IOSR Journal of VLSI and Signal Processing (IOSR-JVSP)

Volume 7, Issue 6, Ver. I (Nov.-Dec. 2017), PP 01-12

e-ISSN: 2319 – 4200, p-ISSN No. : 2319 – 4197
www.iosrjournals.org

Automatic Detection of Cardiac Arrhythmia through ECG Signal

Analysis: A Review
*Gopisetty Ramesh1 D. Satyanarayana2 M. Sailaja3
1
Research Scholar, JNT University, Kakinada, A.P., India
2
Professor, ECE Department, RGM College of Engineering & Technology, Nandyal, A.P., India
3
Professor, ECE Department, JNT University College of Engineering, Kakinada, A.P., India
Corresponding Author: *Gopisetty Ramesh1

Abstract: Analyzing the characteristics of an ECG signal plays a vital role in the detection of various
Cardiovascular Diseases (CVDs). However the analysis of ECG signal is not a simple task. In this paper a
broad survey is carried out over various approaches focused to analyze the characteristics of ECG signal to
perform automatic Cardiac Arrhythmia detection. Initially, the details of ECG acquisition, Characteristics of
ECG and the possible arrhythmias based on the abnormalities in the ECG signal are discussed. Based on the
step by step execution of the system model, all the approaches are classified as preprocessing approaches,
feature extraction approaches and classification approaches. These approaches are further classified in
different ways based on the methodology used for accomplishment. This paper also discussed about various
standard databases which are used for the implementation of developed approaches.
Keywords: ECG, Cardiovascular Diseases, Cardiac Arrhythmia, Feature Extraction, MIT-BIH.
---------------------------------------------------------------------------------------------------------------------------------------
Date of Submission: 03-11-2017 Date of acceptance: 21-11-2017
----------------------------------------------------------------------------------------------------------------------------- ----------

I. Introduction
In recent years, due to the drastic change in the lifestyles of population across the world, various health
related problems are propagated enormously. Especially with the 20 th century the cardiovascular diseases
(CVDs) have become the leading cause of mortality in India. According to the survey carried out by World
Health Organization (WHO), in Western populations only 23% of CVD deaths occur before the age of 70 years;
in India, this number is 52% [1], [2]. Among the possible CVDs, Cardiac Arrhythmia (CA) is one of the leading
heart diseases, is the main cause of about half of deaths due to cardiovascular disease or about 15% of all deaths
globally. Cardiac Arrhythmia [3], also known as irregular heartbeat, is a group of conditions in which
the heartbeat is irregular, too fast (above 100 beats per minute, called as tachycardia), or too slow (below 60
beats per minute, called as bradycardia). Mainly there are four main types of arrhythmia:
 Extra beats,
 Supraventricular tachycardias,
 Ventricular arrhythmias,
 Brady arrhythmias.
Extra beats include premature atrial contractions, premature ventricular contractions, and premature
junctional contractions. Supraventricular tachycardias include atrial fibrillation, atrial flutter, and paroxysmal
supraventricular tachycardia. Ventricular arrhythmias include ventricular fibrillation and ventricular tachycardia
[4]. Arrhythmias are due to problems with the electrical conduction system of the heart. All the above
mentioned arrhythmias are life-threatening, which may cause sudden cardiac arrest and even death if timely
therapy is not conducted within a few minutes. A high quality, easily implementable, fast Cardiac Arrhythmia
detection algorithm will help achieve a high probability of survival from out-of-hospital heart attack incidents.
Hence there is a need of develop an efficient and feasible CA detection technique which helps in the accurate
diagnosis.
In general, the diagnosis of heart related issues such as the proper or malfunctioning are detected
through the Electrocardiogram (ECG) signal [5]. An ECG signal characterizes the electrical activities of a heart,
which are recorded through several electrodes attached to the skin. This quasi-periodic signal contains valuable
information on the functioning of a heart and can be used for the detection of heart disease. The automatic
detection of arrhythmia and distinguishing them from normal heart rhythms could be very useful for an early
detection of heart disease, especially in real time.
Various approaches are proposed in earlier to perform automatic arrhythmia detection based on the
characteristics of ECG signal. Since the automatic detection is a computer aided task, provision of most
DOI: 10.9790/4200-0706010112 www.iosrjournals.org 1 | Page
 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

significant features of ECG is very important by which the accurate diagnosis is possible. The earlier approaches
focused on various aspects like some focused on preprocessing, some on feature extraction and some on
learning techniques. This paper provides a complete literature survey about the earlier developed approaches.
Rest of the paper is organized as follows; Section II gives the basic details of ECG signal. Section III illustrates
the details of earlier proposed approaches and section IV concludes the paper.

II. ECG Signal Analysis

1. ECG Acquisition
The heart is a muscle that contracts in a rhythmical manner, pumping blood throughout the body. This
contraction has its beginning at the atrial sine node that acts as a natural pace-maker, and propagates through the
rest of the muscle. This electrical signal propagation follows a pattern [6]. As a result of this activity, electrical
currents are generated on the surface of the body, provoking variations in the electrical potential of the skin
surface. These signals can be captured or measured with the aid of electrodes and appropriate equipment. The
difference of electrical potential between the points marked by the electrodes on the skin, usually is enhanced
with the aid of an instrumentation (operational) amplifier with optic isolation. Then, the signal is submitted to a
high-pass filter; and as a second stage, submitted to an anti-aliasing low-pass filter. Finally, it appears in an
analogical to digital converter. The graphical registration of this acquisition process is called electrocardiogram
(ECG). A simplified block diagram of ECG acquisition is shown in fig.1.

Fig.1: Architecture for acquisition of an ECG

2. ECG Segments
An ECG signal can be segmented into heartbeats. Each heartbeat consists of five standard waves
labeled with the letters P, Q, R, S and T [7]. These waves indicate the depolarization and the re-polarization
phases of heart muscles. Besides, five more inter-wave timings called PR, PR segment, QRS, QT, ST segment
are used. These intervals are indicated on Fig.2.

Fig.2: Main characteristics of an ECG curve

Now we give a brief introduction about the role of these clinical features:
P wave: P waves are usually a low-amplitude feature that represents the depolarization of the atria prior to atrial
contraction. They are hard to detect, but important to distinguish various cardiac arrhythmias.
QRS complex: It reflects the depolarization of the ventricles. This is the most significant wave of the ECG due
to the large muscle mass of the ventricles. So it can be easily detected and often used to determine the heart rate.
T wave: The T wave represents the re-polarization of the ventricles. It is a recovery phase of the cardiac muscle.
The shape of this wave carries a lot of information about cardiac abnormalities. So it is important to analyze its
geometrical properties such as symmetry, asymmetry and slope.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 2 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

PR interval: It is the time elapsing between the beginning of the P wave and the beginning of the next QRS
complex. It reflects conduction through the AV node1.
PR segment: The PR segment is the flat, usually isoelectric segment between the end of the P wave and the start
of the QRS complex. Most of the delay in the PR segment occurs in the AV node.
ST segment: Represents the period from the end of ventricular depolarization to the beginning of ventricular re-
polarization. ST level shifts are significant markers of cardiac abnormalities
QT interval: It represents the time between the start of ventricular de-polarization and the end of ventricular re-
polarization. The QT interval is inversely proportional to heart rate: shortens at faster heart rates and lengthens
at slower heart rates.

