Electrocardiogram (ECG) is a quasi-periodic signal which reveals the activity of the heart. A lot of information on the normal and pathological physiology of heart can be obtained from ECG signal. However, due to the nonstationary nature of ECG signals, it is very difficult to visually analyze them. Thus there is a need for computer based methods for ECG signal Analysis.

A huge amount of work has been done in the field of ECG signal analysis using various approaches and methods. The basic principle of all the methods, however involves transformation of ECG signal using different transformation techniques including Fourier Transform, Hilbert Transform, Wavelet transform, etc. Physiological signals like ECG are considered to be quasi-periodic in nature. They are of finite duration and non-stationary. Hence, a technique like Fourier series (based on sinusoids of infinite duration) is inefficient for ECG. On the other hand, wavelet, which is a very recent addition in this field of research, provides a powerful tool for extracting information from such signals. There has been use of both Continuous Wavelet Transform (CWT) as well as Discrete Wavelet Transform (DWT). However, CWT has some inherent advantages over DWT. Unlike DWT, there is no dyadic frequency jump in CWT. Moreover, high resolution in time-frequency domain is achieved in CWT.

The signal processing of an ECG wave mainly consists two parts, first is to make the signal noise free and second is the detection of R-peaks in the signal.

A huge amount of research has been done in the area of signal processing, and various algorithms were given for peak detection of signals. In 1996, Jeongwhan Lee, et al. developed a real-time algorithm for detection of the QRS complexes of ECG signals. It reliably recognizes QRS complexes based upon digital analyses of slope, amplitude, and width. A special digital band pass filter reduces false detections caused by the various types of interference present in ECG signals. This filtering permits use of low thresholds, thereby increasing detection sensitivity. The algorithm automatically adjusts thresholds and parameters periodically to adapt to such ECG changes as QRS morphology and heart rate. For the standard 24 h MIT/BIH arrhythmia database, this algorithm correctly detects 99.3 percent of the QRS complexes [1].

Then a computer algorithm for real-time triggering at the onsets of the R waves of a one channel electrocardiogram is described in 1989. The algorithm is intended for timing the deflation of intra-aortic balloon pumping with a control strategy based on real-time triggering. Thus, early detection of the R waves is the primary design goal. The algorithm consists of three steps: algorithm performs robustly and the point of detection is typically before the peak of an R wave [2].

M. Ibrahim Sezan developed a new automatic peak detection algorithm and applied to histogram-based image data reduction (quantization) in the year 1990. The algorithm uses a peak detection signal derived either from the image histogram or the cumulative distribution function to locate the peaks in the image histogram. Specifically, the gray levels at which the peaks start, end, and attain their maxima are estimated. To implement data reduction, gray-level thresholds are set between the peaks, and the gray levels at which the peaks attain their maxima are chosen as the quantization levels [3].

A.M. Bianchi, et al. in their paper introduced a time-variant algorithm of autoregressive (AR) identification and applied to the Heart Rate Variability (HRV) signal in 1993. The power spectrum is calculated from the AR coefficients derived from each single RR interval considered. Time-variant AR coefficients are determined through adaptive parametric identification with a forgetting factor which obtains weighed values on a running temporal window of 50 preceding measurements. Power Spectrum Density (PSD) is hence obtained at each cardiac cycle, making it possible to follow the dynamics of the spectral parameters on a beat-by-beat basis. These parameters are mainly the LF (Low Frequency), and the HF (High Frequency) powers, and their ratio LF/HF. These together account for the balanced sympathovagal control mechanism affecting the heart rate. This method is applied to subjects suffering from transient ischemic attacks. The time variant spectral parameters suggest an early activation of LF component in the HRV power spectrum. It precedes by approximately 1.5-2 min of the tachycardia and the ST displacement, generally indicative of the onset of an ischemic episode. The results suggest an arousal of sympathetic system before the acute attack [4].

