Download Slide 1

BME 452 Biomedical Signal
Processing
Lecture 6
Biological Signals and Event
Detection
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
1
Lecture 6 Outline








Introduction to ECG ,PCG and CP
Event Detection:
• Introduction
• Problem statement
• Detection of events and waves
• QRS detection
– Derivative-based methods
– Pan-Tompkins algorithm
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
2
The Electro Cardio Gram (ECG)
 ECG: electrical manifestation of of heart
recorded from the body-surface
• Heart rate monitoring
• Wave shape change due to cardiovascular
disease and abnormalities.
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
3
Normal ECG
- Slow P wave: 0.1-0.2 mV
60-80 ms
-PQ segment: AV delay 60-80
ms. Isoelectric
- QRS complex: sharp biphasic
or triphasic wave of about
1 mV amplitude and 80ms
duration
- ST segment: 100-120ms.
Isoelectric
- Slow T wave: 0.1-0.3mV
and duration 120-160 ms
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
4
 Rectangular calibration pulse
– 1 mV amplitude and 200 ms duration
Produce a pulse of 1cm height on
the paper plot.

ECG normaly is between 0.05-100
Hz and sampling rate is 500Hz or 1
kHz

12-lead ECG
I,II,II,aVR,aVL,aVF,
V1,V2,V3, V4, V5, V6.
Einthoven’s triangle
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
5
Arrhythmias
 Disturbances in regular rhthym
 Irregular firing patterns from SA node
 Abnormal or additional pacing activity
from other parts of the heart
 VF: Ventricular fibrillation
- Ineffective pumping -> possibly death
- Disorganized contraction of ventricles
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
6
ECG Abnormalities
 Premature beats (PVC), ectopic beats
 ST segment depression or elevation
- Ischemia: reduced blood supply to a tissue
due to a block in an artery.
- Infarction: dead tissue -> incapable of
contraction
 QRS widening
- Hypertrophy
- Bundle-branch block
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
7
Arrhythmic ECG traces
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
8
The Phono Cardio Gram (PCG)
• Stethoscope:heart sound listening device
• PCG: vibration or sound signal
- Contractile activity of the heart and blood together
– Represents a recording of heart sound signal
• Measurement:
- Requires a transducer to convert vibration or sound signal
into an electronic signal
– Microphones or pressure transducers placed on chest
• Diagnosis:
– Cardiovascular diseases and defects cause changes and
additional sounds and murmurs.
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
9
PCG signal
• S1 occurs at the onset of
ventricular contraction
- Corresponds in timing to
the QRS complex in the
ECG signal
• S2 is caused by the
closure of the semilunar
valves (aortic and
pulmonary valves)
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
10
Heart murmurs
• Occur due to certain cardiovascular defects and diseases
• Are caused by turbulance in blood flow
– Valvular stenosis: deposition of calcium or other reasons,
the valve leaflets are stiffened and dont open completely
– Insufficiency: valves can not close effectively and causes
leakage of blood
• The intervals between S1 and S2, S2 and S1 of the next
cycle are normally silent.
• Are high-frequency, noise like sounds that arise when the
velocity of blood becomes high as it flows through an
irregularity (such as a constriction)
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
11
CP: Carotid Pulse
• Pressure signal recorded over carotid artery.
– Near the surface of the body at the neck
• Provides arterial blood pressure and blood
volume with each heart beat
• Similar morphology of the pressure signal at
the root of the aorta
- But can not measure absolute pressure
• Is useful with PCG
- Can assist in the identification of S2 and its
components
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
12
CP signal
- P: percussion
wave
- T: tidal wave
- D: dicrotic
notch
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
13
Introduction to Event Detection
 Biomedical signals carry signatures of physiological events
• Part of a signal related to a specific event of interest is
referred to as an “ epoch”
• Analysis requires identification of epochs
– For monitoring and diagnosis
• The corresponding waveform may be segmented and
analyzed in terms of its
– Amplitude, waveform, time-duration, intervals between
events, energy distribution, frequency content
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
14
Problem statement
• Given a biomedical signal, identify
discrete signal epochs and correlate
them with events in the related
physiological process
