similar to the Hamming window Blacman Window The Blacman window exhibits a lower maximum stop band ripple in the resulting FIR ilter than the Hamming window. It is deined mathematically as: The width o the main lobe in the magnitude response is wider than that o the Hamming window High Pass Filter Design The amplitude response o a low pass ilter is shown in Fig. 5. Low pass ilter is irst applied, and with simple transormations the high pass ilter can then be easily perormed. (8) Fig5. Magnitude response o a low pass ilter. Pass-band and stop-band regions are illustrated with equation 9 and equation 1. The derivation o the transormation is speciied with the ollowing equations: wp 2 pass s, ws 2 stop s & wc 2 The ideal cut o requency, c, is at the midpoint between the pass band and stop band edge requencies set in equation 1 The transition width is deined as: 2 c c pass stop c (1) stop pass (11) Since the role o pass and stop are interchanged in order to design high pass ilter. The ideal high pass impulse response is obtained rom the inverse Fourier transorm o the ideal high pass requency response. It is speciied by equation 12: d ) ( ) sin wc. ( (12) The windowed ilter impulse response is: h( h( M ) M ) sin[ sin[ M ). wc ] M ) M ). wc ] M ) (9) (13) International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 34

4 Noise Removal rom ECG Signal and Perormance Analysis Using Dierent Filter 2.2. IIR Filter Design An IIR ilter is one whose impulse response theoretically continues or ever because the recursive terms eedbac energy into the ilter input and eep it as speciied in the ollowing equation: y( N 1 a( ). y( n ) M b( ). x( n ) M b( ) z H( z) N (14) a( ) z The theory o Butterworth unction is explained here but, the order o the ilter should be high and implementing a ilter o that order is not easy to perorm. In addition to this diiculty, solving these high order equations is not straightorward Wavelet A wavelet [11] is a wave-lie oscillation with amplitude that starts out at zero, increases, and then decreases bac to zero. It can typically be visualized as a "brie oscillation" lie one might see recorded by a seismograph or heart monitor. Generally, wavelets are purposeully crated to have speciic properties that mae them useul or signal processing. Wavelets can be combined, using a "shit, multiply and sum" technique called convolution, with portions o an unnown signal to extract inormation rom the unnown signal. As wavelets are a mathematical tool they can be used to extract inormation rom many dierent inds o data, including - but certainly not limited to audio signals and images. Sets o wavelets are generally needed to analyze data ully. 3. RESULTS AND CONCLUSION In this paper various noise removal techniques are applied to ECG signals[1], ECG database data sample, and the perormance o these approaches are studied on the basis o spectral density and average power o signal. In the irst step, the most simple approach which is linear trend or a piecewise linear trend to remove baseline drit is applied ater that various digital ilters are applied to the noisy ECG data having Baseline noise as shown in ig 4.1 then the wavelet approach is used or overall denoising o ECG signal and inally the digital ilter is applied on the sample ECG signal to remove Power line noise. All o the above steps are perormed using MATLAB sotware 3.1. Calculation o Parameters The two important parameters to chec the suppression o Baseline noises are spectral density and average power o signal [6] Power spectral density Table1. Comparison o various ilters or Removal o noise at ECG sample input 1. Filter Filter Order Spectral Density beore Filtration Spectral Density ater Filtration Wavelet output at 4dB Wavelet output at 6dB Hanning Kaiser Rectangular Chebyshev Eleptic International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 35

5 Manoj Sharma & Hemant dalal Table1 and 2 shows the comparison o dierent ilters. The trade-o between spectral density and average power is best among all the ilters. Spectral density o data 1 using dierent ilters is shown as ollows: Fig6. Spectral Density using Hanning ilter Fig.6 Spectral Density using Hanning ilter International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 36

6 Power/requency (db/hz) Power/requency (db/hz) Power/requency (db/hz) Power/requency (db/hz) Noise Removal rom ECG Signal and Perormance Analysis Using Dierent Filter Fig7. Spectral Density using Kaiser Filter Fig8. Spectral Density using Rectangular ilter Spectral density o orignal signal 1 Spectral density o iltered signal Spectral density o overall denoised signal by Spectral wavelet density ilter o o db4 overall denoised signal by wavelet ilter o db Fig9. Spectral Density using chebyshev ilter International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 37