Soon in 1995, a research reports a simple and efficient method to detect the QRS-segment in ECG signals. The method suggested by the authors is implemented in two steps. The Step I forms the detection of R-peaks which is accomplished after three main operations: band pass Altering, %sample area estimation and M-sample window averaging with thresholding. While the band pass Altering suppresses the noise, the 3-sample area estimation enhances the Signal to Noise Ratio (SNR) and hence the detectability of R-peaks. Finally, M-sample window averaging smooths the R-peak envelopes. The thresholding pinpoints the IC-peaks. After R-peaks are detected, their location is confirmed in Step 1 which also marks the location of Q and S points resulting in determination of respective QRS- segments [5].

Then in 2005, a novel mechanism was proposed to detect the ECG signal based on adaptation Wavelet Transform (WT). The authors adopt the sub-strap coded theory of WT and restore post-combination weak signal through adaptation matching of multi sub-straps' weights. The simulation results show that the proposed method can further improve the ability to detect signal, which is a very effective scheme to detect the property of weak signal and improve SNR [6].

A paper in 2007 described a Predictive Neural Network (PNN) based technique to detect QRS complexes of electrocardiograms (EGGs). The PNN is trained, using the back propagation algorithm, on non-QRS portions of the EGG to predict the signal one-step ahead. High prediction error is then taken as an indication of the occurrence of a QRS complex. A simple peak detection logic is then invoked to mark the exact location and magnitude of either a Q- or an R- or an S- peak within the QRS complex. The performance of the detector software is illustrated with examples representative of different QRS morphologies. The accuracy of QRS complex detection has been tested using bipolar standard limb leads of a standard EGG library; a sensitivity of 98.96% has been achieved. A brief discussion on how well this technique performs in comparison with the other QRS detectors is also presented [7].

An advanced automated algorithm to preprocess RR intervals obtained from a normal ECG was presented in 2010. Validation of this algorithm was performed on one hour ECG signals of 20 pregnant women. R peaks before and after preprocessing were manually revised for spurious MITand missed R peak detections. Before preprocessing, more than 1% of the detected R peaks were incorrect while preprocessing corrected more than 94% of these errors leading to an overall error rate of 0.06%. The automated preprocessing technique therefore restricts the manual data check to the absolute minimum and allows a reliable HRV analysis [8].

The performance of four R-wave detection methods that were applied on newborn piglet ECG data were analysed in the same year. These methods are based on: first derivative, wavelet transform, and nonlinear transform. The results of the performance analysis showed that the nonlinear approach based on the Hilbert transform marginally outperformed the others, with the highest sensitivity (Se) of 99 9.5%, the lowest detection error (ER) of 01.2% and a high positive prediction (+P) of 99 9.3% [9] .

Later in 2010, the evaluation of mental stress measurement using heart rate variability was presented. The heart rate signals are processed first using Fourier transform, and then it is applied to wavelet transform. The activity of the autonomic nervous system is noninvasive studied by means of autoregressive (AR) frequency analysis of the heart-rate variability (HRV) signal. Some methods of noise rejection and robustness for AR recursive identification are presented that make on-line frequency analysis of the heart-rate variability signal more reliable. The goodness of the algorithms is first tested through simulations, and then results obtained on real data during ischemic episodes are presented. Spectral decomposition of the Heart Rate Variability during whole night recordings was obtained, in order to assess the characteristic fluctuations in the heart rate and their spectral parameters during REM and NREM sleep stages. Mental stress is accompanied by dynamic changes in Autonomic Nervous System (ANS) activity. Heart Rate Variability (HRV) analysis is a popular tool for assessing the activities of Autonomic Nervous System. This paper presents a novel method of HRV analysis for mental stress assessment using fuzzy clustering and robust identification techniques. The approach consists of 1) online monitoring of heart rate signals, 2) signal processing (e.g., using the continuous wavelet transform to extract the local features of HRV in time- frequency domain) [10].

