Portal de Periódicos da CAPES

Efficient FPGA‐based VLSI architecture for detecting R‐peaks in electrocardiogram signal by combining Shannon energy with Hilbert transform

2018; Institution of Engineering and Technology; Volume: 12; Issue: 6 Linguagem: Inglês

10.1049/iet-spr.2017.0201

1751-9683

Madam Aravind Kumar, K. Manjunatha Chari,

Analog and Mixed-Signal Circuit Design

IET Signal ProcessingVolume 12, Issue 6 p. 748-755 Research ArticleFree Access Efficient FPGA-based VLSI architecture for detecting R-peaks in electrocardiogram signal by combining Shannon energy with Hilbert transform Madam Aravind Kumar, Corresponding Author Madam Aravind Kumar aravindkumar0417@gmail.com Electronics & Communication Engineering, GVVIT Engineering College, Andhra Pradesh, IndiaSearch for more papers by this authorKamsali Manjunatha Chari, Kamsali Manjunatha Chari Electronics & Communication Engineering, GITAM University, Andhra Pradesh, IndiaSearch for more papers by this author Madam Aravind Kumar, Corresponding Author Madam Aravind Kumar aravindkumar0417@gmail.com Electronics & Communication Engineering, GVVIT Engineering College, Andhra Pradesh, IndiaSearch for more papers by this authorKamsali Manjunatha Chari, Kamsali Manjunatha Chari Electronics & Communication Engineering, GITAM University, Andhra Pradesh, IndiaSearch for more papers by this author First published: 01 August 2018 https://doi.org/10.1049/iet-spr.2017.0201Citations: 13AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract Electrocardiogram (ECG) is a critical application in light of R-peak detection. The R-peaks are impacted by some QRS complex and noises in the current method. Basic testability, snappier execution, and confirmation choices are accomplished by the utilization of the field programmable gate array (FPGA). But, FPGA execution gives less accuracy. To battle execution trouble, another R-peak detector is proposed and relying upon the propelled peak finding logic, which incorporates a Bandpass filter and first-order differentiation process, which are done in the primary phase of the strategy. The noises in the input ECG signal are decreased in the first stage. In the subsequent stage, the smooth Shannon energy envelope (SEE) is acquired by utilizing SEE extraction and the zero phases filtering process. The false R peak is stifled by proposing a Hilbert transform (HT) in the third stage. The HT requires more hardware space and high-power utilization in the current technique. Due to this reason, Real valued Fast Fourier Transform (RFFT) and Inverse RFFT (IRFFT) techniques are proposed in the HT. In the fourth stage, R-peak acknowledgment is evaluated using Massachusetts Institute of Technology, Beth Israel Hospital (MIT-BIH) arrhythmia database and produced the average accuracy of 99.86%, sensitivity of 99.95% and the positive predictivity of 99.90%. 1 Introduction The field programmable gate array (FPGA) generally utilised as a part of the helpful field application in light of its low-power usage and versatile programming. The FPGA consists of reconfigurable logic, I/O and interconnection blocks, but differs from microcontrollers and Digital Signal Processing (DSP) processors that are Von Neumann type of machines [1]. In order to diagnose heart diseases, the R-peak detection in electrocardiogram (ECG) [2-5] is one of the universal biological signals that play a significant role. The existing R-peak detection strategies generally influenced by some QRS complex [6]. For detection of R-peaks [7], Pan-Tompkins method [8], FPGA-based method [9], empirical wavelet transform [10], Hilbert transform (HT) and adaptive thresholding [11], derivative-based algorithm [12], neural network filter [13], lifting wavelet transform and HT [14], entropy method [15], mathematical morphology [16], empirical mode decomposition [17], geometrical matching approach [18], wavelet and artificial neural network [19], are the various methods reported by researchers. Pan-Tompkins (PT) method is an effective algorithm for QRS complex detection and peak identification, yet the accuracy is very low in this algorithm [20]. The wavelet-based strategy is wasteful to diminish the QRS events in light of the fact that the computational load is very high [21-25]. The HT and the empirical mode decomposition (EMD) method give high accuracy in R-peak identification yet these strategies required more hardware space and high-power utilisation [26]. The SE with HT method gives better accuracy in R-peak identification yet these strategies required high memory and delays [27]. The above-mentioned limitations on the preprocessor of the existing R peak finders show that the recognition of R peak is an extremely a testing task. The