Enhancement of ecg classification using ga and pso

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
_______________________________________________________________________________________
Volume: 04 Issue: 05 | May-2015, Available @ http://www.ijret.org 299
ENHANCEMENT OF ECG CLASSIFICATION USING GA AND PSO
Surbhi Bansal1
, Amritpal Singh2
1
M-Tech Student, Rayat Institute of Engineering & Information Technology, Railmajra, Distt, SBS Nagar, Punjab
(India)
2
Assistant Professor, Rayat Institute of Engineering & Information Technology, Railmajra, Distt, SBS Nagar, Punjab
(India)
Abstract
ECG signal classification utilizes for different predictions of heart diseases. These ECG signals have to be classified using
different frequency bands according to different energy levels for better prediction of features. These signals have to be classified
in different five bands P, Q, R, S and T. These sub-bands provide peak information available in different sub-bands. For the
classification various approaches have to be implemented for filtration of signal. In the purposed work Adaptive filter has been
implemented for the noise reduction from these signals. Classification of the ECG signal has been optimized using Genetic
Algorithm and Particle Swarm Optimization approach. These approaches of classification provide better results i.e. 100 and 100
for 106o
and 119o
respectively for energy levels of ECG signal.
Keywords:- ECG, noise reduction, Genetic Algorithm and Particle Swarm Optimization approach
-------------------------------------------------------------------***-------------------------------------------------------------------
1. INTRODUCTION
An electrocardiogram, which is frequently alluded to just as
an ECG or an EKG, is a symptomatic apparatus that
specialists and medicinal expert’s utilization to gauge a
tolerant heart movement by giving careful consideration to
the electric current streaming in the heart [15]. This is a
strategy that is genuinely standard and is performed
constantly. It by and large takes five to ten minutes to do
and it is both straightforward and protected to perform.
Generally, it is standard for a resting ECG to be managed to
patients. A resting ECG happens when the patient is lying
on his or her back and the specialist or professional spots
metal sensors at the individual's wrists, lower legs and
various places in the midsection zone [12]. The sensors have
the capacity locate the electric driving forces of the heart,
which are then recorded as exceptional tracings on pieces of
chart paper. An ECG is not uncomfortable as the current is
continually advancing just from the patient and from no
place else.
In a general sense a regularly pulsating heart achieves the
same example of waves in everybody. In the event that this
example changes whatsoever [9], it is conceivably because
of a huge number of issues which could incorporate
unpredictable heart rhythms, which could be an indication of
coronary illness however is not so much so; harm to the
muscle of the heart; extension of the assemblies of the heart;
a lopsidedness of minerals in the blood lastly, whether a
patient is having or has as of now had, a heart assault [5].
The vast majority who hint at coronary illness will discover
an ECG helpful in serving to disconnect the issue. It is
paramount to note that an electrocardiogram is not idiot
proof.
It is conceivable to have an ECG that is ordinary yet at the
same time be harrowed by heart issues [16]. The inverse can
likewise be genuine - some of the time the diagram from the
ECG can demonstrate issues where there are none
whatsoever. Atherosclerosis, which is the development of
fat in conduit dividers that causes blocked or limited
coronary courses, is not generally uncovered with a resting
ECG because of the way that when the heart is very still, it
is getting enough oxygen [24]. In this example, a specialist
may choose that an anxiety ECG that takes a gander at the
state of the veins of the heart is in place. An anxiety ECG is
carried out while a patient is either riding a stationary
bicycle or practicing on a treadmill.
An anxiety test or anxiety ECG can give suggestions that
there is a lacking supply of oxygen to particular spots of the
heart muscle [14]. The most widely recognized instance of
this issue is narrowing of coronary courses, which are
stopped up because of the development of plaque. The
anxiety test can reveal issues that would be undetected until
an individual ends up or she experiencing an agony in their
midsection while they are taking part in physical movement.
Fig1: ECG Signal