3. Cardiac Arrhythmias
Since the ECG signal is the main contributor in the diagnosis of CA, different types of abnormalities in the
rhythm of ECG declares different types of CAs. Table.1 gives the details of CAs and their respective
abnormalities in the characteristics of ECG signal [8].

Table.1: Differentiation between different types of CAs

Name of CA Heart Condition ECG Abnormality
Normal
Sinus
rhythm

Impulses originates at S-A All complexes are normal, evenly spaced, rate 60-100/min
node at normal rate
Sinus
bradycardia

Impulses originates at S-A All complexes are normal, evenly spaced, rate < 60/min
node at slow rate
Sinus
tachycardia

Impulses originates at S-A

node at rapid rate All complexes are normal, evenly spaced, rate > 100/min
Sinus
arrhythmia

Impulses originates at S-A All complexes are normal, rhythm is irregular, longest R-
node at varying rate R interval exceeds shortest > 0.16s.
Wandering
pacemaker

Impulses originates from

varying points in atria Variation in P-wave contour, P-R and P-P interval
therefore in R-R interval.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 3 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

Atrial Flutter

Impulses travel in circular

course in atria Rapid flutter waves, ventricular response irregular
Atrial
fibrillation

Impulses have chaotic,

random pathways in atria Baseline irregular, ventricular response irregular
Junctional
rhythm

Impulses originates at AV
node with retrograde and P-wave is often inverted, may be under or after QRS
antegrade direction complex, Heart rate is slow
Premature
ventricular
contraction

A single pulse originates at Time interval between R peaks is multiple of R-R interval
right ventricle
Ventricular
tachycardia

Impulses originates at
ventricular pacemaker Wide ventricular complexes, Rate > 120/min
Ventricular
fibrillation

Chaotic ventricular
Rapid, wide irregular ventricular complexes
depolarization

4. Available Databases
Various databases are composed of ECG signals with various types of arrhythmias. The use of five databases is
recommended by the standardization:
MIT-BIH [102]: The Massachusetts Institute of Technology – Beth Israel Hospital Arrhythmia Database (48
records of 30 min each). The recordings were digitized at 360 samples per second per channel with 11-bit
resolution over a 10 mV range. About half (25 of 48 complete records, and reference annotation files for all 48
records) of this database has been freely available
EDB [103]: The European Society of Cardiology ST-T Database (90 records of 2h each). This database consists
of 90 annotated excerpts of ambulatory ECG recordings from 79 subjects. The subjects were 70 men aged 30 to
84, and 8 women aged 55 to 71. Each record is two hours in duration and contains two signals, each sampled at
250 samples per second with 12-bit resolution over a nominal 20 millivolt input range.
AHA [104]: The American Heart Association Database for Evaluation of Ventricular Arrhythmia Detectors (80
records of 35 min each). As for the records in the AHA Database, the data consist of a 3-hour recording of two
ECG signals, for which the last 30 minutes are annotated beat-by-beat.
DOI: 10.9790/4200-0706010112 www.iosrjournals.org 4 | Page
 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

CU [105]: The Creighton University Sustained Ventricular Arrhythmia Database (35 records of 8 min each).
This database includes 35 eight-minute ECG recordings of human subjects who experienced episodes of
sustained ventricular tachycardia, ventricular flutter, and ventricular fibrillation.
NST [106]: The Noise Stress Test Database (12 records of ECG of30 min each, plus 3 records with noise
excess)

III. Literature Survey

Based on the above discussion, the automatic detection of Cardiac Arrhythmia involves the ECG signal
preprocessing, Feature Extraction and Classification phases. The preprocessing phase involves removing the
unwanted noises and interferences in the ECG signal. The feature extraction phase involves the extraction of
significant features which represents the detailed analysis of ECG and further the classification phase involves
the detection of type of arrhythmia based on the features of ECG. Here the literature survey is also carried out in
the same fashion. I.e., initially, the earlier proposed approaches focused on the removal of noise are illustrated
and followed by the approaches focused on feature extraction and finally the approaches belong to
classification. A simple block diagram for the automatic detection of arrhythmia through ECG signal processing
is represented in fig.3.

Fig.3: Simple block diagram of Cardiac Arrhythmia (CA) detection through ECG signal processing

1. Preprocessing
Since real ECG signals are noisy (i.e. white and mains noise) and contaminated with artifacts (i.e.
electromyography signals due to breathing and chest movement) the first step generally consists of band pass
filtering the measured signals.
Among all proposals for reducing noise in ECG signals, the simplest and most widely used is the
implementation of recursive digital filters of the finite impulse response (FIR) [9], [10], which was made
computationally possible with the advance in microcontrollers and microprocessors. Since these filters allow
quick and easy application of reject band filter, they works well for the attenuation of known frequency bands
like the noise added due to the electrical network (frequency range is about 50-60 Hz). However the main
problem is that the frequency of the noise is not known always. This problem is solved by designing the
adaptive filters for various frequencies of the signal. However the undiscriminating use of filters, i.e., low-pass
and high pass filters distorts the signal’s morphological attributes and makes them as unsuitable for the CA
diagnosis. The architectures of [11-13] applied adaptive filters for noise removal form the ECG signal. Least
Mean Square (LMS) Filter is an adaptive which a san ability to remove the unknown frequencies. Ravina [14]
used the LMS filter to de-noise the ECG signal in an adaptive fashion. However, this technique has constraints
and does not offer great advantages over the FIR digital filters.
In the last decade, many methods based on wavelet transforms have been employed to remove noise,
since they preserve ECG signal properties avoiding loss of its important physiological details and are simple
from a computational point of view [15-22]. Sayadi and Shamsollahi [18] proposed a modification of the
wavelet transform called the multi-adaptive bionic wavelet transform and it was applied to reduce noise and
baseline variation of the ECG signal. This method presented superior results when compared to the ones based
on the traditional wavelet transform. Chen et al. [19] use a wavelet denoising stage based on a discrete wavelet
transform, with three levels of decomposition, as the first processing stage for real-time QRS complex detection.
Thus a wavelet denoising operation appears to be suitable for on-line operation while maintaining the ECG
features for further processing stages. In [21], Savitzky-Golay filter and Discrete Wavelet Transform (DWT) are
being used to de-noise ECG signal and a comparison is provided between two methods.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 5 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