proposed technique outlined an advanced R-peak detector fuses Shannon energy envelope (SEE) extraction and HT to diminish the QRS complex and noises. The current R-peak location methodology moreover used the HT; be that as it may, it requires more hardware space and high power use. The essential goal of this paper is to reduce the hardware space and high-power usage and besides increasing the accuracy of the proposed preprocessor. To fight the inconvenience of the current system, we have to display real valued fast Fourier transform (RFFT) and inverse RFFT (IRFFT) procedure in the HT. The HT and the MA filters are not required in the current FPGA implementation strategy. Our proposed strategy has the ideal accuracy over the current FPGA execution technique. This paper is portrayed underneath. In Section 3, the proposed R-peak detection strategy is delineated. The detail about the advanced peak finding logic portrays in Sections 3.3 and 3.4. The experimental result and the conclusion are discussed in Sections 4 and 5. 2 Related works An automated peak finding logic in light of new preprocessing technique was proposed by Manikandan and Soman [28]. In their method, the HT and the moving average (MA) filter are utilised to lessen QRS edifices and undesirable QRS polarities. The heart rate monitoring system in light of the FPGA implementation method was portrayed by Panigrahy et al. [29]. To distinguish the R-peaks in ECG, the five stage methodology was utilised in their method. Also, the accuracy of 99.84%, sensitivity of 99.94%, positive predictivity (PPV) of 99.89% is measured. Lee et al. [30] have developed a smart ECG patch. Keeping in mind the end goal to diminish the noise level, the Analog-to-digital conversion was performed. The smart patch was used to measures ECG and furthermore filters the deliberate ECG signal to limit noise and distinguishes R-peaks. A new ECG signal-processing approach based on EMD and an improved approximate envelope method was presented by Li et al. [31]. In order to eradicate high-frequency noises before the EMD, a butterwort low-pass filter is used. Yu et al. [32] have anticipated a specific fusion algorithm of detecting multi-channel maternal ECG R-peak locations. The wavelet transform (WT) for R-peak detection and radial basis function neural network to classify the ECG signals were employed by Rai et al. [33]. A variable step size delayed LMS adaptive filter was used to remove the artefacts from the ECG signal for improved feature extraction was proposed by Satheeskumaran and Sabrigiriraj [34]. In order to carry out the extraction of R peak in ECG discrete WT-based QRS detection algorithm is used. 3 Proposed methodology for R-peak detection The goal of this paper is to detect the accurate R-peaks in the ECG signal to identify the patient disease types. The R-peaks are impacted by some QRS complex and noises in the current method. The FPGA implementation method is a well-known approach, but it gives less accuracy. To battle execution trouble, the proposed method utilised the RFFT and IRFFT techniques in the HT. The block outline of the proposed R-peak recognition technique is shown in Fig. 1 a digital filtering, envelope extraction; peak finding logic and R-peak detection are the four phases. Bandpass filter and first request separation process are the starting phase of the proposed R-peak detection technique. So as to expand the state of the QRS complex and limit the noises in the input ECG signal, bandpass filter and the first-order differentiation process are utilised. The smooth Shannon energy envelope (SEE) is gotten from the SEE extraction and the zero phase filtering process in the envelope extraction stage. The SEE process additionally handled the peak detection, false R-peak detection, and false noise detection. The third stage incorporates the HT and MA filter. The ideal position of the R-peak is recognised in the third stage. At long last, the precise area of the R-peak is acquired. The detailed discourses on each stage in Fig. 1 a are clarified in underneath subdivisions. Fig. 1Open in figure viewerPowerPoint Proposed System Model (a) Block schematic of proposed cardiac analyser, (b) Architecture of bandpass filter, (c) Block diagram for hilbert transform, (d) Block schematic for RFFT 3.1 Noise reduction and QRS complex enhancement For confirming the electrical action of the heart, the ECG is the course of action. The ECG signal is acquired from the patient's heart cadence. Different types of noise such as muscle contraction, control line impedance, anode contact commotion, movement ancient rarities, large P, T waves damages the ECG signal. In order to minimise the noises in the ECG signal and increase the shape of the QRS complex, the digital filtering process is used. The digital filtering stage includes the bandpass filter and first-order differentiation filter. These are used for noise reduction and the QRS complex enhancement. 