_______________________________________________________________________________________
Electrocardiograms are regularly a routine piece of a
physical checkup after an individual turns forty years of age.
It is exceptionally proposed that an individual have an ECG
before they turn forty so later on it can be utilized for
examination purposes [21]. The principle objective of any
squeezing method is to accomplish greatest information
volume decrease while safeguarding the huge peculiarities
furthermore recognizing and taking out redundancies in
given information set.
Information layering routines can be grouped into two
classes: 1) lossless and 2) misfortune coding techniques.
Misfortune pressure is helpful where a certain measure of
mistake is satisfactory for expanded layering execution [3].
Misfortune less or data saving clamping is utilized as a part
of the stockpiling of restorative or lawful records. In lossless
information packing, the sign specimens are thought to be
acknowledging of an irregular variable or an arbitrary
procedure and the entropy of the source sign decides the
most minimal clamping degree that can be accomplished.
1.1 Factor Affects ECG
Certain variables or conditions may meddle with or
influence the aftereffects of the test. These incorporate, yet
are not restricted to, the accompanying:
 Obesity, pregnancy, or as refers to
 Anatomical contemplations, for example, the extent
of the midsection and the area of the heart inside
the midsection
 Movement amid the technique
 Exercise or smoking before the method
 Certain drugs
 Electrolyte irregularities, for example, a lot of or
excessively little potassium, magnesium, and/or
calcium.
2. PROPOSED WORK
In the ECG signal classification various approaches has
been utilized for ECG signal classification. Various features
have been selected for classification of different signal used
for ECG signal classification. In classification patterns
different bands have to be calculating on the base of P, Q, R,
S, T. these signal classification has been done using
different Adaptive filters. Adaptive filters removes noise
from different input signals. Different optimization
approaches have been implemented for classification
process to find accuracy at different levels.
In this scenario different energy levels have been
implemented on the basis of different ECG signals.
Classification accuracy is main parameters that have to be
recovered on the basis of signal noise ratio at different peaks
of P, Q, R, S and T.
These classification techniques provide different features for
classification of different ECG signals. For performance
evolution of different ECG signals accuracy have to be
measured on the basis of SNR. In this purposed work PSO
has to be implementing for the optimization process of ECG
signal which is an artificial intelligence for calculation of
fitness value.
The results of various wavelets are compared at different
Activation Energy Level by using GA and PSO
Optimization Technique.
This work flow represents flow of work from loading of
signals to classify different ECG signals. In purposed
classification approach different optimization approaches
have to be implemented to classify different results. These
classification approaches provides better results for feature
extraction of ECG signals.
Fig2: Flow chart of Methodology
3. RESULTS
Simulation has been done using ECG signal for
classification for detection of different peak values has to be
done in this section. ECG signal has been uploaded and then
Principal Component Analysis has to be implemented for
the extraction of different features from loaded signal.

_______________________________________________________________________________________
Fig 3: Wavelet Filter Output
This filter output represents removal of noise from ECG
signal using different HAAR wavelet filter. This reduction
of noise from ECG signal helps for classification that has to
be done
Fig 4: Peak Detection Patterns
The figure 4th
defines Peaks detection using P, Q, R, S and T
at different energy levels provides for signal processing.
This peaks selection provides energy information and
distortion in human health for prediction of status of health.
This signal processing has been done using various
classification approaches.
Fig 5: ECG signal in P and S domains
The 5th
figure depicts ECG signal against time in P and S
domain only. In this figure different band has been used for
the spectrogram of ECG signal. These different bands are P,
Q, R, S and T. this representation of signals has been done at
different P and S domain in signal processing band.
Fig 6: Classified Signals
This graph illustrates the classified signal after processing
using Particle Swarm Optimization approach.
Table 1 Classification Accuracy
Wavelet
Total Accuracy (%)
106o
114o
116o
119o
217o
Average
HAAR
BASE PAPER 88.63 98.66 99.54 100 92.79 95.924
PROPOSED WORK 100 100 100 100 99.5 99.9
Meyer
BASE PAPER 90.9 99.13 99.63 100 95.27 96.986
PROPOSED WORK 100 99.9 98.5 100 99 99.48
Sym3 BASE PAPER 93.33 98.6 99.83 100 96.02 97.556