Some more approaches are also proposed including nonlinear Bayesian filters [23], extended Kalman
filtering [24] to remove the noise from the ECG and these approaches measured the performance in terms of
signal to noise ratio. Lannoyet.al., [25] used two median filters to remove the baseline wander.One median filter
of 200ms width to remove QRS complexes and P-waves and other of 600 ms width to remove T-waves. Then
the resulting signal is filtered again with 1 12-tap, low-pass FIR filter with 3-dB point at 35 Hz. A similar
method is accomplished in [26-28] for the removal noises in ECG.Baziet.al., [29] proposed the use of high pass
filter for noise artifacts and a notch filter for power network noise. Lin and Yang [30] uses a second order low
pass filter and two median filters. In [31], the signal is subtracted by its mean and then normalized. Escalona-
Moranet al. [32] used the raw wave i.e., no preprocessing is applied.
2. Feature Extraction
The feature extraction stage is the key to the success in the heart beat classification of the arrhythmia
using the ECG signal. Any information extracted from the heartbeat used to discriminate its type maybe
considered as a feature. The features can be extracted in various forms directly from the ECG signal’s
morphology in the time domain and/or in the frequency domain or from the cardiac rhythm.
Most of the research work focused on the extraction of RR interval. The RR interval is a time period between
two successive R peaks. With exception of patients that utilize a pacemaker, the variations perceived in the
width of the RR interval are correlated with the variations in the morphology of the curve, frequently provoked
by arrhythmias [36]. Thus, the features in the RR interval have a great capacity to discriminate the types of
heartbeats and some authors have based their methods only on using the RR interval features [33-35].
Not only has the RR interval features, some approaches focused on the extraction of other features also.
Among those QRS interval, or the duration of the QRS complex is the most utilized feature. In [37] the ECG
signal is denoised to remove the artifacts and analyzed using Wavelet Transform to detect the QRS complex and
arrhythmia. A similar process for arrhythmia detection is carried out in [38] through the detection of QRS
complex. ECG data was filtered out first and after removing artifacts, QRS complexes were identified. For each
QRS complex its R-peak, slope, sharpness and duration were calculated. Along with these approaches, a new
approach is developed in [39] for intuitive and robust real time QRS detection based on the physiological
characteristics of the electrocardiogram waveform. The proposed algorithm finds the QRS complex based on the
dual criteria of the amplitude and duration of QRS complex. It consists of simple operations, such as a finite
impulse response filter, differentiation or thresholding without complex and computational operations like a
wavelet transformation. Along with these techniques [40-43] are also focused on the extraction of ECG signal
feature alone and combined. In [44], a new method based on the continuous wavelet transform is described in
order to detect the QRS, P and T waves. QRS, P and T waves may be distinguished from noise, baseline drift or
irregular heartbeats. Firstly, our algorithm is validated using fifty 12 leads ECG samples from the CinC
collection. The samples have been chosen in the "acceptable records" list given by Physionet. The detection and
the duration delineation of the QRS, P and T waves given by [44] are compared to expert physician results.
A location, width and magnitude (LWM) [80] model is proposed for extracting each wave's features in
the ECG. The model is a stream of Gaussian function in which three parameters (the expected value, variance
and amplitude) are applied to approximate the P wave, QRS wave and T wave. Moreover, the features such as
the P-Q intervals, S-T intervals, and so on are easily obtained. Then, a mixed approach is presented for
estimating the parameters of a real ECG signal. To illustrate this model's associated advantages, the extracted
parameters combined with R-R intervals are fed to three classifiers for arrhythmia diagnoses. Two kinds of
arrhythmias, including the premature ventricular contraction (PVC) heartbeats and the atrial premature
complexes (APC) heartbeats, are diagnosed from normal beats using the data from the MIT-BIH arrhythmia
database.
Features extracted from the domain of time/frequency [82] together with the features of the RR interval
appear as part of the methods that produced the highest accuracies. The simplest way to extract features in the
time domain is to utilize the points of the segmented ECG curve, i.e., the heartbeat, as features [45]. However,
the use of samples of the curve as features is a technique that is not very efficient, since besides producing a
vector of the features with high dimensions (depending on the amount of samples used to represent the
heartbeat), it suffers from several problems related to the scale or displacement of the signal with respect to the
central point (peak R).
Aiming at reducing the dimension of the feature vector, various techniques have been applied directly
on the samples that represent the heartbeat (in the neighborhood of the R peak) as principal component analysis
(PCA) [46-48], [84, 85] or independent component analysis (ICA) [49, 50], [85], or the combination of PCA
and ICA [51, 52], [85]in which new coefficients are extracted to represent the heart beat. Hani [52] presents a
comparative study between the use of PCA and ICA to reduce the noise and artifacts of the ECG signal and
showed that PCA is a better technique to reduce noise, while ICA is better one to extract features. The ICA
technique enables statistically separate individual sources from a mixing signal. The ECG is a mix of several
action potentials and each action potential could be strongly related to an arrhythmia class. The rationale behind

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 6 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