Again in 2010, P. Sasikala developed a robust R Peak and QRS detection algorithm using Wavelet Transform. Wavelet Transform provides efficient localization in both time and frequency. Discrete Wavelet Transform (DWT) has been used to extract relevant information from the ECG signal in order to perform classification. Electrocardiogram (ECG) signal feature parameters are the basis for signal Analysis, Diagnosis, Authentication and Identification performance. These parameters can be extracted from the intervals and amplitudes of the signal. The first step in extracting ECG features starts from the exact detection of R Peak in the QRS Complex. The accuracy of the determined temporal locations of R Peak and QRS complex is essential for the performance of other ECG processing stages. Individuals can be identified once ECG signature is formulated. This is an initial work towards establishing that the ECG signal is a signature like fingerprint, retinal signature for any individual Identification. Analysis is carried out using MATLAB Software. The correct detection rate of the Peaks is up to 99% based on MIT-BIH ECG database [11].

Then in 2011, a simple and reliable method termed the Correlation Integral Method (CIM) was proposed to detect the R-wave in ECG signal. The proposed method includes two stages. The first stage is to partition several small segments to an ECG signal and realize to the phase space reconstruction. The second stage is to calculate the correlation integral between these small segmentations. According to the position where the R-wave appears is the position at which the value of the correlation integral is at the minimum, the proposed method enables us to detect the R-wave even when there is a significant amount of high frequency noise in the ECG signal, or when the R peak abruptly reverses. Some records of ECG signals in MIT/BIH Arrhythmia Database is tested to show the CIM has a much more precise detection rate and robust than other methods [12].

Then in 2011, H. Rabbani et al. employed Hilbert and wavelet transforms as well as adaptive thresholding method to investigate an optimal combination of these signal processing techniques for the detection of R peak. In the experimental sections of their paper, the proposed algorithms are evaluated using both ECG signals from MITand BIH database and synthetic data simulated in MATLAB environment with different arrhythmias, artifacts, and noise levels. Finally, by using wavelet and Hilbert transforms as well as by employing adaptive thresholding technique, an optimal combinational method for R peak detection, namely WHAT is obtained that outperforms other techniques quantitatively and qualitatively [13] .

In the same year, Z. S. Wang and J. D. Z. Chen investigated whether electrical stimulation is vagally mediated by assessing the Heart Rate Variability (HRV). The study is performed in six healthy female hound dogs implanted with four pairs of bipolar serosal electrodes, which are used to measure gastric myoelectrical activity. A special fuzzy neural network, which is called Evolutionary- Programmingbased Fuzzy Inference System (EPFIS), is developed to identify the R-R wave to precisely extract the R-R interval and derive the HRV data. A high-resolution adaptive timefrequency analysis method based on ARMA modeling developed previously in OUT laboratory is used to obtain high-quality HRV spectral parameters [14].

Then in this year only a paper named "Variable Threshold Method for ECG R-peak Detection" used differentiation process and Hilbert transform as signal preprocessing algorithm. Thereafter, variable threshold method is used to detect the R-peak which is more accurate and efficient than fixed threshold value method. R-peak detection using MIT- BIH databases and Long Term Real-Time ECG is performed in this research in order to evaluate the performance analysis. For the removal of the noise that contained in biomedical signal, monitoring program able to digitalized the ECG using 35 Hz of 4 IIR Digital Low-pass filter, 0.05 Hz high pass filter, and 60 Hz notch filter. In preprocessing part, the input ECG signal undergoes differentiation process, then Hilbert transform. The experimental performance of the ECG preprocessing is performed. Firstly, ECG signals of MIT-BIH Arrhythmia Database100 in the range of 0 to 10 s are used. Then, the signal undergoes differentiation process. The final part of preprocessing is Hilbert transform process. In R-peak detection of ECG signal, it generally used the threshold value which is fixed. However, it does not give accurate results in R-peak detection when the baseline changes due to signal size and motion artifacts [15].

Then in 2012, a paper proposed a simple algorithm for automatic detection of the R-peaks from a single lead digital ECG data. The squared double difference signal of the ECG data is used to localise the QRS regions. The proposed method consists of three stages: sorting and thresholding of the squared double difference signal of the ECG data to locate the approximate QRS regions, relative magnitude comparison in the QRS regions to detect the approximate R-peaks and RR interval processing to ensure accurate detection of peaks. The performance of the algorithm is tested on 12-lead ECG data from the PTB diagnostic ECG database, and a high detection sensitivity of 99.8% with low computational complexity and low sensitivity to low frequency noises is detected [16] .

