Feature Extraction and analysis of ECG signals for detection of heart arrhythmias

Transcription

1 Volume No - 5, Issue No 3, May, 2017 Feature Extraction and analysis of ECG signals for detection of heart arrhythmias Sreedevi Gandham Dept of Electronics and communication Eng Sri Venkateswara University Tirupati, AP,India Anuradha Bhuma Deptof Electronics and communication Eng Sri Venkateswara University Tirupati, AP, India Abstract: ECG is an important and routine tool used in the primary diagnosis, prognosis and survival analysis of cardiac ailments In the present work, Discrete Wavelet Transform (DWT) was used for feature extraction and analysis of ECG signals A total of nineteen long length ECG signals were taken from MIT-BIH database and tested to check the performance of proposed method Arrhythmias detected were Bradycardia, Tachycardia, Ventricular Tachycardia, First degree heart block, Asystole and Second degree heart block The MATLAB tool used, in addition to detecting arrhythmias, helps in analysing ECG through different parameters of ECG like sampling frequency, PR, RR interval and QRS width Keywords: Electrocardiogram (ECG), wavelet transform, Feature Extraction, classification I INTRODUCTION An Electrocardiogram (ECG) is basically a register of the heart s electrical activity, widely used as a routine cardiac diagnostic tool The frequency range of ECG is from Hz A typical ECG tracing of normal heart beat consists of a P wave, a QRS complex, a T wave and a U wave as shown in fig-1 Successive repetition of such PQRST waves in monotony forms ECG The characteristics of normal heart rhythm are called Normal Sinus Rhythm (NSR), any disorder in these parameters results in a pathological condition called Arrhythmia or dysrrhythmia Three common types of arrhythmias observed are supraventricular arrhythmia, ventricular arrhythmia and Bradyarrhythmias Supraventricular arrhythmias are tachycardia (fast heart rates) type which includes atrial fibrillation, atrial flutter, paroxysmal supraventricular tachycardia (PSVT) and Wolff-Parkinsonwhite (WPW) syndrome [1] Ventricular arrhythmias majorly include ventricular tachycardia and ventricular fibrillation Bradyarrhythmia occurs because of slower heart rate ECG signal being non-stationary it is difficult to visually analyse and may take a lot of time, hence we need computer based methods for its analysis ECG feature extraction plays vital role in disease identification, many methods are developed for this in both time domain and frequency domain Time domain approaches are not always adequate to study all the features of ECG signal [2], therefore, frequency representation of signal is needed The features are: mean, median, maxi-mum, minimum, range, standard deviation, variance, and mean absolute deviation In this work we developed simple and accurate algorithm using Discrete Wavelet Transform (DWT)Wavelets can provide a time versus frequency representation of the signal and work well on non-stationary data Figure 1: Normal ECG Signal Most of ECG signals were obtained from the MIT-BIH arrhythmia database [3], which are sampled at the rate of 360 samples/sec We also acquired additional ECG signals from patients recorded to have cardiac ailments by using a bioamplifier together with data acquisition card (Advantech DAQ) Presence of noises posed considerable threat while recording ECG signal These noises include base line wandering noise, power line (50Hz) noise, muscle tremor etc Specially designed filters were used in the bio amplifier module, which include band pass filter, notch filter etc to filter out these potential noises In addition to this, software based filters were also employed to effectively filter the acquired signal for better efficiency Sampling rate of the acquired signal are 250 Hz II WAVELET TRANSFORMS A wavelet is simply a small wave which has energy concentrated in time to give a tool for the analysis of transient, non stationary or time-varying phenomena We used wavelet transform for extracting parameters of ECG Wavelet analysis breaks a signal down into its constituent parts for analysis Mother wavelet transform used here is Daubechies of Db family; if a signal is not adequately represented by one member of the Db family, it may be still efficiently represented by another Daubechies wavelet family has similar RES Publication 2012 Page 40 wwwijmeceorg