ICA for ECG heartbeat classification is to separate the action potentials sources as well as the noise sources. The
PCA technique separates the sources according to the energy contribution to the signal.
Another technique based on PCA, the Kernel Principal Component Analysis (KPCA), was used by
Devy et al. [53]. In that work, a comparison between PCA and KPCA was performed and it was concluded that
KPCA is superior to the PCA technique for classifying heartbeats from the ECG signal. According to Kallas et
al. [54], KPCA performs better, due to its nonlinear structure. Asl et al. [55] used Generalized Discriminant
Analysis (GDA) to reduce the dimensions of the features of the heartbeat interval type to classify rhythmic
arrhythmias. However, the authors did not take care to separate the heartbeats of the same patient used during
training and testing (intra-patient paradigm), which is a serious concern discussed further. The inter-patient
paradigm should be considered for a more realistic scenario.
Although various techniques have been considered, most of the studies presented in literature use
wavelet transform and researchers claim that this is the best method for extracting features from the ECG signal
[57, 58]. Saniet.al., [59] has proposed a robust ECG feature extraction technique suitable for mobile devices by
extracting only 200 samples between R-R intervals as equivalent R-T interval using Pan Tompkins algorithm at
preprocessing stage. The discrete wavelet transform (DWT) of R-T interval samples are calculated and the
statistical parameters of wavelet coefficients such as mean, median, standard deviation, maximum, minimum,
energy and entropy are used as a time-frequency domain feature. Amruthadevi [60] focused on the suggested
Discrete Wavelet Transform (DWT) in processing ECG recordings and also to extract certain attributes. The
process of feature extraction and dimensionality reduction can be effectively performed using Principal
Component Analysis (PCA). Besides DWT, continuous wavelet transform (CWT) has also been used to extract
features from the ECG signals [61], since it overcomes some of the DWT drawbacks, such as the coarse-ness of
the representation and instability.[62] Presents a classification method using Support Vector Machine (SVM)
algorithm. The noise and some electrical disturbances during measurement also affect the signal feature
measurements. So, the signal is transformed into another domain using Wavelet Transformation method
(Continuous Wavelet Transform (CWT) to be precise) to extract certain features of the signal and study their
pattern while comparing the abnormal ECG signal with that of a normally running ECG signal. However, CWT
is not largely used due to the fact that its implementation and its inverse are not available in standard toolboxes
(such as MATLAB wavelet Toolbox) and CWT should be carefully discretized for the use as a CWT analyzer.
3. Classification
Once the set of features has been defined from the heartbeats, models can be built from these data using
artificial intelligence algorithms from machine learning and data mining domains [64-66] for arrhythmia
heartbeat classification.
The four most popular algorithms employed for this task and found in the literature are: support vector machines
(SVM) [54] [62], [83], artificial neural networks (ANN) [67], [71], [75], [78] and linear discriminant (LD) [63],
and Reservoir Computing with Logistic Regression (RC) [68]. Since the most of the research work is carried out
through the ANN and SVM techniques the following section illustrates the proposed approaches based on those
three techniques.
3.1 Artificial Neural Network (ANN)
The ANN architectures mostly used for arrhythmia classification are Multilayer Perceptrons (MLP) and
Probabilistic Neural Networks (PNN). According to Yu and Chen [69], models constructed with PNN are
computationally more robust and efficient than the traditional MLP.A feed forward multilayer neural network
(NN) with error back-propagation (BP) [70] learning algorithm was used as an automated ECG classifier to
investigate the possibility of recognizing ischemic heart disease from normal ECG signals. The proposed ECG
classification in [72] is supervised by ANN. The ECG waveform gives the almost all information about activity
of the heart, which is depending on the electrical activity of the heart. In [72] only five features of ECG signal P,
Q, R, S, T are focused. This is achieved by extracting the various features and duration of ECG waveform P-
wave, PR segment, PR interval, QRS Complex, ST segment, T-wave, ST- interval, QTC and QRS voltage.
Mitraet.al., [73] attempts correlation-based feature selection (CFS) with linear forward selection search. For
classification, [73] used incremental back propagation neural network (IBPLN), and Levenberg-Marquardt
(LM) [76] classification tested on UCI data base. Some more approaches are proposed by combining ANN with
other algorithms. According to Osowski et. al., [74], a combination of classifiers not only reduces the overall
error in the neural networks, but also reduces the incidence of false negatives.
3.2 Support Vector Machine (SVM)
SVM is found to be a most popular and efficient classifier for the classification of ECG signals to
detect cardiac arrhythmias. A novel life-threatening arrhythmias detection algorithm is presented in [77] by
combining the SVM with previously proposed ECG parameters A total of 13 parameters were computed
accounting for temporal (morphological), spectral, and complexity features of the ECG signal. A filter-type
feature selection (FS) procedure was proposed to analyze the relevance of the computed parameters and how
they affect the detection performance. Nitinajibhaskar [78] focused to classify an ECG signal as healthy subject

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 7 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

or subject diagnosed with Myocardial Infarction (MI) using Artificial Neural Networks (ANN) and SVM
(Support Vector Machine). LIBSVM is utilized for the classification with SVM and back propagation artificial
neural networks with varying hidden layers and nodes are also implemented for performance analysis.
Compared to the extraction of feature sin time domain, the features extracted through the transform domain
illustrates gives the more information about the features. Qin et.al., [79] combined the DWT with SVM to
perform arrhythmia beat classification. In classification, 12-element feature vectors characterizing six types
of beats are used as inputs for one-versus-one support vector machine, which is conducted in form of 10-
fold cross validation with beat-based and record-based training schemes.
Since SVM presents a negative behavior for imbalanced classes, database balancing techniques for the training
phase, which are little explored for this problem, can be studied in future research, as for example, more
sophisticated sampling techniques.
3.3 Logistic Regression
RC computing models are dynamical models aiming to process a time series signal in two parts:
represent the signal through a non-adaptable dynamic reservoir and a dynamic readout from the reservoir. More
details regarding RC can be found in [87]. Chaurasia V et.al., [88],developed prediction models for heart disease
survivability by implementing data mining algorithms CART (Classification and Regression Tree), ID3
(Iterative Dichotomized 3) and decision table (DT) extracted from a decision tree or rule-based classifier to
develop the prediction models using a large dataset.
This approach is achieved efficient results and also declared that this can be implemented on hardware due to its
less complexity and it is much suitable for heart beat classification.
3.4 Other Techniques
Many other methods for arrhythmia classification have been developed using other machine learning and data
mining algorithms, such as nearest neighbors [89, 90], clustering [93], decision tree [95-97]etc.
In [89], an automatic ECG beat is classified into 2 categories-Normal and Premature ventricular
contraction using Dempster Shafer Theory (DST). In biomedical signal classification problems, the cost of
making an erroneous decision can be high. Deferring a decision rather than taking a wrong decision might be
beneficial. This is done by using the evidential k nearest neighbors (EKNN) approach which is based on
Dempster Shafer Theory for classifying the ECG beats. RR interval features are used. Analysis is done on the
MIT-BIH database. Performance evaluation is done by considering error rates. However K-NN algorithm is not
much used for the problem of arrhythmia classification, since their efficiency is intimately connected to
previous knowledge to perform the classification of each sample that is represented by the complete training set,
which leads to a high computational cost during the testing phase.
Clustering techniques are widely used along with Artificial Neural Networks. Some works used
unsupervised clustering techniques to agglomerate all of the heart beats in the record of a given patient into
clusters [91] and the final classification of each cluster, i.e., the heart-beats of that group, is then defined by a
human specialist. Abawajy et.al., [92] proposed a novel multistage clustering algorithm that combines various
procedures for dimensionality reduction, consensus clustering of randomized samples and fast supervised
classification algorithms for processing of the highly dimensional large ECG datasets. An unsupervised method
based on relevance analysis to improve ECG heartbeat clustering is described in [94]. A new feature matrix
projection method for unsupervised relevance analysis is described. The proposed scheme computes weighting
feature values that enable data dimensionality reduction along with a proper feature relevance ranking and uses
a least squares optimization of the input feature matrix in a single iteration. Large datasets tend to be sparse thus
making it very hard to identify structure in the dataset for clustering based on distance measures. Another
challenge is that the dataset often contains noisy and/or irrelevant features that may mislead clustering
algorithms. Generally, these challenges are addressed by coupling clustering algorithms with dimensionality
reduction approaches. In this regard, a number of advanced consensus functions for clustering ensembles have
been developed recently. However, many of the existing consensus functions are computationally expensive so
their application for clustering large and highly dimensional datasets such as ECG signals is impracticable.
Methods that use a decision tree allow an interpretation of the decisions made by the model. In [98],
arrhythmia beat classification using ensemble decision tree is studied. Bootstrap aggregating (bagging) decision
tree is used as a type of ensemble learning. ECG signals from 22 patients including five arrhythmia beats and
normal beats are obtained from MIT-BIH arrhythmia database. After the filtering process, 56569 ECG beats are
collected and feature are extracted based on morphological properties including RR, FF, RR and FF ratio to
previous values (RRR, FFR), RR and FF differences from mean values (RRM, FFM). 25% of 56569 beats is
used as test data for bagged decision tree and the rest for training. However, this type of method is not efficient
for continuous features (belonging to a set of real numbers) and feature vectors of large dimensions. Thus,
methods that use decision trees consider only a few features.
HMM is widely used to audio and speech signal analysis and recognition. Andreaoet al. [99] validated
the use of HMM for ECG analysis in medical clinics. A novel ECG classification approach based on HMM

