Electromagnetic Interference in ECG Signals

8/10/2019 Electromagnetic Interference in ECG Signals

1/43

CERT IN INVESTIG TIONS ON

ELECTROM GNETIC INTERFERENCE

IN ECG

Presented by

ARUN B THAHARoll no:T403M.Tech(CommunicationSystems)TKMCE

Guided byDr. UNNI C.

Professor

Dept of ECE

TKMCE


2/43

INTRODUCTION Electromagnetic interference (EMI) is a function of power output and frequency of

transmitting device.

EMI is a self-propagating wave in space with electric and magnetic components.

These components oscillate at right angles to each other and to the direction ofpropagation.

2

EM

Radio Waves, MicroWaves, tetra hertz

radiation

Infra red, Visiblelight, UV radiation,

X& Gamma rays


3/43

The signal voltage of ECG is in milli volt range. Therefore it is very easy to be

influent by the EM environment.

Exponential cellular phone result in level of EM Radiation.

Use of such devices will the productivity but the performance of othermedical equipments working in its vicinity.

3


4/43

PROBLEM STATEMENT In this automated era, ECG devices had undergone developments and

advances in its application.

However, the problem of EMI still exists in ECG signal which will adverselyaffect the accuracy of ECG measurement.

The error thus occurred will lead to a delay or inappropriate treatmentdecision made by the doctor.

This might be a severe problem for those patients who are seriously injuredor to the patients in emergency.

So we should find effective mechanisms to avoid this impact.

4


5/43

OBJECTIVES To study the environment based condition of EMI during ECG acquisition

To study various noises present in the ECG signals

To study Stationary Wavelet Transform (SWT), Wavelet Filtering (WF), WaveletWiener Filtering (WWF) and Adaptive Wavelet Wiener Filtering (AWWF).

To suppress interferences present in ECG signal

5


6/43

LITERATURE SURVEY

6

Works by Stuchlybrothers and Foster on Dielectric properties oftissues and biological materials in 1980s

Impact of Electromagnetic Interference on Critical MedicalEquipments by RF Devices

Work by Hong Ming and co authors in 2006 Work by Janusek and co authors in 2013

Commercially available simulation tools Works on ECG Signal processing


7/43

EXPERIMENT To investigate interference (EMI) of electromagnetic fields from GSM mobile

phones several numerical models of human body from the database of NEVA

Electro-magnetics is used.

This study uses a homogeneous human model with permittivity 50 and conductivity0.5 S/m.

The possibility of charge accumulation by exposing the human body to an electric

field while using GSM mobile phones within in a couple of centimeters is studiedwith the E42 software package

7


8/43

E42 PACKAGE

8


9/43

9


10/43

10


11/43

From the above study it is well clear that the amount of charge accumulated on thehuman body also depend on the posture for a given electric field.

The impact of surface charge accumulation on the human body during the ECGsignal acquisition is a great matter of concern as ECG leads may pick this potential

along with the actual signal resulting in inaccurate diagnosis.

The issue is studied with the help of EM field Solver

11


12/43

EM SOLVER

The basic NEVA EM Solver is robust, fast, and comes with many unique and

detailed human body models. The Solver's available customization is also much

more cost-effective in comparison to most other offerings.

Some Highlights of EM solver are

Small radiation box.

Simulation Environment

Speed

12


13/43


14/43

14


15/43

15

6


16/43

16


17/43

17

8


18/43

OBSERVATIONS

18


19/43

From the above experiments it is clear that the ECG signals might be corrupted byGSM mobile phones working in its vicinity.

It should be noted that even if the mobile phone is not used for any voice or datacommunication it is periodically transmitting low frequency bursts in DTX mode.

The noise level in DTX mode is lower but its characteristics may cause problems asits presence in ECG signal may mimic fibrillation events, makes the situation more

worst.

So we should find out an efficient mechanism to eliminate such interfering signalfrequencies especially at the ELF and LF band signals emitted during DTX mode.

19

20


20/43

WAVELET BASED ADAPTIVE NOISECANCELLATIONDATA SETS

Data base employed to asses the performance of this method is obtained from

MIT-BIH database

The artificial noise used are generated individually for each signal

respecting the original noise level and its time dependence.

20

21


21/43

SWT SWT is a modified version WT

Important parameters- decomposition level, impulse characteristics of initial LPF &HPF.