2 Volume No - 5, Issue No 3, May, 2017 shape to QRS complex and their energy spectrum is concentrated around low frequencies The wavelet transform is a convolution of the wavelet function ψ (t) with the signal x (t) Orthonormal dyadic discrete wavelets are generally associated with scaling functions ϕ (t) The scaling function could be convolved with the signal to produce approximation coefficients [4] The discrete wavelet transform (DWT) can be written as: Tm,n = x(t)ψ m,n (t) dt (1) By choosing an orthonormal wavelet basis, ψm,n (t), and we can reconstruct the original The approximation coefficient of the signal at the scale m and location n can be presented by S m,n = x(t)ϕ m,n (t)dt (2) In practice our discrete input signal S0, n is of finite length N, which is an integer power of 2: N = 2M Thus the range of scales that can be investigated is 0 < m < M A discrete approximation of the signal can be shown as X0(t) =XM (t) + Σ dm (t) (3) Where the mean signal approximation at scale M is XM (t) = Σ S M, n ϕ m, n (t) (4) and the detail signal approximation corresponding to scale m is defined for a finite length signal as dm (t) =Σ Tm, n(t)ψm,n(t) (5) Adding the approximation of the signal at scale index M to the sum of all detailed signal components across scales gives the approximation of the original signal at scale index 0 The signal approximation at a specific scale was a combination of the approximation and detail at the next lower scale Xm (t) = Xm-1(t) -dm (t) (6) If scale m = 3 was chosen, it can be shown that the signal approximation is given by X3(t) = X0 (t) -d1 (t) -d2 (t) -d 3(t) (7) This is referred to as multi resolution analysis of a signal using wavelet transform, and is used in our procedure 0<m<M A discrete approximation of the signal can be shown as Input ECG signal is decomposed in to approximate and detailed coefficients Figure (4), (5) gives all detailed and needed approximate coefficients of input ECG signal Of all those only d3, d4, d5, a6 coefficients are used to reconstruct ECG signal Now reconstructed signal is used for further processing ECG signal was de-noised by removing wavelet coefficients at higher scales Contributions and Results: 1 ECG data base 2 Pre-processing of ECG signal 3 QRS detection, R peak detection 4 Feature extraction of complete ECG 5 Arrhythmia classifications 6 Disease diagnosis Figure 2: Our proposed Algorithm III ECG PREPROCESSING The ECG waveform was first normalized to remove the DC offset After that the ECG signal was passed through squaring and smoothing filters [5] Stage 1 in figure 3 show various steps during ECG pre processing Filtered signal y[n] is obtained using the formula: where x(n) are the ECG samples 1QRS Detection Algorithm Stage 2 in the figure 3 depicts various steps in QRS detection algorithm The signal y[n] from (1) was then differentiated twice to give z[n] using the following equations: (9) (10) The differentiated signal was then squared to give w[n] using (11) Signal (4) denotes the squared QRS complex, from which the start point and end points of the QRS complex is determined In order to confirm the detected signal in a QRS complex, this signal has to pass through threshold and peak and valley checker [6] This was followed by the following threshold operation: (8) (12) This threshold operation produced a series of pulses h[n] having durations equivalent to the width of the QRS complexes The presences of the peak and valley points of the QRS complex were confirmed through Peak and valley checker, which is a distinguished feature of QRS complex RES Publication 2012 Page 41 wwwijmeceorg