F. Scholkmann, et al., presented a new method for automatic detection of peaks in noisy periodic and quasiperiodic signals later in 2012. The new method, called Automatic Multiscale-based Peak Detection (AMPD), is based on the calculation and analysis of the local maxima scalogram, a matrix comprising the scale-dependent occurrences of local maxima. The usefulness of the proposed method is shown by applying the AMPD algorithm to simulated and real-world signals [17] .

An automatic approach for detecting QRS complexes and evaluating related R-R intervals of ECG signals (PNDM) is proposed in the same year. It reliably recognizes QRS complexes based on the deflection occurred between R & S waves as a large positive and negative interval with respect to other ECG signal waves. The proposed detection method follows new fast direct algorithm applied to the entire ECG record itself without additional transformation like Discrete Wavelet Transform (DWT) or any filtering sequence. Mostly used records in the online ECG database (MIT-BIH Arrhythmia) have been used to evaluate the new technique. Moreover it was compared to seven existing techniques; the results show that PNDM has much detection performances according to 99.95% sensitivity and 99.97% specificity. It is also quickest than comparable methods [18].

T.R.G. Nair, et al. in 2013 used an adaptive wavelet approach to generate an appropriate wavelet for R-signal identification under noise, baseband wandering, and temporal variations of R-positions. This study designed an adaptive wavelet and successfully detected R- peak variations under various ECG signal conditions. In most of the cases, the error is either zero or less than 2%. Error was more in cases of missing P-peaks or merged P-peaks. When there was variability in R-R intervals, the algorithm modified the search intervals and detected the R-peak correctly. False detection was avoided by removing peaks which are occurring within less than 120 ms [19].

Then in 2013, H. Lin, et al. showed DWT as an efficient method for analyzing non-stationary signals like ECG signal. The Symlets wavelets (sym5) and soft-thresholding are chosen as the wavelet function and thresholding method to do noise correction at the first denoising stage. The second stage is R wave detection. The MIT- BIH arrhythmia database is used to verify the proposed algorithm. We reconstruct the decomposition level 3 to 5. Choosing the adaptive threshold and window size are the key points to reduce error rate. Applying two thresholds leads to better performance, compared to applying one threshold. At the last stage, noise correction is done again. With the information of R wave position, a novel method is proposed to eliminate the electromyogram (EMG) signal. The algorithm for R wave detection has a sensitivity of 99.70%, and a positive predictivity of 99.65%. The error rate is 0.65% under all kinds of situation (0.37% if ignoring 3 worst cases). For noise correction, the SNR improvement is achieved at least 10,dB at SNR 5,dB, and most of the improvement SNR are better than other methods at least 1dB at different SNR. To apply presented algorithms for the portable ECG device, all R peaks can be detected no matter when people walk, run or move at the speed below 9 km/hr [20].

A method of electrocardiogram (ECG) signal pretreatment is presented by the application of Discreet Wavelet Transform (DWT) by automatically determining the optimal order of decomposition in 2013. After the purification of the original signal, they describe an algorithm to detect R waves based on the Dyadic Wavelet Transform DyWT by applying a windowing process. This algorithm is validated on a sample of synthesis ECG signal with and without noise which they have proposed and on real data. Finally, once the R peaks of real data are detected, three methods of RR intervals analysis were used by calculating the standard deviation of heart rate and applying the Fast Fourier Transform (FFT), and the Wavelet Transform on detected RR intervals to study the Heart Rate Variability (HRV). A comparative study between the analysis results of detected RR intervals in healthy and diseased subjects through the application of the FFT and the Wavelet Transform was given [21].