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 8 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review

model is proposed by Wei Lianget.al.,[100]. In ECG preprocessing, an integral-coefficient-band-stop (ICBS)

filter is applied, which omits time-consuming floating-point computations. In addition, two-layered Hidden
Markov Models (HMMs) are applied to achieve ECG feature extraction and classification. The periodic ECG
waveforms are segmented into ISO intervals; P sub wave, QRS complex and T sub wave respectively in the first
HMM layer where expert-annotation assisted Baum-Welch algorithm is utilized in HMM modeling. Then the
corresponding interval features are selected and applied to categorize the ECG into normal type or abnormal
type (PVC, APC) in the second HMM layer [101].

IV. Conclusion
Cardiac arrhythmia occurs spasmodically at the early stages of heart disease by which the diagnosis
will become difficult. If these CA is not detected in early stages, the effect of treatment will become ineffective
at the advanced stages. In addition some types of CAs like tachyarrhythmia are associated with sudden dead,
occurring less than an hour after the onset of symptoms. Hence the major part of biomedical research is directed
towards the development of an effective ECG signal diagnosing equipment to detect the CAs in the early stages
only and making the effective heart disease treatment.
This paper focused on the earlier approaches developed with the aim of accurate diagnosis of various
CAs through ECG signal. Since the ECG signal carries the most significant information of the status of heart,
i.e., proper or malfunctioning, analysis of the entire characteristics of ECG signal gives better results. For this
purpose the entire system is divided into three phases such as preprocessing, feature extraction and
classification. Initially the approaches which are focused towards the preprocessing of ECG signal are
discussed. All these approaches aimed to remove the unwanted noise added in the ECG signal. Further the
approaches mainly focused on the feature extraction are discussed. In summary the total feature extraction
approaches are categorized as time domain and transform domain. From the survey it was summarized that,
compared with time domain features, transform domain features gives the better results in the ECG diagnosis.
Finally the approaches mainly focused in the optimization of classification are discussed. These methods include
the machine learning algorithms, clustering algorithms and data mining approaches etc. Based upon the above
survey, the concluding remarks can be outlined as follows;
Results presented in literature usually use the MIT-BIH database that is extremely unbalanced. However, this
aspect has been ignored by authors that use the intra-patient scheme.
Various approaches employed the semi-automatic approaches to enhance the results of diagnosis. These semi-
automatic approaches can improve the results around 40% even with less number of features. However the main
drawback is that they demand the expert intervention.
Various machine learning approaches have shown that the size/diversity of the database used for the
construction of methods impacts more than the choice of the learning algorithm and/or employed techniques.

