Bonfring International Journal of Industrial Engineering and Management Science, Vol. 8, No. 2, April 2018 15
Abstract--- Nowadays, cardiovascular diseases are known
as the one of the most dangerous and common problems in
modern society. Analyzing and classifying the ECG signal will
yield an accurate detection of different arrhythmias. Hence
this paper proposes an Integer-Coded Genetic Algorithm
(ICGA) based Particle Clonal Neural Network (ICGA-PCNN)
for classifying the different ECG arrhythmias. Initially the
histogram features and morphological features are extracted
from the Pan-Tompkins based QRS complex. After that the
optimal set of features has been selected using the ICGA for
the extracted features. Then Multilayer feed forward neural
network (MFNN) is used as a classifier to classify the ECG
signal, where the weight and the biased are trained using the
Particle based clonal selection. The MIT-BIH arrhythmias
ECG Database has been presented as a database to train and
test the proposed ICGA-PCNN classifier. The experimental
results show that the proposed classifier approach performs
better than the existing classification approaches in terms of
classification accuracy, sensitivity and specificity.
Keywords--- ECG, Cardiovascular Diseases, Pan-
Tompkins, ICGA-PCNN, Integer-Coded Genetic Algorithm,
MFNN, Particle Swarm Optimization, Clonal Selection.
I. INTRODUCTION
ATA MINING is the process of extracting meaningful
information from various resources which is used to
analyze the decision about particular data. The mining process
used in several applications such as prediction, emotion
classification, hand gesture recognition, heart disease
classification, iris recognition and cancer identification. From
the above applications heart diseases are the most important,
because heart diseases foremost cause of death. Heart disease
factors can be split in two ways such as modifiable and non
modifiable. Modifiable risk factors include obesity, smoking,
lack of physical activity and so on. The non modifiable risk
factors for heart disease are like age, gender, and family
history [1].
Some kind of heart diseases are irregular heartbeat
(arrhythmias), congenital heart defects, weak heart, muscles
(cardiomyopathy), heart valve problems, heart infections and
cardiovascular disease. Cardiovascular disease is the important
cause of death. Cardiovascular disease raised blood pressure,
S. Silvia Priscila, Research Scholar, Bharathiar University, Coimbatore,
India. E-mail:sylviaprem2010@gmail.com
Dr.M. Hemalatha, Research Supervisor, Bharathiar University,
Coimbatore, India. E-mail:hema.bioinf@gmail.com
DOI:10.9756/BIJIEMS.8394
peripheral artery disease, rheumatic heart disease, congenital
heart disease and heart failure problems [2]. In the
cardiovascular diseases may display various syndromes. By
reason of this complexity, there is a need to detect diseases
using diagnostic process. In the cardiovascular diseases may
display various syndromes [3]. Hence, early detection is
considerable to improve the diagnosis of cardiovascular
diseases. As the symptoms are unsteady it is very challenging
to the physicians or radiologists to identify abnormalities if
they diagnose only by their experience. In EU country due to
cardiovascular disease dearth percent in roughly is 45%, and
pay hundred billion Euros for cardiovascular diseases [4].
Several cardiovascular disease cannot be find accurately by
chest X-rays. The Cardiac abnormalities can be identified fast
and efficiently using ECG signals.
Thus, it is important to develop a fully automatic technique
with high sensitivity to assist the early detection of
cardiovascular disease with ECG. In our proposed based
Particle Clonal Neural Network (ICGA-PCNN) for classifying
the different ECG arrhythmias such as Normal, LBBB, APB,
VE and VF to aid in diagnosis. The rest of the paper is
organized as follows: Section 2 describes the recent related
work of automated diagnosis of ECG signals. Section 3
describes the methods and materials. The experimental results
are discussed in the section 4. Finally, section 5 renders the
conclusion.
II. RELATED WORK
Lovepreet Kaur [5] proposed prototype Intelligent Heart
Disease Prediction System with Fuzzy C Means Clustering
algorithm. An intelligent Heart Disease Prediction System
built with the help of miming technique like decision trees,
naive bayes and neural network. The proposed system using
data acquisition, pre-processing, feature extraction, and
classification methods which is used to analysis the heart
diseases in terms of accuracy, time, specificity and sensitivity,
and to make intelligent medical decisions from the traditional
decision support system. This paper have58 records and 14
attributes are used to predict the heart diseases. The prediction
system displays 86.6% accuracy, 32 milliseconds time, 0.44
specificity and 0.45 sensitivity.
Chaitrali S. Dangare et al [6]., proposed mining techniques
such as decision trees, naive bayes and neural networks, in
which to classify the heart diseases from the historical heart
database based on Quality of Service (QoS). Based on the
Quality of Service provides correct and effective treatment. In
this paper takes two input attributes from 13 input attributes
Heart Disease Prediction Using Integer-Coded
Genetic Algorithm (ICGA) Based Particle Clonal
Neural Network (ICGA-PCNN)
S. Silvia Priscila and Dr.M. Hemalatha
D
ISSN 2277-5056 | © 2018 Bonfring