3 IV International Journal of Modern Electronics and Communication Engineering (IJMECE) ISSN: Volume No - 5, Issue No 3, May, 2017 ECG FEATURE EXTRACTION: For ECG feature extraction, algorithm based on Daubechies Wavelet Transform is presented Db4 Wavelet is selected due to the similarity of its scaling function to the shape of the ECG signal R peaks detection is the core of this algorithm s feature extraction All other primary peaks are extracted with respect to the location of R peaks through creating windows proportional to their normal intervals The proposed extraction algorithm is evaluated on MIT-BIH Arrhythmia Database Experimental results indicate that the algorithm can successfully detect and extract all the primary features with a deviation error of less than 10% The study of electrocardiogram (ECG) signals has been a growing research topic due to its role in diagnosing many cardiac diseases [5,6] A normal ECG signal highlighting its main features in terms of its primary waves and time intervals The majority of the medically useful information in the ECG is originated from the intervals and amplitudes defined by its features (characteristic wave peaks and time durations) [6] There is a significant research effort paid to the investigation of methods for extracting useful features carrying medical information from ECG signals Mazomenos, Evangelos et al[7] presented a low complexity algorithm for extracting fiducial points from ECG signals using Discrete Wavelet Transform (DWT) Vaneghi et al [8] compared six distinct approaches for an EGC signal feature extraction Their findings revealed that the Eigenvector method outperformed the other 5 presented schemes One of the closest studies to ours is explored by Espiritu-Santo- Rincon et al [9] Similar to this work, they have devised a mechanism for detecting R peaks followed by the rest of the signal features However their method is totally based on Wave segmentation A number of other techniques have been proposed in for the detection of ECG features [10] Most of the aforementioned methods for ECG signal analysis were based on the time domain But it is not always adequate to study all the features of ECG signals in the time domain Therefore, the frequency representation of a signal is equally important and its processing is more appealing Current study is inspired by the ECG feature extraction algorithm proposed by Mahmoodabadi et al [11] Daubechies family of wavelet is used for the analysis and decomposition of the signal Current study s main objective is the extraction of primary features of the denoised and decomposed ECG signals By extracting these primary features of ECG, it is feasible to obtain some fundamental parameters like the amplitudes of the waves and their durations such as RR, QRS, and PR intervals which could be used for subsequent automatic analysis (1) PEAKS DETECTION For peak detection we used online implementation of state machine logic to determine peaks and location of various ECG signal parameters such as peaks and locations of P Q R S T wave Further, RR interval, ST segment, PR interval and TT interval etc were calculated using the basic logic to find distance between occurrence of the required parameter ECG analysis helps to detect and classify ECG waveform abnormalities This is achieved by extracting various features and durations of the ECG waveform such as RR interval, QRS complex, P wave and PR durations These durations are then compared with normal values to determine the degree and types of abnormalities In conventional methods of ECG diagnosis, physician tries to find whether the ECG signal is different from normal sinus rhythm in terms of the morphology of each component, time intervals and heart rate Considering the large numbers of critical patients and the need for accurate interpretation of their cardiac status and the limitations of human to observe the fine details in ECG waveform led to the introduction of automated arrhythmia monitoring systems to assist cardiologists in detecting the signs of arrhythmia at early stages which can save patients lives Further the processed ECG signal is given as input to the feature extraction algorithm where in the features related to time such as the occurrence and duration of P, Q, R, S and T waves are determined Table 1: Normal ECG Wave Amplitudes Wave Typical Amplitude RES Publication 2012 Page 42 wwwijmeceorg P wave R wave Q wave T wave 025 mv 160 mv 25% of R wave 01 to 05 mv (2) QRS COMPLEX DETECTION OF ECG SIGNAL In recent years, ECG signal plays an important role in the primary diagnosis, prognosis and survival analysis of heart diseases Electrocardiography has had a profound influence on the practice of medicine This work deals with the detection of QRS complexes of ECG signals using wavelet transform based algorithms[11] The electrocardiogram signal contains an important amount of information that can be exploited in different manners The ECG signal allows for the analysis of anatomic and physiologic aspects of the whole cardiac muscle In the wavelet based algorithm, the ECG signal has been denoised by removing the corresponding wavelet coefficients at higher scales Then QRS complexes are detected and each complex is used to find the peaks of the individual waves like P and T, and also their deviations The detection of the QRS complex particularly, the detection of the QRS complex peak, or R wave-in an ECG signal is a difficult problem because it has a time-varying morphology and is subject to physiological variations of the patient and extent of corruption due to noise Since the QRS complexes have a time-varying morphology, they are not always the strongest signal component in an ECG signal Therefore, P-waves or T-waves with characteristics similar to that of the QRS complex, as well as spikes from high frequency pacemakers can compromise the detection of the QRS complex In addition, there are many sources of noise

4 Volume No - 5, Issue No 3, May, 2017 in a clinical environment that can degrade the ECG signal These include power line interference, muscle contraction noise, poor electrode contact, patient movement, and baseline wandering due to respiration [15] Therefore, QRS detectors must be invariant to different noise sources and should be able to detect QRS complexes even when the morphology of the ECG signal is varying with respect to time Table 2: Normal ECG Wave Durations P-R interval 012 to 020 second Q-T interval 035 to 044 second S-T segment 005 to 015 second P wave interval 011 second QRS interval 009 second After the determination of the QRS durations the fiducial points which show the location of R points were determined The cardiac cycle (RR) intervals were obtained and hence the heart rate (HR) for each beat was calculated using the relationship: (13) The P-wave duration and the PR interval for each beat were also determined by scanning each beat waveform using a time window prior to each R point V CLASSIFICATION Based on features extracted (RR, PR intervals and QRS duration) decision rules were formed In our algorithm we took average of 8 RR, PR intervals and QRS width So intervals considered are averaged ones Total of 6 decision rules are formed after profoundly studying many arrhythmias Some of the arrhythmia classifying decision rules are For Bradycardia If (QRS==011&&PR>02 && abnormalbeats==0&&pr<02) (14) For Tachycardia If (QRS==011&&PR<02&&RR<085&& (abnormalbeats==0)) (15) Using this decision rules we can detect totally six Arrhythmias those are Bradycardia, Tachycardia, Ventricular Tachycardia, Asystole (Complete Heart Block), First Degree AV block, Second Degree AV block Sixteen arrhythmia decision rules were developed after consultation with a group of cardiology specialists The rules were applied to the features extracted from each ECG data files There were two types of analysis applied, one is the beat analysis and second is the beat to beat analysis In beat analysis, each beat was examined against standard values, where as in the beat to beat analysis the similarities in the beat morphology between adjacent beats were examined Some of the arrhythmia decision rules are explained in the equations (16) and (17) (16) (17) Based on the decision rules the arrhythmias were classified as Tachycardia (TA), Bradycardia (BR), Premature Ventricular contraction (PVC), Sick Sinus Syndrome (SSS), Premature Atrial Contraction (PAC), Wandering Pacemaker (WP), Complete Heart Block (CHB), First Degree Heart Block (FDHB), Bigeminy (BI) and Quadrigeminy (Q) VI RESULTS In this work original input signal (Actual Signal) taken from MIT-BIH data base is shown in figure 3, decomposed signal is shown in figure 4, preprossesing stage using DWT and reconstructed signal is shown in figure 5 Smoothing filters were used to remove the baseline wander (figure 6) Figure 7 determines peaks and location of various ECG signal parameters in P Q R S T wave Further, RR interval, ST segment, PR interval and TT interval etc were calculated using the basic logic to find distance between occurrence of the required parameter Figure 8 represents detection of r-peak in actual signal In figure 9 detection of P, QRS &T waves are depicted Figure 3:Original Signal Figure 4:Decomposed Signal using DWT RES Publication 2012 Page 43 wwwijmeceorg