3.1.1 Bandpass filter The accurate R peak detection is the primary objective of this paper. A bandpass filter is utilised to stay away from the undesirable frequencies in the signal and decreasing the interference and rivalry among signals as it grants signals between two particular frequencies as it were. So in the proposed work, an FIR bandpass filter (6–18 Hz) is utilised for subtracting the noises in the ECG signal. For the execution of the FIR bandpass filter, a direct form-1 is used and it is given as (1)In the above equation, the yield of the bandpass filter is spoken to as y, the shifted matrix is spoken to as and the coefficient matrix is spoken to as and N is the length of the coefficients. Fig. 1 b shows the architecture of the digital bandpass filter (condition 1) which joins D-flip flops, multipliers () and adders () for the usage of the bandpass filter. To process the coefficient matrix of the FIR bandpass filter, at first, every one of the coefficients are changed over to the fixed point 16-bits format of the real number. The input values are put away by utilising the shifted shape. At first, the ECG signal x and the clock signal are going about as the input of the D-flip flop. A similar signal x and the parameter value are multiplied by the assistance of the multiplier and the output of the multiplier is summed by using the adder block for the. A similar procedure is rehashed until the last stage. At long last, the last adder block output is the output of the bandpass filter. 3.1.2 First-order differentiation filter Keeping in mind the end goal to reduce the large P and T wave interference and improve the shape of the QRS complex, the first-order differentiation filter is utilised. Here, the yield of the bandpass filter (filter signal z) is going about as the input of the first-order differentiation filter. The first-order differentiation filter separating the filtered signal z by utilising one subtraction block and it yields the data about QRS complex. In this way, the output of the first-order differentiation filter is a bipolar signal. It is figured as (2)where is the differentiated signal and is filtered signal. The negative R-peaks are identified by utilising adaptation and afterwards the differentiated signal is standardised after the first-order differentiation process and it is registered as (3)where is the normalised signal of and K is the number of samples in the ECG signal. 3.2 Envelope extraction The envelope extraction stage incorporates the SEE extraction and the zero-phase filtering process. 3.2.1 SEE extraction The R peaks related to quick and high accuracy in SEE extraction stage. The state of the QRS complex is kept up in view of the first-order differentiation filter output . This SEE extraction arrange used for keeping up the shape of the QRS complex, subtracting the extensive P and T waves and diminishing the non-stationary noises. In any case, is a bipolar signal and it incorporates both positive and negative qualities. In any case, in the peak detection process, just the positive value of the R peak is required. So the bipolar signal is changed over into a unipolar signal (incorporates particular regards). The numerical equation of Shannon energy is given as follows: (4)where is the end product of differentiator and is the end product of Shannon energy value. The function, multiplication operation and square operation are used to perceive the better R-peak in our proposed method. The differentiated output signal acts as an input of the square operation and the input is p. In this paper, multiplier-2 algorithm is utilised for square operation. Consider . For calculating p, the squaring operation is used. So (4) becomes (5)where p is the output of the square operation. To apply Taylor series expansion for (6)The equations are not affected by the higher order terms in the Taylor series of , so the higher order terms are neglected. The multiplication, subtraction, addition and the division operations are required for calculating . At first using the subtraction block is calculated. Then the is found by using the squaring operation and then is calculated by using the multiplication operation. Then the is calculated by using the subtraction operation. Finally, the Shannon energy envelope is obtained. 