21

22


22/43

WAVELET FILTERING Based on appropriate adjustment of wavelet coefficients

On the basis of ECG wavelet coefficients it is easy to separate the interference &signal via thresholding

Effective thresholding requires

Evaluate right value of threshold

To choose right thresholding method

22

23


23/43

THRESHOLD LEVEL

x(n) is the noisy signal

x(n) =s(n) + w(n); n=0,1N-1

N length of signalo On transforming using dyadic SWT

ym(n) = um(n) + vm(n)

Threshold level for modifying coefficients is set on the basisof noise level vm.

23

24


24/43

Threshold value is optimized by

If level is too low- Occurrence of noise

If level is too high- damage the noise free signal

Adaptivity in threshold level selection is achieved by

24

25


25/43

WAVELET WIENER FILTERING

25

26


26/43

HW is Wiener correction factor

Using HW modified coefficients are obtained

Y(n) is obtained by taking inverse SWT2

26

27


27/43

ADAPTIVE WAVELET WIENERFILTERING

27

28


28/43

WWF has many parameters that has to be set manually

Appropriate setting of these parameters has great influence on thefiltering results.

Its not clear that which parameter is suitable for ECG

Parameters vary on noise levels

The robust filtering algorithm should change the parameters on the basisof noise level

28

29


29/43

SETTING PARAMETERS

3 important parameters are taken into consideration here

Level of decomposition

Filter banks

Thresholding methods

29

30


30/43

EVALUATION CRITERIA

SNR improvement

30

31


31/43

SYNTHETIC NOISE Most of the authors used white Gaussian noise as artificial

interference

This paper focuses on removing interferences in the LF range.

Therefore noise having power spectrum similar to powerspectrum of interference is created.

Noise = Sin(2*pi*f*length of input)+

SNRin*random(length of input)

31

32


32/43

AWWF ALGORITHM Initialize the necessary parameters

Set SNRin

Noise = Sin(2*pi*f*length of input)+SNRin*random(length of input)

Noisy ECG = i/p ECG+Noise

Set parameters for WWF(threshold level, threshold type, filter

bank,) Perform SWT on noisy ECG

Compute Noise estimate using noisy signal and above o/p.

32

33


33/43

Use parameters above discussed to perform AWWF

Iterate through each parameters in loops to obtain optimumvalues and hence the noise estimate.

Compute the best SNR estimate from stored noise estimate

Filter the ECG signal using the above results.

33

34


34/43

SIMULATION RESULT

34

35


35/43

35

36


36/43

3

37


37/43

37

38


38/43

3

39


39/43

CONCLUSIONS The study shows the worst case level of the interference

induced in ECG electrodes.

Intensity of interference depends on the position of the mobilephone.

The observations from the simulated study shows that duringECG acquisition mobile phones should be either switched off

or it should be located at a distance greater than 15 cm from

any of the ECG electrodes to prevent artifacts in the

electrocardiographic signal.

40


40/43

The proposed algorithm provides better filtering results than otheralgorithm based on WWF algorithm for ELF and LF ranges.

The new algorithm is adaptive and adaptation lies in the division of signalinto individual segments each with constant level of noise.

These segments are filtered using parameters appropriate for given noiselevel.

Thus its evident that the filter can deal with dynamically changingnoise.

41


41/43

REFERENCES1) Stuchly, M. A., and S. S. Stuchly, Dielectric Properties of Biological

SubstancesTabulated, J. of Microwave Power, Vol. 15, No. 1, 1980,

pp. 1926.

2) Foster, K. R., and H. P. Schwan, Dielectric Properties of Tissues and

Biological Materials: A Critical Review,Critical Reviews in Biomedical

Engineering, Vol. 17, No, 1, 1989, pp. 25104.

3) Duck, F. A., Physical Properties of Tissue: A Comprehensive Reference

Book, New York: Academic Press, Harcourt Brace Jovanovich, 1990.

42


42/43

4) Hong ming, Yajun zhang and Weijiang Pan, Evaluation and Removal of

EMI between ECG Monitor and GSM Mobile Phones,ICWMMN 2006

Proceedings.

5) T. Buczkowski, D. Janusek, H. Zavala-Fernandez, M. Skrok, M. Kania,A. Liebert, Influence of Mobile Phones on the Quality of ECG Signal

Acquired by Medical Devices, Measurement Science Review, Vol.13,

No. 5, 2013.

6) P. M. Agante and J. P.Marques de Sa, ECG noise filtering usingwavelets with soft-thresholding methods, Comput. Cardiol., vol. 26, pp.

535538, Sep. 1999. DOI: 10.1109/CIC.1999.826026.

43


43/43

THANK YOU

Electromagnetic Interference in ECG Signals

Documents

Transcript of Electromagnetic Interference in ECG Signals