You are on page 1of 28

Technion Israel Institute of Technology

Department of Electrical Engineering


VLSI Systems Research Center & Laboratories

Personal Heart
Attack Detector
Nir Gluzman
Alexei Iolin
March 3, 2005
1

Did you know?


More than 1.2 million Americans will suffer a
heart attack, where about 502,000 of these
people will die!
340,000 of them in ER or they reach a hospital.
Surviving a heart attack depends on the
treatment that is given within the first hour of the
heart attack.
Most common reason for treatment delay is
patient unawareness.
Source:

American Heart Association, 2004 estimation.


2

The solution

Personal Heart
Attack Detector

AGENDA
ECG waveform.
Heart attack & ST-Elevation.
Project objective.
System block diagram.
Background on wavelets.
DWT and filter-bank.
Detection of ECG Characteristic points with DWT.
SIMULINK implementation.
Summary.

ECG waveform
ECG = Electrocardiogram
Recording of hearts electrical activity.
Measures rate and regularity.
Used to diagnose diseases.

ECG waveform
Basic ECG complex
R

T
P
Q
S

Heart attack & ST-Elevation


Heart attack
(Myocardial Infarction)
An event that results in permanent damage to the
heart.
myocardium = heart muscle.
infarct = die.
Q-waves and inverted T-waves
are two diagnostic wave forms
seen in old heart attacks and new
heart attacks respectively.
7

ST-Elevation (STEMI)
Elevation of ST-segment relative to PR or TP
segment.

ST-elevation is a sign for coming heart attack


(acute myocardial infarction).
Short PR interval - very difficult for computer
algorithms to find the isoelectric reference level.
8

Project objective
Develop a chip for measuring ST-elevation and
alerting the user to heart attack.
Detection algorithm is implemented in hardware for
power efficiency.

Winter semester objective


Develop an algorithm for detection of ECG
characteristic points.
Develop an algorithm for detection of ST-elevation.
Implement the above in SIMULINK for feasibility
demonstration.
9

System block diagram


Biosensors

A/D
DWT algorithm for
detection of ECG characteristic points
Detection of
ST-Elevation
ST-Elevation
alert

10

Background on wavelets
Fourier Transform (FT)
Break down a signal into constituent sinusoids of
different frequencies.
Transform the view of the signal from time-base to
frequency-base.

Source:

Wavelet applications using MATLAB toolbox, SIPL.

11

Background on wavelets
Short Time Fourier Transform (STFT)
Analyze a small section of a signal, based on FT
and using windowing.
STFT maps a signal into a two-dimensional function
of time and frequency.

Source:

Wavelet applications using MATLAB toolbox, SIPL.

12

Background on wavelets
Wavelet Transform (WT)
STFT with variable size window.
Long time intervals when low frequency information
is needed.
Short time intervals when high frequency
information is needed.

Source:

Wavelet applications using MATLAB toolbox, SIPL.

13

Wavelet main advantage

LOCAL ANALYSIS
Sinusoid with a small discontinuity
1

0 .5

-0. 5

-1

20

40

60

80

1 00

time
14

Fourier coefficients

Wavelet coefficients

45
40

scale

35
30
25
20
15
10
5
0
0

10

20

30

40

50

60

frequency

70

80

90

100

20

40

time

60

80

100

Wavelet transform can help to detect


signal discontinuities !!
15

DWT and Filter-Bank


CWT is continues in two means:
Scale/frequency.
Time shift (position).
CWT is very hard for implementation by hardware.
Discrete Wavelet Transform (DWT) chooses only a
subset of scales and positions, based on power of
two (dyadic choice).
Mallat was the first to implement DWT by two
channel subband coder (filter bank).
Filter bank can be easily implemented by hardware.

16

DWT and Filter-Bank


Mallats fast implementation for DWT

17

Detection of ECG
Characteristic points with DWT
Algorithm is based on the article:

Detection of ECG characteristic points


using wavelet transform (1995)
The article uses a CWT algorithm and implements it
on a 486-DX33 PC with C-language.
Based on Mallats work we develop a system that
implements the detection algorithm with DWT.

18

Power spectrum of QRS complex, P- and T-waves

QRS complex can be separated from other


interfering signals
Source:

Estimation of QRS complex power spectra for design of a QRS filter,


IEEE Transaction of biomedical engineering, Nov. 1984

19

SIMULINK implementation

20

SIMULINK implementation
Filter bank output
R
1.5

ECG original
signal
1

Voltage [mV]

ECG approx.
0.5

d4

-0.5

4.2

4.3

4.4

4.5

4.6

4.7

4.8

4.9

Time [sec]
21

Summary
DWT algorithm for ECG
detection was presented.

characteristic

points

Algorithm is implemented with filter-bank.


Detection of ECG characteristic points is a crucial
stage before calculating ST-elevation.
Project involves utilization of sophisticate signal
processing algorithm for medical problem.

22

Questions?

23

ECG is a representation of an electrical activity


created by depolarization and repolarization of the
atria and ventricles

24

Existed solutions for


heart monitoring & heart attack alert
Holter heart monitor
Continuous recording of heart rhythm during normal
activity.
Holter monitoring is used
to determine how the
heart responds to normal
activity.

25

Heart rate monitoring for


sport activities
Heart rate gives a precise measurement of exercise
intensity.
Very common with professional athletes and
amateurs.

Source:

Polar Electro.

26

Triage Cardiac Panel


Diagnosis of heart attack based on blood test.
Results delivered in approximately 15 minutes.

Source:

Biosite inc.

27

Automated External Defibrillator


(AED)
Jump-start a heart that has stopped beating, almost
always due to a heart attack.
The machine won't give a shock unless it detects
that a person's heart has stopped beating and is just
fluttering.

Source:

Philips HeartStart Defibrillators.

28

You might also like