3.2.2 Zero phase filtering After the SEE extraction, the output signal is given to the zero phase filtering process. This filtering process is used to smooth out the spikes of the signal and that the signal is like the QRS complex part of the ECG signal. The smooth R peaks can be gotten, depending upon the filter length L. The conceivable more extensive QRS complex is equivalent to the length L. 3.3 Peak detection technique The HT and MA filters are incorporated into the peak finding logic. The ideal position of R-peak is seen by the HT. The issues, for example baseline drift, motion artefacts, and muscular noise, are decreased with HT. In the theoretical side, the HT is formulated obviously yet in practical side, it is yet to be illustrated. The greater part of the technique has been utilised as a part of R peak detection procedure. Be that as it may, in this paper radix2 algorithm with eight focuses (butterfly structure) single path delay feedback real fast Fourier transform-(single path delay feedback (SDF) RFFT)-based technique for the detection of the analytical signal is used. The theoretical thought is examined and the computation of diagnostic signal utilising RFFT is assessed. 3.3.1 Hilbert transform In theoretical consideration, the HT is used to achieve an analytic signal is written as (7)where s [n] is the HT of the discrete time signal n = 0, 1, …N − 1 and . The property of analytic signal is (8)Then, Fourier transform of can be written as (9)The complex sinusoidal signal can be written as (10)The time-dependent amplitude may be reconstructed from (11) 3.3.2 Computation of the analytical signal using RFFT In the computation of analytic signal, the RFFT processor is utilised. The discrete Fourier transform (DFT) of the signal , is defined as (12)where the inverse transform is defined as (13)where The computation of the analytical signal using the RFFT-based method is based on the property (2) of the Fourier spectrum of the analytic signal. If represents the DFT coefficients of the real signal and represents the negative frequency band values. So the IRFFT is used to neglect the negative frequency components. Applying the IRFFT, now the analytical signal is written as (14)The input signal is passed to the first block of the RFFT to acquire forward RFFT signal for the computation of HT. So as to disregard the non-existent values, the forward RFFT signal is handled. For acquiring IRFFT the output of the RFFT is given to the next block. At long last, the output of the HT signal is produced. Both DFTs and RFFT can be used to create the real values if the signals are discrete. In HT, the logical signal fuses both real and complex esteems. For decreasing the complex operational, the RFFT and IRFFT architectures are utilised. 3.3.3 Real valued fast Fourier transform For getting a real-valued signal the RFFT design was made. The proposed design gave radix2 algorithm 8 points are utilised to stifle the superfluous operation. Four butterfly units, our shuffling units, and two complex multiplier units are the four stages in the proposed RFFT design. The RFFT design has less computation cycle and lower equipment costs. The computational cycle normally gets diminished by extending the processing element. The block schematic for RFFT is delineated in Fig. 1 d. The first stage has one butterfly unit. There are two inputs are given to the first stage. The expansion and subtraction are utilised as a part of typical sort if the inputs are real values. The multiplexer allows the input to the output with no calculation if the inputs are imaginary values. At that point the output of the first stage is given to the second stage. The shuffling unit, butterfly unit, and the complex multiplier are incorporated into the second stage. Swapping is required at each stage of the RFFT unit. So the shuffling unit is utilised as a part of the second stage for the swapping operation. In the second stage, the output of the first stage is swapped. At that point, the second stage swapped output goes about as an input of the third stage. Butterfly unit and two shuffling units are incorporated into the third stage. The output of the third stage is given to the last stage. In the last stage, the twiddle factor coefficients are ignored. At last, the single real-valued output is gotten. 3.3.4 Moving average filter The output signal is passed through the MA filter after the HT. For smoothing the R peak signal and suppresses the interference of those signals the MA filter is commonly used. The mathematical equation of the MA filter is (15)where is an output of the Shannon energy method and is the output of MA filter. 