A paper titled "Design of Digital IIR Filter for Noise Reduction in ECG Signal" proposed in the same year is a digital infiniteimpulse response (IIR) filter design. This includes an implementation and evaluation of butterworth low pass infinite impulse response filter method to remove high frequency noise and for this filter is applied to noisy ECG data sample and original samples are taken as reference signal. The suggested method considers the magnitude response for choosing the cutoff frequency and the FFT spectrum estimate response to find the lowest filter order. The structure and the coefficients of the digital IIR filter are designed using FDA tool in MATLAB. The filter output's average power before and after filtration are calculated using FFT and for simulation of this filter, the hardware is designed using microcontroller ATmega 16A. For hardware designing, the samples taken are record no. 108 and record no.119 (taken from MIT-BIH database, ML II signal). Here samples are taken from MIT-BIH arrhythmia database (mitdb) ML II are used [22].

Later in this year, R. Alonso, et al. compared three of the best QRS detection algorithms, regarding their results, to check the performance and to elucidate which get better accuracy. They implemented two algorithms based on digital filters, Pan & Tompkins algorithm and Hamilton & Tompkins algorithm and a new algorithm based on the phasor transform. They show that the performance of Pan & T. and Hamilton & T. is good, with values between 99.42 and 99.84. Phasor transform has worse performance, specially the positive predictivity. In terms of sensitivity, they concluded that Pan & T. algorithm achieved the best results [23].

In the next year, Indu Saini, et al., explored a classifier motivated from statistical learning theory, i.e., Support Vector Machine (SVM), for detection and delineation of these wave components. Digital filtering techniques are used to remove interference present in ECG signal. The feature extraction is done using a modified definition of slope of the ECG signals. The performance of the proposed algorithm is validated using ECG recordings from dataset-3 of the CSE multi-lead measurement library. The results in terms of accuracy, i.e., 94.4%, obtained clearly indicate a high degree of agreement with the manual annotations made by the referees of CSE dataset-3 [24].

A method is discussed for detection of R peak of ECG signal in a research work in 2014. The main objective of this work is to test the method using ECG of Indian patients. ECG signals from Modified Lead II (MLII) are chosen for processing. The location of the R peak is detected using the proposed algorithm. The error in the location of the R peaks is within the permissible limit. So the algorithm is acceptable for ECG signal processing. Using the location of the R peaks the RR interval can also be calculated and on the basis of the measured value, the cardiac arrhythmias can be detected. The algorithm can be modified further to calculate all the ECG parameters automatically at the output. Thus, it can be used for the processing of real time ECG signals in future [25].

An algorithm for R-peak detection is implemented in 2014 using filter and Discrete Wavelet Transform (Haar transform). The information about R-peak obtained is very useful for classification, analysis and arrhythmia detection such as Tachycardia and Bradycardia. From this technique, Tachycardia and Bradycardia can be easily identified from R-peak location and this technique gives the best possible locations for R-peak. The approximation and detailed coefficient are also plotted using Discrete Wavelet Transform (Haar transform). From the R-peak location R-R interval can be easily estimated that will provide best possible heart beats for humans. The main advantage of this kind of detection is less time consuming for long time ECG signal [26].

In this year H.M.R.A. Trivedi and K.C.S. Shukla employed the Daubechies Wavelet Transform (WT) for R-peak detection and Radial Basis Function Neural Network (RBFNN) to classify the electro- cardiogram (ECG) signals in their paper. Five types of ECG beats: normal beat, paced beat, Left Bundle Branch Block (LBBB) beat, Right Bundle Branch Block (RBBB) beat, and Pre-mature Ventricular Contraction (PVC) were classified. 500 QRS complexes were arbitrarily extracted from 26 records in Massachusetts Institute of Technology- Beth Israel Hospital (MIT-BIH) arrhythmia database, which are available on Physionet website. Each and every QRS complex was represented by 21 points from p1 to p21 and these QRS complexes of each record were categorized according to types of beats. The system performance was computed using four types of parameter evaluation metrics: sensitivity, positive predictivity, specificity, and classification error rate. The experimental result shows that the average values of sensitivity, positive predictivity, specificity and classification error rate are 99.8%, 99.60%, 99.90% and 0.12%, respectively with RBFNN classifier. The overall accuracy achieved for Back Propagation Neural Network (BPNN), multilayered perceptron (MLP), support Vector Machine (SVM) and RBFNN classifiers are 97.2%, 98.8%, 99% and 99.6%, respectively. The accuracy levels and processing time of RBFNN is higher than or comparable with BPNN, MLP and SVM classifiers [27] .

