“Classification of ECG-signals using Artificial Neural

Networks” Gaurav D.Upadhyay

1 Akshay S. Thaware

2 Sumit M. Pali

3 Prateek A. Madne

4

Abstract – An electrocardiogram (ECG) is a bioelectrical

signal which records the heart's electrical activity with

respect to time. Early and accurate detection is important

in detecting heart diseases and choosing appropriate

treatment for a patient. ECG signals are used as the

parameter for detection of Cardiac diseases and most of

the data comes from PhysioDataNet and MIT-BIH

database .The pre-processing of ECG signal is performed

with help of Wavelet toolbox and also used for feature

extraction of ECG signal. The complete project is

implemented on MATLAB platform. The performance of

the algorithm is evaluated on MIT–BIH Database. This

paper presents the application of Probabilistic Neural

Networks (PNN) for the classification and detection of

Electrocardiogram (ECG).

Keywords: Electrocardiogram (ECG), MIT-BIH database,

Probabilistic Neural Networks (PNN), Wavelet toolbox.

I. INTRODUCTION

Electrocardiography deals with the electrical activity of the

heart. Bio-signals being non-stationary signals, the

reflection may occur at random in the time-scale.

Therefore, for effective diagnostic, ECG signal pattern and

heart rate variability may have to be observed over several

hours. Thus the volume of the data being enormous, the

study is tedious and time consuming. Therefore, computer-

based analysis and classification of cardiac diseases can be

very helpful in diagnostic. The ECG may roughly be

divided into the phases of depolarization and repolarisation

of the muscle fibers making up the heart. The

depolarization phases correspond to the P-wave (atrial

depolarization) and QRS-wave (ventricles depolarization).

The re-polarization phases correspond to the T-wave.

Arrhythmia is a heart disorder representing itself as an

irregular heartbeat due to malfunction in the electrical

system cells in the heart. It causes the heart to pump blood

less effectively and causing disorders in the heart

conduction process. Early detection of heart disease is very

helpful for living a long life and increase the improvement

of our technique detection of arrhythmias. The technique

used in ECG pattern recognition comprises: ECG signal

pre-processing, QRS detection, feature extraction and

neural network for signal classification. Probabilistic

Neural Network (PNN) is used as a classifier to detect QRS

and non-QRS regions. Most of the QRS detection

algorithms reported in literature detects R-peak locations

and separate rules are applied for the delineation of QRS

i.e. to locate the onsets and offsets of the QRS complexes.

Fig. 1. Normal ECG waveform

II. LITERATURE SURVEY

Nazmy et al [1] described adaptive neuro-fuzzy inference

system (ANFIS) algorithm for classification of ECG wave

.The feature extraction is done with the help of

Independent Component Analysis (ICA) and Power

spectrum and input is provided by the RR interval of ECG.

In this paper the classified ECG signals are normal sinus

rhythm (NSR), premature ventricular contraction (PVC),

atrial premature contraction (APC), Ventricular

Tachycardia (VT), Ventricular Fibrillation (VF) and

Supraventricular Tachycardia (SVT).using ANFIS

approach the classification accuracy is also obtained.

Alan and Nikola in [2] presented that use chaos theory for

classification of ECG signal and feature extraction. In this

paper also consist of including phase space and attractors,

correlation dimension, spatial filling index, central

tendency measure and approximate entropy. A new

program is developed for ECG classification which is

based on the chaos method and also developed semi-

automatic program for feature extraction. The program is

helpful to classify the ECG wave and extract the features of

the signal successfully.

Castro et al. in [3] describe the feature extraction with the

help of wavelet transform technique and also present an

algorithm which will utilize the wavelet transform for

extracting the feature of ECG wave. Their proposed

method first denoised by use of soft or hard threshold then

the feature of ECG wave divided in to coefficient vector by

optimal wavelet transformation. In the proposed method

choose the mother wavelet transform set of orthogonal and

biorthogonal wavelet filter bank by means of the best

correlation with the ECG signal was developed. After the

analysis of ECG signal coefficient are divided QRS

complex, T wave, P wave then sum to obtain feature

extraction.