3.4 R peak detection After the peak finding logic, the real R peaks are recognised in the R peak recognition strategy. Now, there are two subtraction blocks and three samples are expected for the detection of real R peaks. They are u, v and w (assume w is the output of the MA filter). At that point we find q = u − v and r = w − v. On the off chance that q and r values are negative, we plot the R peak location is 1 generally 0. Identified R peak and the actual R peak positions are not equivalent. At that time, the quantities of previous samples are utilised to detect the real R peak. From the distance between the highest amplitude, the actual R peaks are detected. Presently, the input signal given to the current position of the R peak is automatically incremented by 1. At that point from the current location, the separation between the highest amplitude is suppressed. Finally, the exact position of R-peak is recognised. 4 Experimental results 4.1 Performance evaluation The MIT-BIH arrhythmia databases are used for the experimental evaluation of the proposed method [25, 34]. The MIT-BIH arrhythmia database fuses large P and T WAVE, unexpectedly alters in the QRS complex, wide QRS complex, little QRS complex, false R peaks, negative R peaks. Remembering the ultimate objective to diminish these sorts of issues, the genuine R-peak detection is basic. The proposed technique is executed by using Xilinx (ISE 14.5) to get the right location of the real R peaks. The sensitivity (Se), PPV, detection error rate (DER) and accuracy (Acc) are indispensable to figure the execution of the real R peaks. These qualities are handled by using true positive (TP), false negative (FN) and false positive (FP) parameters. From the finally perceived real R peaks, these parameters are enrolled. The Se, PPV, DER and Acc are prepared by using taking after conditions (16) (17) (18) (19)Here, TP is the quantity of accurately recognised R peaks, FN is the quantity of missed R peaks and FP is the quantity of noise peaks distinguished as R peaks [32]. The performance of our proposed R peak detection rates is revealed in Table 1. The proposed peak detection method generates 35,991 TP peaks, 35 FP peaks, 18 FN peaks. The sensitivity is 99.95%, PPV is 99.90% and the accuracy is 99.86%. Table 1. Results of R-peak detection DATA NO TP FN FP Se, % PPV DER, % ACC 100 994 1 0 99.89950 100 0.100603 99.89950 101 533 0 0 100 100 0 100 102 925 1 1 99.89200 99.89200 0.216216 99.78425 103 595 0 0 100 100 0 100 104 944 0 2 100 99.78858 0.211864 99.78858 105 737 1 0 99.86449 100 0.135685 99.86449 106 971 0 1 100 99.89712 0.102986 99.89712 107 610 0 0 100 100 0 100 108 993 1 1 99.89940 99.89940 0.201409 99.79899 109 723 0 1 100 99.86188 0.138313 99.86188 111 607 0 2 100 99.67159 0.329489 99.67159 112 725 0 1 100 99.86225 0.137931 99.86225 113 513 0 0 100 100 0 100 114 537 0 2 100 99.62894 0.372439 99.62894 115 558 1 1 99.82111 99.82111 0.358423 99.64286 116 687 0 0 100 100 0 100 117 439 0 0 100 100 0 100 118 650 0 0 100 100 0 100 119 567 0 1 100 99.82394 0.176367 99.82394 121 983 1 0 99.89837 100 0.101729 99.89834 122 707 0 1 100 99.85876 0.141442 99.85876 123 997 1 1 99.89979 99.89979 0.200601 99.79980 124 991 0 2 100 99.79859 0.201816 99.79859 200 746 0 0 100 100 0 100 201 861 1 0 99.88399 100 0.116144 99.88399 202 610 1 0 99.83633 100 0.163934 99.83633 203 853 0 1 100 99.88290 0.117233 99.88290 205 759 0 0 100 100 0 100 207 531 0 0 100 100 0 100 208 843 1 0 99.88152 100 0.118624 99.88152 209 858 1 0 99.88359 100 0.116550 99.88359 210 752 0 0 100 100 0 100 212 785 0 2 100 99.74587 0.254777 99.74587 213 929 2 0 99.78518 100 0.215285 99.78518 214 755 0 0 100 100 0 100 215 961 0 1 100 99.89605 0.104058 99.89605 217 631 0 1 100 99.84177 0.158479 99.84177 219 615 0 0 100 100 0 100 220 585 0 0 100 100 0 100 221 693 0 2 100 99.71223 0.288600 99.71223 222 709 0 4 100 99.43899 0.564175 99.43899 223 745 1 1 99.86595 99.86595 0.268456 99.73226 228 587 2 2 99.66044 99.66044 0.681431 99.72318 230 645 1 0 99.84520 100 0.155039 99.84520 231 949 0 1 100 99.89474 0.105374 99.89474 232 939 0 2 100 99.78746 0.212992 99.78746 233 877 1 1 99.88610 99.88610 0.228050 99.77247 234 787 0 0 100 100 0 100 total 35,991 18 35 99.95215 99.90242 0.145761 99.863 4.2 Performance comparison Table 2 shows the performance of the proposed method is compared with the other FPGA-based method. The total beats: FP and FN beat rates are included in proposed method. Table 2 reveals that the proposed method accuracy is higher than the other existing methods. In our proposed method, the accuracy is 99.86% and the standard deviation of the accuracy is 14.41%. Table 2. Correlation of performance of our proposed method with different techniques Method Total beats (TP + FN) FN (beats) FP (beats) Se, % +P, % DER, % ACC, % Standard deviation of Acc, % our proposed method 35,991 18 35 99.95215 99.90242 0.145761 99.863 14.41 heart rate monitoring system [21] 109,474 58 116 99.94705 99.89415 0.158942 99.84131 