7 Power/requency (db/hz) Power/requency (db/hz) Power/requency (db/hz) Power/requency (db/hz) Manoj Sharma & Hemant dalal Spectral density o orignal signal 1 Spectral density o iltered signal Spectral density o overall denoised signal by Spectral wavelet density ilter o o db4 overall denoised signal by wavelet ilter o db Fig1. Spectral Density using Eleptic ilter But it can also visualize that the waveorm got distorted to some extend in case o rectangular window. The Kaiser Window and rectangular window is also showing better results at the expense o some more computational load as the order o the ilter is large. But in case o remaining windows i.e. Hamming and Blacman windows, the order o ilter easily grow very much high. It increases the number o ilter coeicients which increases the large memory requirement and problems in hardware implementation. So, the Kaiser Window ilter can be best choice or the removal o Baseline wandering among ilters [2]. Average power Comparison o various ilters or Removal o noise at ECG sample input 1 in Table 2 4. CONCLUSION This paper concludes the wor in this thesis; digital FIR and IIR ilter with wavelet or removal o Baseline noise were implemented in MATLAB. It is observed that the choice o the cut-o requency is very important, a lower than required cut-o requency does not ilter the actual ECG signal component, however some o the noise successully, but the ECG signal is distorted in the process. Cut-o requency varies corresponding to heart rate and baseline noise spectra. Thus, constant cut-o requency is not always appropriate or baseline noise suppression; it should be selected ater a careul examination o the signal spectrum. Table2. Average power Comparison o various ilters or Removal o noise at ECG sample input 1 Filter Filter Order Average Power beore Filtration Average power ater Filtration Wavelet output at 4dB Wavelet output at 6dB Hamming Kaiser Rectangular Chebyshev Elliptic When FIR ilter with wavelet is applied on signal it can be observe that the combination o Kaiser International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 38

8 Noise Removal rom ECG Signal and Perormance Analysis Using Dierent Filter and wavelet yield the smallest phase delay among all the FIR ilters combination. It can remove the Baseline noises without distorting the waveorm. But the order o ilter is 45.However, high ilter orders are required to obtain this satisactory result and this increases the computational complexity o the ilter. Furthermore, there is signiicant delay in the ilter result, thus this combination can be applied to long data window. Thereore, this combination is appropriate only or oline application, but or real time application, in which short intervals o data is iltered and ast implementation is important, FIR is not an appropriate iltering method.iir and wavelet combination is more appropriate or real time iltering application due to its lower computational complexity, and its better trade-o between average power and spectral density. It completely eliminates the oscillations produced at the starting o the waveorm called ringing eect. For perormance analysis we use dierent baseline noise removal methods or the purpose o comparison. The results are presented in the tabulation orm. From the table it can conclude that it outperorm the other method. REFERENCES [1] Allen, J.; Anderson, J. McC.; Dempsey, G.J.; Adgey, A.A.J., Eicient Baseline Wander Removal or Feature Analysis o Electrocardiographic Body Surace Maps, IEEE proceedings o Engineering in Medicine and Biology Society. vol. 2, pp , [2] Arunachalam, S.P.; Brown, L.F., (Real-Time Estimation o the ECG Derived Respiration (EDR) Signal Using A New Algorithm or Baseline Wander Noise Removal, IEEE Conerence o Engineering in Medicine and Biology Society. pp , 29. [3] Barati, Z.; Ayatollahi, A., Baseline Wandering Removal by Using Independent Component Analysis to Single-Channel ECG data, IEEE conerence on Biomedical and Pharmaceutical Engineering, pp , 26. [4] Carr, J. J. and Brown John M., Introduction to Biomedical Equipment Technology (3rd ed.), Prentice Hall, Inc., [5] Chavan M. S., R.A. Aggarwala, M.D.Uplane, Intererence reduction in ECG using digital FIR ilters based on Rectangular window, WSEAS Transactions on Signal Processing, Issue 5, Volume 4, May, pp.34-49, 28. [6] Chavan M. S., Agarwala R., and Uplane M.D., Suppression o Baseline Wander and power line intererence in ECG using Digital IIR Filter, International Journal O Circuits, Systems And Signal Processing, issue 2,volume 2, 28. [7] Chendeb, M.; Mohamad, K.; Jacques, D., Methodology o Wavelet Pacet Selection or Event Detection, Signal Processing archive vol. 86, issue 12, pp , 26. [8] Dai Min and Liana Shi-Liu, Removal o Baseline Wander rom Dynamic Electrocardiogram Signals, IEEE Conerence on Image and Signal Processing. pp [9] Dansereau, R. M; Kinsnea, W. and V. Clevher, Wavelet Pacet Best Basis Search Using Generalized Renyi Entropy, Proceedings o the IEEE Canadian Conerence on Electrical & Computer Engineering. pp 15-18, 22. [1] Daqrouq, K., ECG Baseline Wandering Reduction Using Discrete Wavelet Transorm, Asian Journal o Inormation Technology, vol. 4. Issue 11, pp , 25. [11] Dhillon S. S., Charabarti S., Power Line Intererence removal From Electrocardiogram Using A Simpliied Lattice Based Adaptive IIR Notch Filter, Proceedings o the 23rd Annual EMBS International conerence, October 25-28, Istanbul, Turey, pp ,21 [12] EE416 Lecture homepage, Last accessed date August 26. [13] Frau D., Nova D, Electrocardiogram Baseline Removal Using Wavelet Approximations, Proceeding o the 15th Biennial Eurasip Conerence Bio signal, pp , 2. [14] Gabbanini, F. Vannucci M., Wavelet pacet methods or the analysis o variance o time series with application to crac widths on the Brunelleschi dome, Journal o Computational & Graphical Statistics. pp , 24. [15] S Salivananan.,AVallavraj C Gnanapriya, Digital Signal processing, Mc Graw Hill,21. International Journal o Innovative Research in Electronics and Communications (IJIREC) Page 39