Application of Interference Canceller in Bioelectricity Signal Disposing

Transcription

1 Available online at Procedia Environmental Sciences 10 (011 ) rd International Conference on Environmental Science and Information Conference Application Title Technology (ESIAT 011) Application of Interference Canceller in Bioelectricity Signal Disposing Liangling GU a, Nanquan ZHOU b, Haotian WU a b * a School of Electronics and Automation, Chongqing University of Technology, Chongqing , China b Department of Electronic Engineering, Chongqing Aerospace Polytechnic College, Chongqing 40001,China Abstract Bioelectricity signal is in strong interfering environment. When it is abstracted, filter is an important hinge. Adaptive interference canceller which based on LMS algorithm is excellent. It can adjust system parameter automatically. When signals are abstracted or disposed, it can play better performance. With this algorithm, this paper dispose ECG(Electrocardiograph)signal as an example in two aspect: canceling power line interference and canceling baseline shift. Both get well effect. 011 Published by Elsevier Ltd. Open access under CC BY-NC-ND license. Selection and/or peer-review under responsibility of Conference ESIAT011 Organization Committee. keywords: LMS algorithm, adaptive filter, ECG, power line interference, baseline shift 1. Introduction When bioelectricity signals such as EEG (electroencephalograph) and ECG (Electrocardiograph) abstracted, they are mixed with plenty of other signals. We call the other signals interfering signals. The main interfering signal is 50Hz electromagnetic signal, because it exists in all kinds of bioelectricity signals and its intensity is greatly more than bioelectricity signals. So it is necessary for us to filter 50Hz interfering signal. There are some technique to remove this interfering signal today, such as reasonable grounding or twisted-pair. These means can reduce interference to a certain extent, but it doesn t get well effects. Moreover, to filter 50Hz power line interference, traditional method is 50Hz notch filter which include * * Corresponding author. Tel.: address: b znq55555@16.com Published by Elsevier Ltd. Open access under CC BY-NC-ND license. Selection and/or peer-review under responsibility of Conference ESIAT011 Organization Committee. doi: /j.proenv

2 Liangling GU et al. / Procedia Environmental Sciences 10 ( 011 ) analog means and digital means. These means can play good role when electrified wire netting is stable. But when electrified wire netting is unstable, it can t get well effect because it can t change when interference change.. LMS adaptive algorithm LMS algorithm which is put forward by Widrow and Hoff get extensive application for its small calculation and easiness to realize. LMS algorithm which based on the steepest descent method is a simple and practical grads estimate means. Its core idea is to use square error e ( n) to replace mean square error (n) Fig. 1. Adaptive noise canceller general diagram Figure 1 shows general diagram of adaptive noise canceller. Signal s which plus an irrelevant noise n 0 form the original input signal of the canceller. Reference signal n 1 is irrelevant to s but it is relevant to n 0. The noise n 1 is processed by an adaptive filter that automatically adjusts its own impulse response through a least-squares algorithm such as LMS that responds to an error signal dependent. The output y produced by the filter is a close replica of n 0. This output is subtracted from the primary input s+n 0 to produce the system output, s+n 0 -y. The aim of LMS algorithm is to obtain output of the system s+n 0 -y which is very close to signal s. Assuming that s,n 0,n 1 and y are statistics steady and have zero mean, according to schematics, output s n0 y (1) Because s is irrelevant to n 0 and n 1, and n 1 is relevant to n 0, its expected value E[ ] E[ s ] E[( n0 y) ] () To minimize [ E ] E[( n ) ], 0 y should arrive minimum value. Then output of the filter y is the minimum square estimate of original noise. The minimum power output of the system is E[ ] E[ s ] (3) Then y=n 0,=s. And noise signal n 0 is cancelled.

3 816 Liangling GU et al. / Procedia Environmental Sciences 10 ( 011 ) Adaptive interference canceller principle Based on above algorithm, here a notch filter realized with adaptive canceller is put forward. It can provide bandwidth which is easy to control and Extremely deep zero point. Moreover, it has ability to trace interference s frequency and phase adaptively, and it has good Robust because it can adjust the system parameters automatically which make the system obtain the best effect. Fig.. 50 Hz adaptive canceller diagram. Figure shows an adaptive interference canceller which has two factors. It is equal to a noise cancelling system which has a complex number factor. In other words, the amplitude and phase of a single frequency sine wave are adjusted with two real factor. Assuming that the original input signal is arbitrary type or the combination of various types, and reference input is pure sine wave at frequency f 0, So x( t) C cos(f 0t ) (4) The first factor s input is reference input and the second factor s input is 90 degree delay of the first factor. It can be expressed as: x1 k C cos( k 0 ) (5) xk C sin( k 0 ) At above formula, 0 f 0 T,which T is sampling period. Using LMS algorithm, at k+1 moment, factor updated as: w w 1, k 1, k 1 w 1k w k x k x In formula 6, is step. Its transfer function is: H ( z) z z C z k 1k k cos( ) 1 (7) z (1 ) cos( C 0 0 ) 1 (6)

