INTERNATIONAL JOURNAL OF ELECTRONICS AND COMMUNICATION ENGINEERING & TECHNOLOGY (IJECET)

Similar documents
ECG classification and abnormality detection using cascade forward neural network

Keywords: Adaptive Neuro-Fuzzy Interface System (ANFIS), Electrocardiogram (ECG), Fuzzy logic, MIT-BHI database.

Keywords : Neural Pattern Recognition Tool (nprtool), Electrocardiogram (ECG), MIT-BIH database,. Atrial Fibrillation, Malignant Ventricular

Testing the Accuracy of ECG Captured by Cronovo through Comparison of ECG Recording to a Standard 12-Lead ECG Recording Device

Neural Network based Heart Arrhythmia Detection and Classification from ECG Signal

ECG Signal Characterization and Correlation To Heart Abnormalities

IDENTIFICATION OF NORMAL AND ABNORMAL ECG USING NEURAL NETWORK

An ECG Beat Classification Using Adaptive Neuro- Fuzzy Inference System

Robust system for patient specific classification of ECG signal using PCA and Neural Network

Analysis of Fetal Stress Developed from Mother Stress and Classification of ECG Signals

ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 2, Issue 10, April 2013

ECG Signal Analysis for Abnormality Detection in the Heart beat

Assessment of Reliability of Hamilton-Tompkins Algorithm to ECG Parameter Detection

II. NORMAL ECG WAVEFORM

A Review on Arrhythmia Detection Using ECG Signal

ECG SIGNAL PROCESSING USING BPNN & GLOBAL THRESHOLDING METHOD

ECG Beat Recognition using Principal Components Analysis and Artificial Neural Network

CHAPTER 5 WAVELET BASED DETECTION OF VENTRICULAR ARRHYTHMIAS WITH NEURAL NETWORK CLASSIFIER

A MULTI-STAGE NEURAL NETWORK CLASSIFIER FOR ECG EVENTS

Artificial Neural Networks in Cardiology - ECG Wave Analysis and Diagnosis Using Backpropagation Neural Networks

CARDIAC ARRYTHMIA CLASSIFICATION BY NEURONAL NETWORKS (MLP)

Automatic Detection of Heart Disease Using Discreet Wavelet Transform and Artificial Neural Network

Comparison of ANN and Fuzzy logic based Bradycardia and Tachycardia Arrhythmia detection using ECG signal

DISEASE CLASSIFICATION USING ECG SIGNAL BASED ON PCA FEATURE ALONG WITH GA & ANN CLASSIFIER

Various Methods To Detect Respiration Rate From ECG Using LabVIEW

Comparison of Feature Extraction Techniques: A Case Study on Myocardial Ischemic Beat Detection

CHAPTER 4 CLASSIFICATION OF HEART MURMURS USING WAVELETS AND NEURAL NETWORKS

IJRIM Volume 1, Issue 2 (June, 2011) (ISSN ) ECG FEATURE EXTRACTION FOR CLASSIFICATION OF ARRHYTHMIA. Abstract

DETECTION OF HEART ABNORMALITIES USING LABVIEW

ECG Rhythm Analysis by Using Neuro-Genetic Algorithms

Performance Identification of Different Heart Diseases Based On Neural Network Classification

Classification of heart signal using wavelet haar and backpropagation neural network

Automated Diagnosis of Cardiac Health

Detection of Atrial Fibrillation by Correlation Method

Developing Electrocardiogram Mathematical Model for Cardiovascular Pathological Conditions and Cardiac Arrhythmia

Vital Responder: Real-time Health Monitoring of First- Responders

Quick detection of QRS complexes and R-waves using a wavelet transform and K-means clustering

Delineation of QRS-complex, P and T-wave in 12-lead ECG

Genetic Algorithm based Feature Extraction for ECG Signal Classification using Neural Network