References
[1]. Harikrishnan S, Leeder S, Huffman M, Jeemon P, Prabhakaran D, A Race against Time: The Challenge of Cardiovascular
Disease in Developing Economies. 2nd ed. New Delhi, India: New Delhi Centre for Chronic Disease Control; 2014.
[2]. Xavier D, Pais P, Devereaux PJ, Xie C, Prabhakaran D, Reddy KS, Gupta R, Joshi P, Kerkar P, Thanikachalam S, Haridas KK,
Jaison TM, Naik S, Maity AK, Yusuf S; CREATE registry investigators. Treatment and outcomes of acute coronary syndromes
in India (CREATE): a prospective analysis of registry data. Lancet., 371, 2008, pp: 1435–1442.
[3]. A.S. Adabag, G. Peterson, F.S. Apple, J. Titus, R. King, R.V. Luepker, “Etiology of sudden death in the community: results of
anatomic, metabolic, and genetic evaluation”, Am. Heart. J. 159, 2010, pp: 33–39.
[4]. J.J. Goldberger, A.E. Buxton, M. Cain, O. Costantini, D.V. Exner, B.P. Knight, D.Lloyd-Jones, A.H. Kadish, B. Lee, A. Moss,
R. Myerburg, J. Olgin, R. Passman, D.Rosenbaum, W. Stevenson, W. Zareba, D.P. Zipes, “Risk stratification for arrhythmic
sudden cardiac death: identifying the roadblocks”, Circulation, 123, 2011, pp: 2423–2430.
[5]. M. Velic, I. Padavic, S. Car, “Computer aided ECG Analysis – State of the Art and Upcoming Challenges”, 2013,
[6]. M.S. Spach, J.M. Kootsey, “The nature of electrical propagation in cardiac muscle”, Am. J. Physiol. Heart Circ.Physiol, 1983,
pp: 3–22.
[7]. Peter Kovacs, “ECG Signal Generator based on Geometrical Features”, Annales Univ. Sci. Budapest., Sect. Comp. 37, 2012, pp:
247-260.
[8]. Macfarlane PW, Lawrie TDV, Comprehensive Electrocardiology: Theory and Practice in Health and Disease, Pergamon Press,
New York, 1st ed., Vols. 1, 2, and 3, 1989, pp. 1785.
[9]. K sravankumar, P Rajesh Kumar, “Removal of noise from electrocardiogram using digital FIR and IIR filters with various
methods”, International Conference on Communications and Signal Processing (ICCSP), 2015.
[10]. S. Sundar, “Filtering Noise from Electrocardiogram using FIR filter with CSD Coefficients”, International conference on
Innovations in Information, Embedded and Communication Systems, 2014.
[11]. Mohammad Zia, “Denoising ECG Signals Using Transform Domain Adaptive Filtering Technique”, Annual IEEE India
Conference (INDICON), 2009.
[12]. J Jenitta, “Denoising of ECG signal based on improved adaptive filter with EMD and EEMD”, IEEE Conference on Information
& Communication Technologies (ICT), 2013.
[13]. ChinmayChandrakar, Kowar, “Denoising ECG Signals Using Adaptive Filter Algorithm”, International Journal of Soft
Computing and Engineering (IJSCE), Vol 2, Issue 1, March 2012.
[14]. Ravina Bhatia , SupriyaGoelb , GurjitKaurc and PradeepTomard, “Denoising of ECG using Adaptive Filter Algorithm”,
International Journal of Control Theory and Applications, Vol 9, Number 46, 2016.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 9 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review
[15]. B.N. Singh, A.K. Tiwari, “Optimal selection of wavelet basis function applied to ECG signal denoising”, Digit. Signal Process,
Vol.16, Issue.3, 2006, pp:275–287.
[16]. S.-W. Chen, H.-C.Chen, H.-L. Chan, “A real-time QRS detection method based on moving-averaging incorporating with wavelet
denoising”, Comput. Method Programs Biomed. Vol.82, 2006, pp: 187–195.
[17]. A.E. Zadeh, A. Khazaee, V. Ranaee, “Classification of the electrocardiogram signals using supervised classifiers and efficient
features”, Comput. Method Programs Biomed, Vol.99, 2010, pp: 179–194.
[18]. O. Sayadi, M.B. Shamsollahi, “Multi adaptive bionic wavelet transform: application to ECG denoising and baseline wandering
reduction”, EURASIP J. Adv. Signal Process, Vol.14, 2007, pp: 1–11.
[19]. S.W. Chen, H.C. Chen, H.L. Chan, “A real-time QRS detection method based on moving-averaging incorporating with wavelet
denoising”, Comput. Meth.Prog. Bio, Vol.82, 2006, pp: 187–195.
[20]. MounaimAqil, “ECG Signal Denoising by Discrete Wavelet Transform”, IJOE, Vol. 13, No. 9, 2017.
[21]. HarjeetKaur, “ECG Signal Denoising with Savitzky-Golay Filter and Discrete Wavelet Transform (DWT)”, International
Journal of Engineering Trends and Technology (IJETT), Vol36 Number 5, June 2016.
[22]. SaifEddineHadji, “Wavelet-based Performance in Denoising ECG Signal”, ICSPS 2016.
[23]. R. Sameni, M.B. Shamsollahi, C. Jutten, G.D. Clifford, “A non-linear Bayesian filtering framework for ECG denoising”, IEEE
Trans. Biomed. Eng., 2007, pp: 2172–2185.
[24]. O. Sayadi, M. Shamsollahi, “ECG Denoising and Compression using A Modified Extended Kalman Filter Structure”, IEEE
Transactions on Biomedical Engineering, 2008, pp: 2240-2248.
[25]. G. de Lannoy, D. Franc¸ois, J. Delbeke, M. Verleysen, “Weighted conditional random fields for supervised inter-patient
heartbeat classification”, IEEE Trans. Biomed. Eng., 2012, pp: 241–247.
[26]. T. Mar, S. Zaunseder, J.P. Martínez, M. Llamedo, R. Poll, “Optimization of ECG classification by means of feature selection”,
IEEE Trans. Biomed. Eng., 2011, pp:2168–2177.
[27]. Z. Zhang, J. Dong, X. Luo, K.-S.Choi, X. Wu, “Heartbeat classification using disease-specific feature selection”, Comput. Biol.
Med., Vol 46, 2014, pp: 79–89.
[28]. Z. Zhang, X. Luo, “Heartbeat classification using decision level fusion”, Biomed. Eng. Lett., 2014, pp: 388–395.
[29]. Y. Bazi, N. Alajlan, H. AlHichri, S. Malek, “Domain adaptation methods for ECG classification, International Conference on
Computer Medical Applications (ICCMA), 2013, pp: 1–4.
[30]. C.-C. Lin, C.-M.Yang, “Heartbeat classification using normalized RR intervals and morphological features”, Math. Problem
Eng., 2014, pp: 1–11.
[31]. H. Huang, J. Liu, Q. Zhu, R. Wang, G. Hu, “A new hierarchical method for inter-patient heartbeat classification using random
projections and RR intervals”, Biomed. Eng., 2014, pp: 1–26.
[32]. M.A. Escalona-Moran, M.C. Soriano, I. Fischer, C.R. Mirasso, “Electrocardiogram classification using reservoir computing with
logistic regression”, IEEE J. Biomed. Health Inform. 19(3) (2015) 892–898.
[33]. ChandrakarKamath, “A Novel Approach to Arrhythmia Classification Using RR Interval And Teager Energy”, Journal of
Engineering Science and Technology Vol. 7, No. 6, 2012, pp: 744 – 755.
[34]. Tsipouras MG, “An arrhythmia classification system based on the RR-interval signal”, Artificial Intelligence in Medicine,
Volume 33, Issue 3, March 2005, pp: 237-250.
[35]. R.G. Kumar, Y.S. Kumara swamy, “Investigation and classification of ECG beat using input output additional weighted feed
forward neural network”, International Conference on Signal Processing, Image Processing &Pattern Recognition (ICSIPR),
2013, pp: 200–205.
[36]. G.D. Clifford, F. Azuaje, P. McSharry, “Advanced Methods and Tools for ECG Data Analysis”, 1st ed., Artech House
Publishers, 2006.
[37]. Kaur, “ECG Signal Analysis and Arrhythmia Detection using Wavelet Transform”, Journal of the Institution of Engineers,
Volume 97, Issue 4, 2016, pp: 499–507.
[38]. Yusuf Khan, “Arrhythmia detection based on derivative analysis of QRS complex”, 3rd International Conference on Signal
Processing and Integrated Networks (SPIN), 2016.
[39]. Jinkwon Kim, Hangsik Shin, “Simple and Robust Real time QRS Detection Algorithm Based on Spatiotemporal Characteristic
of the QRS Complex”, Research article, PLOS ONE, March 4, 2016.
[40]. A. Peterkova, M. Stremy, “The raw ECG signal processing and the detection of QRS complex”, IEEE European Modeling
Symposium, 2015.
[41]. Tekeste, “Adaptive ECG interval extraction”, IEEE International Symposium on Circuits and Systems (ISCAS), 2015.
[42]. DeboleenaSadhukhan, “R-peak detection algorithm for ECG using double difference and RR interval processing”, Procedia
Technology, 2012, pp: 873 – 877.
[43]. Hussain A. Jaber AL-Ziarjawey and IlyasÇankaya, “Heart Rate Monitoring and PQRST Detection Based on Graphical User
Interface with Matlab”, International Journal of Information and Electronics Engineering, Vol. 5, No. 4, July 2015.
[44]. MaximeYochum, Charlotte Renaud, SabirJacquir, “Automatic detection of P, QRS and T patterns in 12 leads ECG signal based
on CWT”, Biomedical Signal Processing and Control, Elsevier, 2016.
[45]. C. Wen, T.-C.Lin, K.-C.Chang, C.-H. Huang, “Classification of ECG complexes using self-organizing CMAC”, Measurement,