4 Liangling GU et al. / Procedia Environmental Sciences 10 ( 011 ) Notch bandwidth is: BW C C rad Hz (8) T Bandwidth divide center frequency makes quality factor: 0 Q C When reference input is sine wave, single frequency adaptive interference canceller is equivalent to a stable notch filter. When reference frequency changed very slowly, adaptive process can adjust right phase relation which canceller need. So the depth of zero is better than that of fixed filter. (9) 4. Power line interference cancellation of bioelectricity We use adaptive interference canceller principle to dispose a set of ECG data which mixed 50Hz power line interference recur to MATLAB. Here step 0.05 is selected. Step has to be selected properly because it decides the stability and astringency of iteration. Figure 3 shows a set of ECG data which mixed 50Hz power line interference. From the figure, basal profile of ECG which has not been disposed can be seen. But it is seriously polluted. For gain clearer wave, it is filtered with above method. From the filtered wave, power line interference has been removed completely less than half of ventricular tachycardia period. Moreover, the wave is very smooth. Fig. 3. ECG waves before and after cancelling power line interference To show superiority of adaptive algorithm, IIR filter is compared. With the Filter Design Toolbox in MATLAB, a second order IIR notch filter is designed and its transfer function is: z z H z) (10) ( z z

5 818 Liangling GU et al. / Procedia Environmental Sciences 10 ( 011 ) So curve of amplitude-frequency characteristic is shown as figure Magnitude (db) Frequency (Hz) Fig. 4. Curve of amplitude-frequency characteristic With this IIR filter, above ECG data has been disposed and the waves before and after filter is shown as figure 5. From the figure, we can see that it should spend longer time to gain steady wave. Moreover, it is more rough than wave filtered with adaptive algorithm. Fig. 5. ECG waves before and after cancelling power line interference with IIR notch filter 5. Elimination of Bioelectricity Signal base line shift In the detection of biological signals, restraining baseline shift is also very important, because it is hard for us to observe data if biological signal is too high. Above adaptive filter could be used as high-pass filter if the notch position is put at zero frequency. In such circumstances, DC or low-frequency drift in

6 Liangling GU et al. / Procedia Environmental Sciences 10 ( 011 ) the original input signal will be removed only one factor is needed and the corresponding reference input signal is arbitrary non-zero DC. Figure 6 shows the restraint of baseline drift in ECG with adaptive interference canceller. From the waveform before filter, all of the ECG data float at the voltage of 1V. After two ventricular tachycardia period s filtering, ECG get to steady and the wave is restricted in the vicinity of 0V. Fig. 6. ECG waveform before and after restraining baseline shift conclusion 6. conclusion The frequency range of bioelectricity signal determines the signal processing is an important issue. Here two interference noise in ECG which include power line interference and baseline interference has been eliminated with adaptive algorithm and both get well effect. Adaptive interference canceller which based on LMS algorithm also can be realized with modern hardware tools such as DSP or EDA. And it is also a hot topic of study today. References [1] Xuefei Yu, principle and design of medical electronic equipment. Guangzhou: Publishing House of South China University,000. [] R. Ramos, A. MBnuel, G. Olivar, E. Trullols and J. Del RioApplication by means of FPGA of an adaptive canceller 50 Hz interference in electro-cardiographyieee Instrumentation and Measurement,Technology Conference Budapest, Hungary, May [3] J. C. Huhta and J. G. Webster, 60-Hz interference in electrocardiography, IEEE Trans. Biomed. Eng., vol. BME-0, pp , Mar [4] J. R. Glover, Adaptive noise cancelling applied to sinusoidal interferences, IEEE Trans. Acoust. Speech Signal Process., vol. ASSP-5, pp. 484, Dec [5] M. J. Shensha, Non-Wiener solutions of the adaptive noise canceller with a noisy reference, IEEE Trans. Acoust. Speech Signal Process. vol. ASSP-8, p. 468, Aug [6] Fuming Shen, Adaptive Signal Processing[M]. Xi an: Publishing House of Xidian University, 001.