CLASSIFICATION OF CARDIAC SIGNALS USING TIME DOMAIN METHODS

Cardiac Arrest Prediction to Prevent Code Blue Situation

The Use of Artificial Neural Network to Detect the Premature Ventricular Contraction (PVC) Beats

USING CORRELATION COEFFICIENT IN ECG WAVEFORM FOR ARRHYTHMIA DETECTION

COMPRESSED ECG BIOMETRIC USING CARDIOID GRAPH BASED FEATURE EXTRACTION

CHAPTER IV PREPROCESSING & FEATURE EXTRACTION IN ECG SIGNALS

Ventricular Parasystole

LABVIEW based expert system for Detection of heart abnormalities

Wavelet Decomposition for Detection and Classification of Critical ECG Arrhythmias

ABCs of ECGs. Shelby L. Durler

Premature Ventricular Contraction Arrhythmia Detection Using Wavelet Coefficients

PCA and SVD based Feature Reduction for Cardiac Arrhythmia Classification

Real-time Heart Monitoring and ECG Signal Processing

Wavelet Neural Network for Classification of Bundle Branch Blocks

REVIEW ON ARRHYTHMIA DETECTION USING SIGNAL PROCESSING

Multi Resolution Analysis of ECG for Arrhythmia Using Soft- Computing Techniques

Temporal Analysis and Remote Monitoring of ECG Signal

DIFFERENCE-BASED PARAMETER SET FOR LOCAL HEARTBEAT CLASSIFICATION: RANKING OF THE PARAMETERS

Body Surface and Intracardiac Mapping of SAI QRST Integral

Fuzzy Based Early Detection of Myocardial Ischemia Using Wavelets

Prediction of Diabetes by using Artificial Neural Network

COMPARING THE IMPACT OF ACCURATE INPUTS ON NEURAL NETWORKS

Classıfıcatıon of Dıabetes Dısease Usıng Backpropagatıon and Radıal Basıs Functıon Network

Detection and Classification of QRS and ST segment using WNN

Automatic Detection of Abnormalities in ECG Signals : A MATLAB Study

International Journal of Advance Engineering and Research Development

PCA Enhanced Kalman Filter for ECG Denoising

Assessment of the Performance of the Adaptive Thresholding Algorithm for QRS Detection with the Use of AHA Database

Comparison of Different ECG Signals on MATLAB

SPECTRAL ANALYSIS OF LIFE-THREATENING CARDIAC ARRHYTHMIAS

Heart Rate Calculation by Detection of R Peak

Extraction of Unwanted Noise in Electrocardiogram (ECG) Signals Using Discrete Wavelet Transformation

Coimbatore , India. 2 Professor, Department of Information Technology, PSG College of Technology, Coimbatore , India.

Simulation Based R-peak and QRS complex detection in ECG Signal

Electrocardiography Biomedical Engineering Kaj-Åge Henneberg

Classification of Cardiac Arrhythmias based on Dual Tree Complex Wavelet Transform

Speed - Accuracy - Exploration. Pathfinder SL

POWER EFFICIENT PROCESSOR FOR PREDICTING VENTRICULAR ARRHYTHMIA BASED ON ECG

Identification of Arrhythmia Classes Using Machine-Learning Techniques

ECG DE-NOISING TECHNIQUES FOR DETECTION OF ARRHYTHMIA

Robust Detection of Atrial Fibrillation for a Long Term Telemonitoring System

ARTIFICIAL NEURAL NETWORKS TO DETECT RISK OF TYPE 2 DIABETES

Application of Artificial Neural Networks in Classification of Autism Diagnosis Based on Gene Expression Signatures

SPPS: STACHOSTIC PREDICTION PATTERN CLASSIFICATION SET BASED MINING TECHNIQUES FOR ECG SIGNAL ANALYSIS

AFC-ECG: An Intelligent Fuzzy ECG Classifier

I PART I. Timing Cycles and Troubleshooting Review COPYRIGHTED MATERIAL

Classification of electrocardiographic ST-T segments human expert vs artificial neural network