In the same year, S. S. M. Sabrigiriraj used a variable step size delayed LMS adaptive filter in his paper to remove the artifacts from ECG signal for improved feature extraction. Moreover, the extraction of R peak in ECG is carried out using Discrete Wavelet Transform based QRS detection algorithm. Due to the high speed of this method, the ECG de-nosing and R Peak extracting could be both realized at real time, which is an effective method to monitor the patients in biotelemetry applications. A variable step size delayed LMS adaptive filter is presented to remove the artifacts from ECG signal. Moreover, the extraction of R peak in ECG is carried out using DWT. The proposed denoising algorithm requires 2 N multiplications, 2 N + 1 additions and one delay which is comparably less than other existing algorithms. There is a considerable reduction in the MSE which leads to high Signal-to-Noise Ratio. The proposed denoising technique with Meyer wavelet has a SNR of 46.89 dB which is much higher than other wavelets. The denoised ECG is processed using the R peak detection algorithm with Coif wavelet to determine the maximum possible beats per minute. The proposed implementation is suitable for biotelemetry applications, where large signal to noise ratios with less computational complexity are required [28].

Next year, R-peak detection has been proposed by A.T. Bhatti in his paper because it is adaptive to the nonlinear and time varying features of ECG signal. It can be trained to recognize the normal waveform and filter out the unnecessary artifacts and noises. Usually R-peak detection in QRS complex can be improved by considering multiple features, including RR interval, pulse duration and amplitude. This research paper, takes the difference of maximum original signal and minimum original signal to obtain the filtered R-peak ECG signal after 1^st and 2^nd pass to observe noise for 512 sample points at sampling frequency of 1 KHz using High Order Statistics Algorithm. Once the R- peak is detected, compute the Mean Square Error (MSE), and Peak Signal to Noise Ratio (PSNR). From analytical perspective of this research, it appears to be exceedingly robust, correctly detects R-peaks even aberrant QRS complexes in noise-corrupted ECG signal compression [29].

In 2015, a paper titled "R-Peak Detection using Wavelet Transforms" presented a technique based on wavelet transforms to analyze the electrocardiogram signal (ECG) for the detection of the R peaks. They located the QRS complexes of this signal using the Dyadic Wavelet Transform (DyWT) and detect the R peaks using the Direction Change Mark (DCM) method. Applied on the ECG signal database MIT-BIH, the algorithm that they presented gave good results for detecting the R waves with a rate of 99.91%, a sensitivity of 99.95 % and a positive predictivity of 99.96 % [30].

Again in 2015, W.K.L.Ee, et al. developed a smart electrocardiography (ECG) patch. The smart patch measures ECG using three electrodes integrated into the patch, filters the measured signals to minimize noise, performs analog-to-digital conversion, and detects Rpeaks. The measured raw ECG data and the interval between the detected R-peaks can be recorded to enable long-term HRV analysis. Experiments were performed to evaluate the performance of the built-in wave detection, robustness of the device under motion, and applicability to the evaluation of mental stress. The Rpeak detection results obtained with the device exhibited a sensitivity of 99.29%, a positive predictive value of 100.00%, and an error of 0.71%. The device also exhibited less motional noise than conventional ECG recording, being stable up to a walking speed of 5 km/h. When applied to mental stress analysis, the device evaluated the variation in HRV parameters in the same way as a normal ECG, with very little difference. This device can help users better understand their state of health and provide physicians with more reliable data for objective diagnosis [31].

Later in 2015, a paper proposed a new approach which combines matched filtering and Hilbert Transform. The RRintervals and cross-correlation are used in conjunction to not only automatically locate the R-peaks, but also to display the candidate ambiguous peaks via an interactive graphical user interface. The performance of the proposed approach is compared to the well-known Pan- Tompkins algorithm and is evaluated for two types of ECG databases: standard stationary data and low-SNR ECG data obtained from wearable ECG. The proposed method results in a distinctly higher positive predictivity and leads to more satisfying overall outcomes, especially for the critical call of low-SNR data [32].