Wisnu Jatmiko, et al. employed Back-Propagation Neural

Networks and Fuzzy Neuro Learning Vector Quantization

(FLVQ) as classifier in ECG classification [3]. In their

work, they used only the MLII lead as source data. The

classes that are considered are Left Bundle Branch Block

beat (LBBB), Normal beat (NORMAL), Right Bundle

Branch Block beat (RBBB), Premature Ventricular

Contraction (PVC). They used training classification

methods namely Back-propagation and FLVQ for their

experiment. It produces an average accuracy 99.20% using

Back- Propagation and 95.50% for FLVQ. The result

shows that back-propagation leading than FLVQ but, back-

propagation has disadvantages to classified unknown

category beat but not for FLVQ. FLVQ has stable accuracy

although contain unknown category beat.

Maedeh Kiani Sarkaleh, [4], proposed a Neural Network

(NN) based algorithm for classification of Paced Beat (PB),

Atrial Premature Beat (APB) arrhythmias as well as the

normal beat signal. They applied Discrete Wavelet

Transform (DWT) for feature extraction and used it along

with timing interval features to train the Neural Network.

About 10 recordings of the MIT-BIH arrhythmias database

have been used for training and testing the neural network

based classifier. The model results show that the

classification accuracy is 96.5%.

Karpagachelvi.S, [5], a novel ECG beat classification

system using RVM is proposed and applied to MIT/BIH

arrhythmia database to classify five kinds of abnormal

waveforms and normal beats. In exacting, the sensitivity of

the RVM classifier is tested and that is compared with

ELM. The obtained results clearly confirm the superiority

of the RVM approach when compared to traditional

classifiers.

Ruchita Gautam and Anil Kumar Sharma [6] proposed a

method is based on the Dyadic wavelet transform (DyWT)

technique this method is applied for finding the QRS

complex. In these method focused on the interval of the

two consecutive R wave and calculate the heartbeat. This

method is applied on the ECG waveforms for detect the

dieses Ventricular Late Potentials (VLP’s), and separate the wave P R & T which is associated with features of

ECG waveforms, In theses method the main consideration

is to find out the R waves and threshold is set to 75% of the

maximum peak.

Manpreet Kaur, A.S.Arora [7] shows with the help of K

clustering technique the output signal is analyzed, the

parameter is wave shape, duration and amplitude. With the

help of K clustering technique minimize the sum of point

to centroid distance, this clustered K summed. In these

technique first phase give information about the points are

resigned to the closest cluster around the centroid. The

second phase gives information on line value where values

are self-resigned. The data comes from MIT-BIH for

analysis. The success rate of classification for set 2, set 3,

set 4, set 5 and set 7 is 100%, for set 1 it is 87.5% and for

set 6 it is 75%.

III. Probabilistic Neural Network

Artificial neural networks have been used to solve a wide

variety of tasks that are hard to solve using ordinary rule-

based programming. In this work, Probabilistic Neural

Network (PNN) was used for classification. A probabilistic

neural network (PNN) is a feed-forward neural network,

derived from the Bayesian network and a statistical

algorithm called Kernel Fisher Discriminant analysis. In a

PNN, the operations are organized into a multilayered feed-

forward network with four layers namely Input layer,

Pattern layer and Decision layer as shown in figure 13.

There is one neuron in the input layer for each predictor

variable value. The input neurons then supply the values to

each of the neurons in the pattern layer. Pattern layer has

one neuron for each case in the training data set. The

neuron stores the values of the predictor variables for the

case beside with the target value.

IV. Wavelet Transform

The wavelet transform is a convolution of the wavelet

function ψ (t) with the signal x (t). Orthonormal dyadic discrete Wavelets are associated with scaling functions

ϕ(t). Wavelet transform: For extracting parameters of ECG

we use wavelet transform, wavelet analysis breaks a signal

down into its constituent parts for analysis. The scaling

function can be convolved with the signal to produce

approximation coefficients. The discrete wavelet

transform (DWT) can be written as:

Tm,n =∫ x(t)*ψ m,n (t)dt

A. Performance Measure

We have used three parameters for evaluating performance

of our algorithm. Those are accuracy, sensitivity.These

parameters are defined using 4 measures True Positive

(TP), True Negative (TN), False Positive (FP), and False

Negative (FN).