Signal Processing of Stress Test ECG Using MATLAB

Discrete Wavelet Transform-based Baseline Wandering Removal for High Resolution Electrocardiogram

AUTOMATIC ANALYSIS AND VISUALIZATION OF MULTILEAD LONG-TERM ECG RECORDINGS

A RECOGNITION OF ECG ARRHYTHMIAS USING ARTIFICIAL NEURAL NETWORKS

ATRIAL FIBRILLATION FROM AN ENGINEERING PERSPECTIVE

DIABETIC RISK PREDICTION FOR WOMEN USING BOOTSTRAP AGGREGATION ON BACK-PROPAGATION NEURAL NETWORKS

Electrocardiography for Healthcare Professionals

CONVERSION OF ECG GRAPH INTO DIGITAL FORMAT AND DETECTING THE DISEASE. G. Angelo Virgin 1, Dr.M.Sangeetha 2

Clinical and Electrocardiographic Characteristics of Patients with Brugada Syndrome: Report of Five Cases of Documented Ventricular Fibrillation

Detection of Atrial Fibrillation Using Model-based ECG Analysis

ECG based Atrial Fibrillation Detection using Cuckoo Search Algorithm

Research Article Artificial Neural Network-Based Automated ECG Signal Classifier

MRI Image Processing Operations for Brain Tumor Detection

VLSI Implementation of the DWT based Arrhythmia Detection Architecture using Co- Simulation

Transcription:

INTERNATIONAL JOURNAL OF ELECTRONICS AND COMMUNICATION ENGINEERING & TECHNOLOGY (IJECET) International Journal of Electronics and Communication Engineering & Technology (IJECET), ISSN 0976 ISSN 0976 6464(Print) ISSN 0976 6472(Online) Volume 5, Issue 4, April (2014), pp. 19-24 IAEME: www.iaeme.com/ijecet.asp Journal Impact Factor (2014): 7.2836 (Calculated by GISI) www.jifactor.com IJECET I A E M E ECG INTERPRETATION USING BACKWARD PROPAGATION NEURAL NETWORKS Kritika Parganiha #1, Prasanna Kumar Singh #2 #1 M.Tech student, Dept. of ECE, Lingayas University, Faridabad, India #2 Associate Professor, E&C Dept., Lingayas University, Faridabad, India ABSTRACT Electro Cardiogram (ECG) is a non invasive technique and is used as a primary diagnostic tool for cardiovascular diseases. ECG Signal provides necessary information about electrophysiology of heart diseases and ischemic changes that may occur. Arrhythmias are amongst the most common ECG abnormalities. There are various Arrhythmias like Ventricular Premature Beats, Asystole, Couplet, Fusion beats etc. This information can be extracted from ECG but its diagnosis simply depends on the experience of the physician. Previously many methods have been used for the analysis and automisation of analysis. In this paper we make use of MATLAB based Artificial Neural Networks (ANN) to judge whether the patient is normal or not. We use Back Propagation Neural Networks Levenberg Marquardt (LM) Algorithm by taking the ECG input in the form of digital time series signal. These results are compared with previous neural network techniques and found that the method proposed in this paper gives best results. Keywords: Arrhythmia, MATLAB, Artificial Neural Networks, Radon Transform, Back Propagation, Levenberg Marquardt Algorithm. 1. INTRODUCTION Heart related problems are the major concern now-a-days. Monitoring heart helps to determine the abnormalities of a cardiac patient. Electrocardiography (ECG) is used for this purpose in the hospitals since a long time. ECG plays an important role in healthcare and with time its volume has increased by a large amount. ECG is printed in a thermal paper and is kept in hospitals for further diagnosis and as records. This requires immense storage space and manpower requirement which is time consuming and not economical. This can be eliminated by converting. 19

