lecture ecg noise

11
1 Noise in ECG and how Noise in ECG and how to deal with it to deal with it Djordje Djordje Popovic, MD Popovic, MD Outline Outline Frequency characteristics of ECG Frequency characteristics of ECG Most common sources of noise, Most common sources of noise, characteristics and examples characteristics and examples How to deal with some of them (filtering How to deal with some of them (filtering techniques) techniques)

Upload: jobin-jose

Post on 21-Jul-2016

16 views

Category:

Documents


0 download

DESCRIPTION

ECG biometric technology is still very much in the research phase, and there are many concerns which remain to be investigated. To achieve accurate real time extraction of the identifying features from an ECG trace the effects of noise need to be well understood. Noise introduces artifacts which make it difficult to extract the desired identification data. In this paper the primary noise sources are analyzed and their characteristics identified.

TRANSCRIPT

Page 1: Lecture Ecg Noise

1

Noise in ECG and how Noise in ECG and how to deal with itto deal with itDjordjeDjordje Popovic, MDPopovic, MD

OutlineOutline

Frequency characteristics of ECGFrequency characteristics of ECG

Most common sources of noise, Most common sources of noise, characteristics and examplescharacteristics and examples

How to deal with some of them (filtering How to deal with some of them (filtering techniques)techniques)

Page 2: Lecture Ecg Noise

2

Frequency characteristics of Frequency characteristics of normal ECGnormal ECG

Sources of noiseSources of noise

Power line interferencePower line interferenceElectrode contact noise, baseline drift and Electrode contact noise, baseline drift and motion artifactsmotion artifacts

EMG from the chest wallEMG from the chest wall

Instrumentation noiseInstrumentation noiseElectrosurgical noiseElectrosurgical noise

Page 3: Lecture Ecg Noise

3

Power line interferencePower line interference

60 Hz & multiples60 Hz & multiplesUp to 50% of QRS amplitudeUp to 50% of QRS amplitude

SolutionsSolutions

Notch filter at 60 HzNotch filter at 60 Hz

Comb filterComb filterSampl.rateSampl.rate=n x 60 Hz=n x 60 HzY(iY(i)=)=X(i)X(i)--X(iX(i--nn))

0 0.02 0.04 0.06 0.08 0.1 0.12-250

-200

-150

-100

-50

0

Frequency (kHz)

Mag

nitu

de (d

B)

Magnitude Response (dB)

0 0.05 0.1 0.15 0.2 0.25-70

-60

-50

-40

-30

-20

-10

0

10

Frequency (kHz)

Mag

nitu

de (d

B)

Magnitude Response (dB)

Page 4: Lecture Ecg Noise

4

Adaptive filtering of power humAdaptive filtering of power hum

Electrode contact noiseElectrode contact noise

Loose contactLoose contact

Motion artifactMotion artifact

Baseline driftBaseline driftdue to respirationdue to respiration

Page 5: Lecture Ecg Noise

5

SolutionsSolutions

Typically <0.5 Hz signals (except for the Typically <0.5 Hz signals (except for the abrupt shifts due to motion)abrupt shifts due to motion)

HighHigh--pass filtering with cutpass filtering with cut--off frequency at off frequency at ~ 0.5Hz, realized in hardware or software~ 0.5Hz, realized in hardware or software

Abrupt motion artifact cancellationAbrupt motion artifact cancellation

Adaptive recurrentAdaptive recurrentfilteringfiltering

Page 6: Lecture Ecg Noise

6

EMGEMG

Amplitude Amplitude ~ 10% ECG~ 10% ECG

Broadband Broadband (20 (20 –– 1000 Hz1000 Hz

SolutionsSolutions

LowLow--pass filter with cutpass filter with cut--off frequency off frequency >40Hz>40HzAdaptive filter with Adaptive filter with aVfaVf as primary and as primary and

aVraVr--aVlaVl as referenceas reference

Page 7: Lecture Ecg Noise

7

SummarySummary

Many different sources of noise, frequency Many different sources of noise, frequency content of some of them overlaps with content of some of them overlaps with ECGECG

Limited capabilities of linear filteringLimited capabilities of linear filtering

Final solution dependent on each Final solution dependent on each particular case, and our goalsparticular case, and our goals

Questions?Questions?

Page 8: Lecture Ecg Noise

8

Use of adaptive filters in Use of adaptive filters in ECG processingECG processing

DjordjeDjordje Popovic, MDPopovic, MD

Structure of an AFStructure of an AF

S1/S2 or n1/n2 need not be the same, only correlated in some wayS1/S2 or n1/n2 need not be the same, only correlated in some way !!!!

Page 9: Lecture Ecg Noise

9

Noise cancellationNoise cancellation

Power line interferencePower line interference Baseline driftBaseline drift

EMG cancellationEMG cancellation

Page 10: Lecture Ecg Noise

10

Arrhythmia detection Arrhythmia detection

Goal: to cancel normal QRS complexes, Goal: to cancel normal QRS complexes, and to analyze the signal that remains and to analyze the signal that remains after the cancellationafter the cancellation

Detection of Detection of ectopicectopic beatsbeats

Page 11: Lecture Ecg Noise

11

PP--wave analysiswave analysis

AtrialAtrial fibrillation and/or flatterfibrillation and/or flatter