Vol 42, issue 3, 2009, pp: 399–407.
[46]. Nadi Sadr, “A fast approximation method for principal component analysis applied to ECG derived respiration for OSA
detection”, IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society (EMBC), 2016.
[47]. Jukka A Lipponen1 and Mika P Tarvainen, “Principal component model for maternal ECG extraction in fetal QRS detection”,
Physiological Measurement, Volume 35, Number 8, 2014.
[48]. Daniel Raine, “Principal component analysis of atrial fibrillation: Inclusion of posterior ECG leads does not improve correlation
with left atrial activity”, Med Eng Phys. Feb., Vol 37, issue 2, 2015, pp: 251–255.
[49]. M. Sarfraz, A.A. Khan, F.F. Li, “Using independent component analysis to obtain feature space for reliable ECG arrhythmia
classification”, IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 2014, pp: 62–67.
[50]. Sarfraz, “Role of independent component Analysis in intelligent ECG signal Processing”, PhD thesis, University of Salford,
2014.
[51]. Mayank, “ECG signal decomposition using PCA and ICA”, National Conference on Recent Advances in Electronics &
Computer Engineering (RAECE), 2015.
[52]. DhaniDharmaprani, “A comparison of independent component analysis algorithms and measures to discriminate between EEG
and artifact components”, IEEE 38th Annual International Conference of the Engineering in Medicine and Biology Society
(EMBC), 2016.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 10 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review
[53]. Devy, “Application of Kernel Principal Component Analysis for Single-lead-ECG-Derived Respiration”, IEEE Transactions on
Biomedical Engineering, Volume: 59, Issue: 4, April 2012.
[54]. M. Kallas, C. Francis, L. Kanaan, D. Merheb, P. Honeine, H.Amoud, “Multi-class SVM classification combined with kernel
PCA feature extraction of ECG signals”, International Conference on Telecommunications (ICT), 2012, pp: 1–5.
[55]. FatemehShahbaz, “Generalized discriminant analysis for congestive heart failure risk assessment based on long-term heart rate
variability”, Published by Elsevier Inc. 2015.
[56]. Yogendra, “Human recognition using Fisher's discriminant analysis of heartbeat interval features and ECG morphology”, Elsveir
science publishers, Volume 167 Issue C, November 2015.
[57]. C. Lin, Y. Du, T. Chen, “Adaptive wavelet network for multiple cardiac arrhythmias recognition”, Expert Syst.Appl., Vol 34,
issue 4, 2008, pp: 2601–2611.
[58]. Y. Kutlu, D. Kuntalp, “Feature extraction for ECG heartbeats using higher order statistics of WPD coefficients”, Comput.
Method Program Biomed., Vol 105, issue 3, 2012, pp: 257–267.
[59]. SaniSamu, “Wavelet feature extraction for ECG beat classification”, IEEE 6th International Conference on Adaptive Science &
Technology (ICAST), 2014.
[60]. Amutha Devi C, “Effective ECG beat classification using colliding bodies’’, An International Journal of Medical Sciences, 2017.
[61]. P.S. Addison, “Wavelet transforms and the ECG: a review”, Physiol. Meas., Vol 26, issue 5, 2005, pp: 155–199.
[62]. TowfeeqFairooz and HediKhammari, “SVM classification of CWT signal features for predicting sudden cardiac death”,
Biomedical Physics & Engineering Express, Volume 2, Number 2, 2016.
[63]. M. Llamedo, J.P. Martínez, “Heartbeat classification using feature selection driven by database generalization criteria”, IEEE
Trans. Biomed. Eng., Vol 58, 2011, pp: 616–625.
[64]. R.O. Duda, P.E. Hart, D.G. Stork, Pattern Classification, 2nded., Wiley-Inter science, 2000.
[65]. C.M. Bishop, Pattern Recognition and Machine Learning, 1sted., Springer, 2006.
[66]. S. Theodoridis, K. Koutroumbas, Pattern Recognition, 4thed., Elsevier, 2009.
[67]. E.D. Übeyli, “Combining recurrent neural networks with eigenvector methods for classification of ECG beats”, Digit. Signal
Process, 2009, pp: 320–329.
[68]. M.A. Escalona-Moran, M.C. Soriano, I. Fischer,C.R. Mirasso, “Electrocardiogram classification using reservoir computing with
logistic regression”, IEEE J. Biomed. Health Inform.,Vol 19, issue 3, 2015, pp: 892–898
[69]. S.-N. Yu, Y.-H. Chen, “Electrocardiogram beat classification based on wavelet transformation and probabilistic neural network”,
Pattern Recogn. Lett.,Vol 28, issue 10, 2007, pp: 1142–1150.
[70]. Sahar H. El-Khafifand Mohamed A. El-Brawany, “Artificial Neural Network-Based Automated ECG Signal Classifier”, ISRN
Biomedical Engineering Volume 2013 (2013).
[71]. Andrew, “Classification of the ECG Signal Using Artificial Neural Network”, Proceedings of the 3rd International
Conference on Intelligent Technologies and Engineering Systems, 2014, pp: 545-555.
[72]. Gaurav Kumar Jaiswal and Ranbir Paul, “Artificial neural network for ECG classification”, Recent Research in Science and
Technology 2014, pp:36-38.
[73]. MalayMitra, R.K.Samanta, “Cardiac Arrhythmia Classification Using Neural Networks with Selected Features”, Proceedia
Technology, Volume 10, 2013, pp: 76-84.
[74]. S. Osowski, T. Markiewicz, L.T. Hoai, “Recognition and classification system of arrhythmia using ensemble of neural
networks”, Measurement, Vol 41, issue 6, 2008, pp: 610–617.
[75]. Mayank Kumar Gautam, “A Neural Network approach and Wavelet analysis for ECG classification”, IEEE International
Conference on Engineering and Technology (ICETECH), 2016.
[76]. KritikaParganiha, Prasanna Kumar Singh, “ECG Interpretation Using Backward Propagation Neural Networks”, IJECET,Volume
5, Issue 4, April 2014, pp. 19-24.
[77]. Alonso-Atienza F, Morgado E, Fernández-Martínez L, García-Alberola A, Rojo-Álvarez JL, “Detection of life-threatening
arrhythmias using feature selection and support vector machines”, IEEE Trans Biomed Eng., Vol 61, 2014, pp: 832-840.
[78]. NitinAjiBhaskar, “Performance Analysis of Support Vector Machine and Neural Networks in Detection of Myocardial
Infarction”, Procedia computer science, Volume 46, 2015, Pages 20-30.
[79]. Mert, A., Kilic, N. & Akan, A, “Evaluation of bagging ensemble method with time -domain feature extraction for
diagnosing of arrhythmia beats”, Neural Computing and Applications, Vol 24, 2014, pp: 317–326.
[80]. Zhu, J. J., He, L. S. &Gao, Z. Q, “Feature extraction from a novel ECG model for arrhythmia diagnosis”, Biomed. Mater.
Eng., Vol 24, 2014, pp: 2883–2891.
[81]. Li, P. F. et al.,“High-performance personalized heartbeat classification model for long-term ECG signal”, IEEE Trans.
Biomed. Eng.,2017, pp: 78–86.
[82]. Li, H. Q. et al., “Arrhythmia classification based on multi-domain feature extraction for an ECG recognition
system”, Sensors 16, 2016, pp: 1–16.
[83]. Shen, C. P. et al., “Detection of cardiac arrhythmia in electrocardiograms using adaptive feature extr action and modified
support vector machines”, Expert Systems with Applications, Vol 39, 2012, pp: 7845–7852.
[84]. Martis, R. J. et al., “Characterization of ECG beats from cardiac arrhythmia using discrete cosine transform in PCA
framework”, Knowledge-based Systems, Vol 45, 2013, pp: 76–82.
[85]. Martis, R. J., Acharya, U. R. & Min, L. C, “ECG beat classification using PCA, LDA, ICA and discrete wavelet
transform”, Biomedical Signal Processing and Control, Vol 8, 2013, pp: 437–448.
[86]. Escalona-Morán MA, Soriano MC, Fischer I, Mirasso CR, “Electrocardiogram classification using reservoir computing with
logistic regression”, IEEE J Biomed Health Inform., 19(3), 2015pp: 892-898.
[87]. M. Lukoˇseviˇcius, H. Jaeger, “Reservoir computing approaches to recurrent neural network training”, Comput.Sci. Rev., 2009,
pp: 127–149.
[88]. Chaurasia V, Pal S, “Early Prediction of Heart Diseases Using Data Mining”, Caribbean Journal of Science and Technology,
Vol.1, 2013, pp:208-217.
[89]. ShameerFaziludeen, Praveen Sankaran, “ECG Beat Classification Using Evidential K -Nearest Neighbours”, procedia computer
science, Volume 89, 2016, pp: 499-505.
[90]. InduSaini, Dilbag Singh, and ArunKhosla, “QRS detection using K-Nearest Neighbor algorithm (KNN) and evaluation on
standard ECG databases”, J Adv Res., Vol 4, issue 4, 2013, pp: 331–344.
[91]. Y.-C. Yeh, C.W. Chiou, H.-J. Lin, “Analyzing ECG for cardiac arrhythmia using cluster analysis”, Expert Syst. Appl., 2012, pp:
1000–1010.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 11 | Page

 Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis: A Review
[92]. J.H.Abawajy, A.V.Kelarev, M.Chowdhury, “Multistage approach for clustering and classification of ECG data”, computers
methods and programming in biomedicine, Volume 112, Issue 3, December 2013, pp: 720-730.
[93]. M. Lagerholm, C. Peterson, G. Braccini, L. Edenbrandt, L.Sornmo, “Clustering ECG complexes using hermite functions and
self-organizing maps”, IEEE Trans. Biomed.Eng., 47 (7), 2000, pp: 838–848.
[94]. J. Rodrguez-Sotelo, D. Peluffo-Ordoez, D. Cuesta-Frau, G. Castellanos-Domnguez, “Unsupervised feature relevance analysis
applied to improve ECG heartbeat clustering”, Comput. Methods Programs Biomed., 108, 2012, pp: 250–261.
[95]. Leigang Zhang, “An approach for ECG classification based on wavelet feature extraction and decision tree”, International
Conference on Wireless Communications and Signal Processing (WCSP), 2010.
[96]. BaljitKaur, “ECG analysis with signal classification using Decision Tree Induction (DTI)”, Proceedings of the International
Conference on Advances in Information Communication Technology & computing, 2016.
[97]. Smita L. Kasar, “Analysis of Multi-Lead ECG Signals using Decision Tree Algorithms”, International Journal of Computer
Applications, Volume 134, No.16, January 2016, pp: 0975 – 8887.
[98]. AhmetMert, “ECG Signal Classification Using Ensemble Decision Tree”, 16th International Research/Expert Conference
Trends in the Development of Machinery and Associated Technology, TMT 2012, Dubai, UAE, September 2012, pp: 10-12.
[99]. R.V. Andreao, B. Dorizzi, J. Boudy, “ECG signal analysis through hidden markov models”, IEEE Trans. Biomed. Eng., Vol 53,
issue 8, 2006, pp: 1541–1549.
[100]. Wei Liang, Yinlong Zhang, Jindong Tan and Yang Li, “A Novel Approach to ECG Classification Based upon Two-Layered
HMMs in Body Sensor Networks”, Sensors2014, pp:5994-6011.
[101]. K.Vimala, “Stress causing Arrhythmia Detection from ECG Signal using HMM”, International Journal of Innovative Research
in Computer and Communication Engineering, Vol. 2, Issue 10, October 2014.
[102]. Greenwald SD, “Development and analysis of a ventricular fibrillation detector”, M.S.Thesis, MIT Dept. of Electrical
Engineering and Computer Science, 1986.
[103]. Taddei A, Distante G, Emdin M, Pisani P, Moody GB, “The European ST-T Database: Standard for evaluating systems for the
analysis of ST-T changes in ambulatory Electrocardiography”, European Heart Journal, 1992.
[104]. https://www.ecri.org/components/Pages/AHA_ECG_USB.aspx?tab=3
[105]. Nolle FM, Badura FK, Catlett JM, CREI-GARD, “a new concept in computerized arrhythmia monitoring systems, computers in
cardiology”, 1986, 13, pp: 515-518.
[106]. Moody GB, Muldrow WE, Mark RG, “A noise stress for arrhythmia detectors”, computers in cardiology, 1984.pp: 381-384.

IOSR Journal of VLSI and Signal Processing (IOSR-JVSP) is UGC approved Journal with Sl.
No. 5081, Journal no. 49363.

Gopisetty Ramesh Automatic Detection of Cardiac Arrhythmia through ECG Signal Analysis:
A Review.” IOSR Journal of VLSI and Signal Processing (IOSR-JVSP) , vol. 7, no. 5, 2017,
pp. 01-12.

DOI: 10.9790/4200-0706010112 www.iosrjournals.org 12 | Page