ECG records into digital time series signal[1]. Now these ECG samples are analysed by experienced doctors who depending upon their knowledge predict the problems associated with patient. This experience based analysis gives different interpretations. Hence there is a need of system that could analyse ECG signals properly and accurately so that there is a less chance of mistake and the problem gets spotted in its early stage. For this purpose many works in the field of Image processing, Digital signal processing etc has been done making use of Artificial Neural Networks [11] which has given effective results to such complex problems. This paper has been divided into five sections. Section 1 gives basic Introduction. Section 2 depicts the Methodology used. Section 3 gives information about the Database. Section 4 represents the type of Input taken. Section 5 shows the Results obtained and in Section 6 we discuss the Conclusion. 2. METHODOLOGY The database provided by MIT-BIH Arrhythmia database [3] regarding different kinds of heart rhythm abnormalities for different class of patients is used for training, testing and validation of neural networks. In this paper, 12 lead ECG signals were recorded at 25mm/sec and printed in thermal paper. These ECG trace are scanned at 600dpi (dots per inch) black and white images and stored in jpeg format. Then using Radon Transform the skewness of images is detected and corrected, which may have incurred during scanning. The de-skewed or corrected image is then adaptively binarized by choosing local thresholds and then it is filtered by morphological filters. In this work Otsu s algorithm [9] has been performed for image adaptive binarization. Finally the peaks of ECG signal input are detected and these peaks are used as an input for ECG Classification Using Neural Networks. Then the parameters namely, Standard Deviation, Correlation and Wavelet Coefficient are extracted which helps to decide the performance of the algorithm. The features are then encrypted and testing process is done to determine whether the signal is normal or abnormal. 3. DATABASE The database used in this paper to train and test the neural network, is the standard MIT-BIH arrhythmia database [3].The input database consists of 48 half-hour excerpts of two channel ambulatory ECG recordings, obtained from subjects studied by BIH Arrhythmia Laboratory. The recordings were digitized and then taken as input in the network. 4. INPUT The input for network was selected keeping in view following criteria[2]: a. The input must be of standard size so that it is neither too small nor too large. b. The input must be arranged in such a way that the R-peak in QRS complex is placed at centre of signal cycle. 20

6464(Print), ISSN 0976 6472(Online), Volume 5, Issue 4, April (2014), pp. 19-24 IAEME Figure 1: Schematic representation of normal ECG waveform Here in figure 1, A clear P wave before the QRS complex represents sinus rhythm. Absence of P waves may suggest atrial fibrillation, junction rhythm or ventricular rhythm. The QRS complex is the largest voltage deflection of approximately 10 20 mv but may vary in size depending on age, and gender. The voltage amplitude of QRS complex may also give information about the cardiac disease. T wave represents ventricular repolarization. 5. RESULTS The following table represents the results obtained using Levenberg Propagation Neural Networks. Marquardt Back Table 1: Back Propagation Neural Network Design Analysis Results TRAINLM Training Algorithm HN 10 Time ( in Sec) 0:00:02 Epochs ( Max: 30) 6 MSE (0.0001) 0.00753 Gradient 0.110 Result Passed HN- Hidden neurons, representing the number of neurons in the Hidden layer. Time- Maximum Training Time. MSE- Mean Squared Error the errorr goal being fixed at 0.0001 and hence here the difference MSE- 0.0001 is being tabulated. TRAINLM-Levenberg Marquardt Back propagation training algorithm. 21

The following images are analysis plots for this work, each representing different properties about the network. Figure 2: Training Process Results In the above figure 2, less number of epochs means that network is a good learner and it learns in small repetitions. Less time means network achieving goal easily and in short span of time. Performance indicates final Mean Square Error (MSE) achieved in which lower value corresponds to higher network accuracy. Figure 3: Mean Squared Error ( MSE) Plot In Figure 3, the Mean Squared Error (MSE) Plot shows achieved error value. Lower value depicts less probability of false predictions. Hence, our network achieved quite low error probability. 22