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
15
Normal ECG
• Slow P wave: 0.1-0.2 mV
60-80 ms
• PQ segment: AV delay
60-80 ms
– isoelectric
• QRS complex: sharp
biphasic or triphasic wave
of about 1 mV amplitude and
80 ms duration
• ST segment: 100-120 ms
– Isoelectric
• Slow T wave: 0.1-0.3 mV
and duration 120-160 ms
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
16
PCG signal
• S1 occurs at the onset of
ventricular contraction
– Corresponds in timing to
the QRS complex in the
ECG signal
• S2 is caused by the closure
of the semilunar valves
(aortic and pulmonary
valves)
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
17
EEG signals
• Delta waves
– 0.5<= f < 4 Hz, appear at deepsleep stages
• Theta waves
– 4 <= f < 8 Hz, appear at the
beginning stages of sleep
• Alpha waves
– 8 <= f < 13 Hz, principal resting
rhythm
– Auditory and mental arithmetic
tasks with eyes closed
• Beta waves
– f > 13 Hz, background activity in
tense and anxious subjects
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
18
Detection of Events and Waves
• QRS detection
– Derivative- based methods
– Pan-Tompkins algorithm
• Correlation analysis of EEG channels
– Detection of EEG rhythms
– Template matching for EEG spike-and
wave detection
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
19
Detection of Events and Waves
• Matched filter
• P-wave detection
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
20
Applications
 ECG rhythm analysis
 ECG Waveform Classification
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
21
QRS Detection
• Derivative-based methods
• Pan-Tompkins algorithm
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
22
Derivative-based methods
• QRS might not always be the highest wave in
a cardiac cycle
– artifacts may upset the peak search
algorithm
• QRS complex has the largest slope (rate of
change of voltage)
• Rate of change = derivative operator (d/dt )
• Derivative operator:
– P and T waves will be suppressed
– Output is the highest at the QRS
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
23
Derivative-based algorithm
• Balda et al proposed an algorithm
– Three-point first derivative
• y0[n] = | x[n] – x[n-2] |
– Second derivative
• y1[n] = | x[n] – 2x[n-2] + x[n-4] |
– The two results are weighted and combined as y2[n]
– The result y2[n] is scanned with a threshold of 1.0
– Whenever threshold is crossed
• Subsequent 8 samples also tested against the same
threshold
• If at least one pass the threshold test
– The segment of eight samples is taken to be a part of a
QRS complex
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
24
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
25
The Pan-Tompkins algorithm
• Pan and Tompkins proposed a realtime QRS
detection algorithm based on
– Slope, amplitude, and width of QRS
complexes
Bandpass
Filter
Differentiator
Squaring
Operation
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
Moving
Integrator
26
Algorithm details
• Recursive LPF
–H(z) = (1/32)( (1-z-6)2 )/( (1-z-1)2 )
–y[n] = 2 y[n-1] - y[n-2] + (1/32)[ x[n]2x[n-6]+x[n-12] ]
• Sampling rate = 200Hz, fc = 11 Hz
• Filter introduces 5 samples of delay (25
ms)
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
27
Algorithm details
• HPF
• Allpass filter minus a LPF
– H_lp(z) = (1-z-32)/(1-z-1)
– y[n] = y[n-1] + x[n] - x[n-32]
• H_hp(z) = z-16 – (1/32)H_lp(z)
– p[n] = x[n-16] – (1/32)[y[n-1] + x[n] x[n-32]]
• fc =5 Hz
• Filter introduces 80ms of delay
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
28
Derivative operator
• y[n] = (1/8) [ 2x[n] + x[n-1] – x[n-3] –
2x[n-4] ]
–Approximates the ideal d/dt operator up
to 30 Hz
• Suppresses P and T waves
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
29
Squaring Operation
• Makes the result positive and emphasizes
large differences resulting from QRS
complexes
• Small differences arising from P and T
waves are suppressed
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
30
Integration
• Multiple peaks within the duration of a
single QRS complex
• Smoothing of the output of the preceding
operations through a moving window
integration filter
– y[n] = (1/N) [ x[n – (N-1)] + x[n – (N-2)
+ …+ x[n] ]
– N: window width ( 30 found to be suitable
for fs=200 Hz)
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
31
Adaptive thresholding
• Thresholding procedure adapts to
changes in ECG signal by computing
running estimates of signal and noise
peaks
• A peak is said to be detected whenever
the final output changes direction within
a specified interval
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
32
Lecture 6 BME452 Biomedical Signal Processing 2013 (copyright Ali Işın, 2013)
33