5 Volume No - 5, Issue No 3, May, 2017 VII CONCLUSION By using DWT method, feature extraction and analysis of ECG signals was done The MATLAB tool helped to detect different arrhythmias and to increase accuracy of our algorithm Hither to, hardly any MATLAB based software could detect these many arrhythmias and at the same time analyse ECG signal in detail The obtained results indicate that the proposed algorithm can support various types of arrhythmias detection in clinical tests Figure 5:Reconstruted Signal using (DB4) REFERENCES [1] Sayantan Mukhopadhyay1, Shouvik Biswas, Wavelet Based QRS Complex Detection of ECG Signal, International Journal of Engineering Research and Applications,Vol2, Issue3, pp , May-Jun 2012 [2] Mujeeb Rahman, Mohamed Nasor, An Algorithm for Detection of Arrhythmia, International Journal of Biological Engineering, Volume 2 Issue 5,pp44-47, 2012 Figure 6:Removing Base line noice &Smoothed Signal Figure 7: Detection of R-Peak Figure 8:Detection of R-Peak in Actual Signal Figure 9: Detection of P,QRS,&T waves [3] S Z Mahmoodabadi, A Ahmadian, ECG Feature Extraction Using Daubechies fifth iasted international conference visualization, imaging and image processing, september Issue, pp7-9, 2005 [4] Introduction to graphical user interface (GUI), The Mathworks [5] Jaylaxmi C Mannurmath, Prof Raveendra M, MATLAB Based ECG Signal Classification, International Journal of Science, Engineering and Technology Research,Vol 3, Issue 7, July 2014 [6] MSabarimalai Manikandana, KP Soman, A novel method for detecting R-peaks in electrocardiogram(ecg) signal, Elsevier, Biomedical Signal Processing and Control, Issue 7, pp , 2012 [7] C Saritha, V Sukanya, ECG Signal Analysis Using Wavelet Transform, Bulg J Phys Issue 35, pp68 77, 2008 [8] Gordan cornelia, reiz romulus, ECG signals processing using wavelets, [9] Anuja sas, Shreetam behera, Identification of Tachycardia and Bradycardia Heart Disorders using Wavelet Transform based QRS Detection, International Journal of Image Processing and Vision Sciences,Volume-2 Issue-1, pp , 2013 [10] Nahit Emanet, Ecg Beat Classification By Using Discrete Wavelet Transform And Random Forest Algorithm, IEEE, 2009 [11] SSumathi, DrMY Sanavullah, Comparative Study of QRS Complex Detection in ECG Based on Discrete Wavelet Transform, Int Journal of Recent Trends in Engineering, Vol 2, No:5, pp , 2009 [12] Massachusetts Institute of Technology, MITBIH arrhythmia database [ /physiobank/database/mitdb] RES Publication 2012 Page 44 wwwijmeceorg

6 Volume No - 5, Issue No 3, May, 2017 [13] Massachusetts Institute of Technology, CU database [ physionet org / physiobank /database/cudb] [14] Massachusetts Institute of Technology, MITBIH SupraventricularArrhythmiadatabase[ etorg /physiobank/database /svdb] [15] DCReddy Biomedical signal processing principles and techniques (1st edition), New Delhi: Tata McGraw- Hill Publishing Company Ltd, 2005 RES Publication 2012 Page 45 wwwijmeceorg