Figure 4: Gradient & Validation Check Plots In Figure 4, Lower value of Gradient plot shows that network is learning upto large extent representing fine adjustments in weights and biases making the network more accurate and reliable avoiding any false predictions. Validation plot represents the point where network learned sufficiently and has passed the validation. The magnitude of the gradient and the number of validation checks are used to terminate the training. The gradient will become very small as the training reaches a minimum of the performance. The number of validation checks represents the number of successive iterations that the validation performance fails to decrease. 6. CONCLUSION This network gives quite low value of MSE and is near 0.00753 in just 6 epochs. Levenberg Marquardt Algorithm proves to be fastest method for training moderate sized neural networks. The network based on Back Propagation Neural network algorithm with trainlm training algorithm was best for case of normal beat analysis giving an accuracy of about 99.9% as well as low memory requirement. Hence this method was preferred for normal beat analysis. By this work we conclude that by using MATLAB based Neural network design [6]; such networks can be made with capability to understand different class of inputs which are fed to be analyzed. Though the objective of this research was not to use MATLAB or Neural Networks and these were used to get higher accuracy for analysis of ECG which is more useful for the mankind. The results obtained with other methods are compared with our results [4][5][7][8][10]. Table 2 shows the comparison of results. 23

Table 2: Comparison of Results METHOD % OF CORRECT CLASSIFICATION Multilayer Perceptron 98.87 (MLP) Hybrid Neuro- Fuzzy System 98.68 (HFNS) Principal Component Analysis 98.73 (PCA) Weightless 99.63 Our Method 99.9 These networks are not tested with the current real patients record but it will give the same high accuracy, the network being trained and tested with sufficient number of inputs. In this research paper, all arrhythmias are not classified by their name, they are just tested as normal or abnormal. The classification needs to be done as the next step of the research. REFERENCES [1] A. R. Gomes e Silva, H.M. de Oliveira, R.D. Lins., Converting ECG and other paper legated biomedical maps into digital signals, XXV Simposio Brasileiro de Telecomunicacoes, Setembro 3-6, Recife PE, Brasil. [2] Ayub and Saini / International Journal of Engineering, Science and Technology, Vol. 3, No. 3, 2011, pp. 41-4. [3] Brown G., 2006, MIT-BIH Arrhythmia database, MIT. [4] Chickh M. A.N. Belgacem, F. Bereksi-Reguig, 2002, Neural classifier to classify ectopic beats. Acte des IX emes rencontre de la Societe Francophone de Classification, Toulouse, le 16-18. [5] Chow H.S,. Moody G. B., and Mark R.G.1993, Detection of ventricular ectopic beats using neural networks, Computers in Cardiology, pp 659-662. [6] Demuth Horward, Beale Mark, Hagan Martin, 2008, MATLAB Neural Network Toolbox, MATHWORKS INC., MATLAB Version R2008b, October. [7] Gao D, Kinouchi Y, Ito K, Zhao X, 2003, Neural Networks for Event extraction from Time Series, A back propagation Algorithm Approach, Future generation, Computer System. [8] Nadal J. and Bosson M.C.,1993, Classification of cardiac arrhythmias based on principal component analysis and feedforward neural networks, Computers m Cardiology, pp 341-344. [9] Rafael C. Gonzalez and Richard E. Woods,, Digital Image Processing (3rd Edition), Prentice Hall, 2008. [10] Thomson D. C., Soraghan J. J, and Durrani T.S, 1993, An artificial neural-network based SVT/VT classification system, Computers in Cardiology,(pp 333-336. [11] Zurada, 1999, Introduction to Artificial Neural Systems, West Publishing House. [12] Shivajirao M. Jadhav, Sanjay L. Nalbalwar and Ashok A. Ghatol, Performance Evaluation of Multilayer Perceptron Neural Network Based Cardiac Arrhythmia Classifier, International Journal of Computer Engineering & Technology (IJCET), Volume 3, Issue 2, 2012, pp. 1-11, ISSN Print: 0976 6367, ISSN Online: 0976 6375. 24