_______________________________________________________________________________________
PROPOSED WORK 97.5 100 99.9 100 98.2 99.12
Bior3.9
BASE PAPER 90.65 98.38 99.63 100 95.77 96.886
PROPOSED WORK 100 99.8 99.2 100 99.3 99.66
Db3
BASE PAPER 90.01 98.82 99.54 100 93.53 96.38
PROPOSED WORK 100 100 100 100 98.2 99.64
Db5
BASE PAPER 91.1 98.7 99.58 100 96.26 97.128
PROPOSED WORK 99.7 100 99.8 100 99.5 99.8
Db8
(NW1)
BASE PAPER 94.02 99.25 99.58 100 96.02 97.774
PROPOSED WORK 99.4 99.1 99.7 99 99.2 99.28
Particle optimization approach use different particle values
to find out fitness value for each parameter used for the
selection of different peaks of ECG signal. PSO utilize
different signal processing tool for the calculation of
different ECG signals fitness values.
This table comprises of different accuracy values comprised
at different energy levels using different classification
approaches. This shows classification accuracy using
different artificial intelligence approaches that are GA and
PSO. This comparison table results provides different
accuracies using GA and PSO at various energy levels.
4. CONCLUSION
An ECG is used to measure the heart’s electrical conduction
system. It picks up electrical impulses generated by the
polarization and depolarization of cardiac tissue and
translates into a waveform. The ECG is a one of the
important physiological signal which depicts the electrical
activity of a heart. ECG processing is a topic of great
interest in the scientific community because based on the
ECG’s a diagnosis is done for detecting abnormalities in the
heart functioning. Basically, a data coding algorithm seeks
to minimize the number of code bits stored by reducing the
redundancy present in the original signal.
In this we have many parameters like feature extraction, Q R
S detected wave form, filtered output, to classify the GA
(Genetic Algorithm) with the help of signals in time domain,
GA (Genetic Algorithm) with the help of Fourier transform,
GA (Genetic Algorithm) with the help of Spectrogram, GA
(Genetic Algorithm) with the help of PSD of signals, GA
(Genetic Algorithm) with the help of PSD of signals with
Accuracy. In this PSO (Particle Swarm Optimization) has
been implemented to find out fitness value of different EEG
signal classification approaches. Features extracted for
detection of QRS signal
5. REFERENCES
[1] Dae-Seok Lee ―An ECG Analysis on Sensor Node
for Reducing Traffic Overload in u-Healthcare with
Wireless Sensor Network‖, ISSN 978-1-4244-1262-
4, Page No. 256 – 259, IEEE, 2007.
[2] Elbuni, a ―ECG Parameter Extraction Algorithm
using (DWTAE) Algorithm‖, ISSN 978-0-7695-
3892-1, Page No. 57 – 62,IEEE,2009.
[3] Kai-chao Yao ―Multi-Function ECG Measurement
System for Design and Implementation‖, ISSN 978-
1-4244-5543-0, Page No. 643 – 646, IEEE, 2009.
[4] Dusit Thanapatay, ―ECG beat classification for ECG
printout with PCA and SVM‖, IEEE, 2010.
[5] Annam, J.R. ―Time series clustering and the Analysis
of Electro Cardiogram heart-beats using Dynamic
Time warping‖, ISSN 978-1-4577-1110-7, Page No.
1 – 3, IEEE, 2011.
[6] Dongxin Lu ―A new system of electrocardiogram
diagnose based on telemedicine‖ ISSN 978-1-61284-
485-5, PP 374 – 377, IEEE, 2011.
[7] Edla, S. ―Electrocardiogram signal modeling using
interacting multiple models‖ ISSN1058-6393, PP
471 – 475, IEEE, 2011
[8] Sajedin, A ―Electrocardiogram beat classification
using classifier fusion based on Decision Templates ‖
ISSN 978-1-4673-0687-4, PP 7 – 12,IEEE, 2011.
[9] Yang Liu ―Enhancing interoperability of ECG
machine to support ECG telediagnosis service ‖
ISSN 978-1-4244-9351-7, Page No. 1093 –
1096,IEEE,2011.
[10] Binfeng Xu ―Extracting Fetal Electrocardiogram
based on a modified fast independent component
analysis‖ ISSN 978-1-4673-0025-4,PP 1787 – 1791,
IEEE, 2012.
[11] Carvalho, H.H ―An electrocardiogram diagnostic
system implemented in FPGA‖ ISSN 978-1-4673-
2476-2, PP 1 – 5, IEEE, 2012.
[12] Dedhia, P. ―Low cost solar ECG with Bluetooth
transmitter‖ ISSN 978-1-4577-1990-5, Page No. 419
– 423, IEEE, 2012
[13] Gwo Giun Lee ―Gabor feature extraction for
electrocardiogram signals‖ ISSN 978-1-4673-2292-8,
PP 304 – 307, IEEE, 2012
[14] Tarmizi, I.A ―A journal of real peak recognition of
electrocardiogram (ECG) signals using neural
network‖, ISSN 978-1-4673-0733-8, PP 504 – 509,
IEEE, 2012.
[15] Wang, J.J ―Transformations for estimating body
surface potential maps from the standard 12-lead
electrocardiogram‖ ISSN 2325-8861, PP 17 –
20,IEEE, 2012
[16] Yong-Hong Hsu ―Health care platform based on
acquisition of ECG for HRV analysis ‖ ISSN 978-1-
4673-0312-5,Page No 464 – 469,IEEE,2012.

_______________________________________________________________________________________
[17] Lashgari, E. ―Manifold learning for ECG arrhythmia
recognition‖ ISSN 14211208, Page No. 126 – 131,
IEEE, 2013.
[18] Ravindra kumar, S ―Fetal ECG extraction and
enhancement in prenatal monitoring — Review and
implementation issues‖, ISSN 978-1-4244-9007-3,
Page No.16 – 20, IEEE, 2013.
[19] Vullings, R. ―Novel Bayesian Vectorcardiographic
Loop Alignment for Improved Monitoring of ECG
and Fetal Movement‖ ISSN 0018-9294, Page No.
1580 – 1588, IEEE, 2013.
[20] Shin Chi Lai ―Low-Cost and Low-Complexity
Electrocardiogram Signal Recorder Design Based on
Arduino Platform‖ ISSN 978-1-4799-5389-9, PP 309
– 312, IEEE, 2014
[21] Sharma, L.N ―Denoising pathological multilead
electrocardiogram signals using multiscale singular
value decomposition‖ ISSN 2157-0981, PP 1 – 5,
IEEE, 2014
[22] Yoon Geon Kim ―Radiation Efficiency-Improvement
Using a Via-Less, Planar ZOR Antenna for Wireless
ECG Sensors on a Loss Medium‖, ISSN 1536-1225,
Page No. 1211 – 1214, IEEE, 2014.

Enhancement of ecg classification using ga and pso

More Related Content

What's hot (19)

Similar to Enhancement of ecg classification using ga and pso (20)

More from eSAT Journals (20)

Recently uploaded (20)

Enhancement of ecg classification using ga and pso