A novel Genetic Optimized Wavelet Thresholding (GOWT) approach was proposed in a research work presented by H. He, et al., in the same year. A Quadratic Curve Thresholding Function (QCTF) was devised to realize the smooth connection of threshold points. Moreover, in terms of the root mean square error and the filtering smoothness, a new genetic algorithm was devised to automatically search the optimal parameters of QCFT for different noisy signals. Finally, the GOWT was evaluated and compared with hard thresholding and soft thresholding by means of MIT-BIH arrhythmia database ECG records. The filtering results indicate that the GOWT can realize smooth threshold transition, avoiding the oscillation at the cutoff threshold point caused by the hard thresholding and the wavelet coefficient bias brought by the soft thresholding. Its adaptability to various signals has been strengthened by the genetic algorithm. The GOWT can find a trade-off between the smoothness and distortion of signal filtering, generating the desirable noise-free signal for feature extraction [33].

G. Chen, et al. presented in the same year, an R peak adaptive threshold extraction algorithm by setting thresholds and re-examining the mechanism for each cardiac cycle to improve the accuracy of the threshold algorithm. Simulation results show that the method is much simpler and more effective, and can perform real-time processing, which significantly improves the accuracy of the R peak detection and meets the requirements of clinical applications [34].

Then in 2016, S.S. Mehta and Lingayat presented two algorithms in their paper for the detection of QRScomplexes in Electrocardiogram (ECG). The first uses single lead ECG at a time for the detection of QRS complexes, while the second uses 12-lead simultaneously recorded ECG. The ECG signal is filtered using digital filtering techniques to remove power line interference and base line wander. Support Vector Machine (SVM) is used as a classifier for detection of QRS-complexes in ECG. Using the standard CSE data-set3, both the algorithms performed highly effectively. The performance of the algorithm with sensitivity (Se) of 99.70% and positive prediction (+P) of 97.75% is achieved when tested using single lead ECG. It improves to 99.93% and 99.13% respectively for simultaneously recorded 12-lead ECG signal. The percentage false positive and false negative is low. The proposed algorithm performs better as compared with published results of other QRS-detectors tested on the same database [35].

In 1997, K.D. Rao presented an algorithm based on Discrete Wavelet Transforms for detection of R-peaks, computation of R-R wave time interval and data compression ECG signals. The advantage of the DWT based data reduction is that no assumptions need to be made regarding the nature of the signal to be compressed and the percentage data reduction and PRD are superior to that were obtained with the other data reduction algorithms. The advantages of the DWT based detection of R-peaks and computation of R-R wave time interval are that it does not assume stationarity or quasi stationarity within the analysis segment, detects R peaks and estimates R-R wave interval very accurately exhibiting robustness to noise and this algorithm is computationally simple since it needs to compute the DyWT at three or four scales [36] .

Initially there were time based algorithms, which were used to detect the peaks by analyzing the magnitude, slope and width. These algorithms were proved well enough, but error rate was high (i.e) false peak detection. Digital band pass filters were used to decrease the false peak detection in these algorithms. Then frequency based algorithms were being used as FFT, DFT, IDFT, etc. These algorithms were good, but proved to be inferior when wavelets were being used for peak detection, especially Morlet which looks very similar to the QRS segment. Then some advanced methods were used to identify the peaks as ANN, entropy based methods, Support Vector Machine based algorithms, etc. with the accuracy of more than 99.9%.

In this paper, several techniques for signal processing of an ECG wave are discussed. From the beginning till present, various techniques and algorithms were used to have a more predictive method with low value of error rate. The error rate has been reduced from around 1% to 0.3% in last few years. The review reveals that different techniques starting from Hilbert Transform to wavelets and then adaptive wavelets, etc. were used. Some techniques like ANN, genetic algorithm, entropy based methods, SVM, etc. have shown effective results and can be analyzed and used for ECG signal preprocessing in future.