True Positive: arrhythmia detection coincides with

decision of physician

True Negative: both classifier and physician suggested

absence of arrhythmia

False Positive: system labels a healthy case as an

arrhythmia one

False Negative: system labels an arrhythmia as healthy

Accuracy: Accuracy is the ratio of number of correctly

classified cases, and is given by,

Accuracy= (TP+TN) / N

Total number of cases are N

Sensitivity: Sensitivity refers to the rate of correctly

classified positive. Sensitivity may be referred as a True

Positive Rate. Sensitivity should be high for a classifier.

Sensitivity = TP / (TP+FN).

V. METHODOLOGY

Denoising and detection of the QRS complexes in an ECG

signal provides information about various cardiac

abnormalities. It supplies evidence for the diagnosis of

cardiac diseases. For this very important reason, it has

earned a great respect in medical community.

Unfortunately, the presence of noise and time-varying

morphology makes the detection difficult.

Fig. 3 Block diagram of ECG classification

Preprocessing ECG signals helps us remove contaminants

from the ECG signals. ECG contaminants can be classified

into the following categories: Power line interference,

contact noise, Patient–electrode motion artifacts,

Electromyography (EMG) noise, Baseline wandering.

Digital filtering methods as well as wavelet based methods

are used to remove baseline wandering and the other

wideband noise. The baseline wandering and the above

noises are removed by taking two approximation level

coefficients.

Detection of R peaks is very important because they define

the cardiac beats. Heart rate is the important parameter that

is detected for analyzing the abnormality in the heart. Heart

rate is calculated based on R-R interval. The detection of

the QRS-complex is the most important task in automatic

ECG signal analysis. Q and S points are detected after

detecting the R peak by the slope inversion method. Wave

shape and the signal are classified into various arrhythmia

cases.

VI. CONCLUSION

This study is on detection and classification of arrhythmia

beats. The heart beats are different for different person

and all these beats are having different variations with

nonlinear nature. Thus the proposed computerized system

will be helpful for early detection of heart status and to

decrease the death percentage of human which occurs

due to the heart disease.

REFERENCE

T. M. Nazmy, H. El-Messiry and B. Al-bokhity. 2009. Adaptive Neuro-

Fuzzy Inference System for Classification of ECG Signals, Journal of

Theoretical and Applied Information Technology.

Alan Jovic, and Nikola Bogunovic, 2007.Feature Extraction for ECG

Time-Series Mining based on Chaos Theory, Proceedings of 29th

International Conference on Information Technology Interfaces.

B. Castro, D. Kogan, and A. B. Geva, 2000. ECG feature extraction using

optimal mother wavelet, The 21st IEEE Convention of the Electrical and

Electronic Engineers in Israel, pp. 346-350.

Wisnu Jatmiko, Nulad W. P., Elly Matul I.,I Made Agus Setiawan, P.

Mursanto,” Heart Beat Classification Using Wavelet Feature Based on

Neural Network ,” Wseas Transactions on Systems, ISSN: 1109-2777

Issue 1, Volume 10, January 2011.

Maedeh Kiani Sarkaleh and Asadollah Shahbahrami, “Classification of ECG Arrhythmias using Discrete Wavelet Transform and Neural

Networks”, International Journal of Computer Science, Engineering and Applications (IJCSEA) Volume 2, Issue 1, February 2012.

Karpagachelvi.S, Dr.M.Arthanari and Sivakumar M, “Classification of Electrocardiogram Signals with Extreme Learning Machine and

Relevance Vector Machine”, International Journal of Computer Science Issues, Volume 8, Issue 1, January 2011 ISSN (Online): 1694-0814.

V. Vijaya, K. Kishan Rao, V. Rama, “Arrhythmia Detection through ECG Feature Extraction using Wavelet Analysis”, European Journal of Scientific Research, Vol. 66, pp. 441-448, 2011.