15 level crossing sampling [23] 109,428 1216 651 98.89 99.4 1.71 98.37 70 PSEE [24] 109,494 93 91 99.92 99.92 0.168 99.832 28.7 WT [25] 104,184 65 112 99.94 99.89 0.170 99.830 31.6 SEHT [20] 109,496 79 140 99.93 99.87 0.200 99.800 29.7 DOM [26] 109,809 58 166 99.95 99.85 0.204 99.796 25 WT [27] 109,428 220 153 99.80 99.86 0.340 99.660 — WT [28] 110,159 322 120 99.89 99.70 0.402 99.599 56.5 PT [29] 109,809 507 277 99.54 99.75 0.712 99.288 20 Table 3 shows the performance of the proposed method and it is compared with the existing FPGA-based method for the accurate R-peak detection. The accuracy of the proposed method is very high when compared with the other FPGA implementation methods like heart rate monitoring system, entropy measure of fuzziness, integer WTs, adaptive lifting scheme, and median-based threshold method. In our proposed methods, the standard deviation of the accuracy is 14.41%. Table 3. Correlation of performance of our proposed method with other FPGA-based method Method Se, % +P, % Acc, % Standard deviation of Acc, % our proposed method 99.95215 99.90242 99.863 14.41 heart rate monitoring system [21] 99.94705 99.89415 99.84131 15 entropy measure of fuzziness [30] 99.74 99.74 99.5 19.2 integer WT [31] 97.06 98.04 95.21 40 adaptive lifting scheme [32] 98.78 99.8 98.68 68 median-based threshold method [22] 99.05 97.67 96.76 75 4.3 System synthesised reports The synthesised performance of the proposed strategy contrasted and other existing strategies have appeared in Table 4. The existing method utilised the vertex 5 family for synthesised the report. We have utilised advanced vertex 5 family (XC5VLX20T) for synthesising the proposed method. In the existing method the availability of slice registers is very high (607,200); however, in our method the availability of slice register is low (12,480). So, the utilisation of the register is high (8%) in the proposed method when compared with the existing method. A comparable thing is going ahead in other parameters. Table 4. Results of synthesised report Device utilisation summary (estimated values) Logic utilisation Our proposed method Heart rate monitoring system Used Available Utilisation Used Available Utilisation number of slice registers 1086 12,480 8% 5728 607,200 0% number of slice LUTs 1408 12,480 11% 88,456 303,600 29% number of fully used LUT-FF pairs 1054 1440 73% 188 93,996 0% number of bonded IOBS 82 172 47% 114 700 16% number of BUFG/BUFGCTRLS 2 32 6% 1 32 3% Fig. 2 a exhibits the reallocation of R peak utilising Xilinx (ISE 14.5) from 100 of the MIT-BIH arrhythmia database. Here, K [15:0] is the input ECG signal and R peak [15:0] is the peak of the ECG signal, R location [15:0] is the location of the ECG signal and the value of R location [15:0] is 665. (b) Demonstrates the position of perceived real R peak of the ECG signal from 100 of the MIT-BIH arrhythmia database. Fig. 2Open in figure viewerPowerPoint R-peak detection using Record 100 (a) Location of R-peak for record 100, (b) Placement of the distinguished R-peak for Record 100 Fig. 3 a exhibits the actual location of R peak utilising Xilinx (ISE 14.5) from 101 of the MIT-BIH arrhythmia database. Here, K [15:0] is the input ECG signal and R peak [15:0] is the peak of the ECG signal, R location [15:0] is the location of the ECG signal and the value of R location [15:0] is 714. Fig. 3 b shows the position of recognised R peak on the ECG from 101 of the MIT-BIH database. It has three actual R-peak, the location of the R peaks are 85, 398, 714. Fig. 3Open in figure viewerPowerPoint R-peak detection using Record 101 (a) Location of R-peak for Record 101, (b) Placement of the distinguished R-peak for Record 101 Fig. 4 a presents the location of the R peak utilising Xilinx (ISE 14.5) from 106 of the MIT-BIH arrhythmia database. Here, K [15:0] is the input ECG signal and R peak [15:0] is the peak of the ECG signal, R location [15:0] is the location of the ECG signal and the VALUE of R location [15:0] is 355. Fig. 3 b shows the position of detecting R peak of the ECG signal from 106 MIT-BIH databases. It has two actual R peaks. The location of the R peaks is 355, 729. Fig. 4Open in figure viewerPowerPoint R-peak detection using Record 106 (a) Location of R-peak for Record 106, (b) Placement of the distinguished R-peak for Record 106 In Fig. 5, the location of the R peak and the situation of detecting R peak on ECG signal from 111 of the MIT-BIH database introduce in Figs. 5 a and b. Here, K [15:0] is the input ECG signal and R peak [15:0] is the peak of the ECG signal, R location [15:0] is the location of the ECG signal and the value of R location [15:0] is 800, in Fig. 5 a. Fig. 5 b shows three actual R peaks, the actual location of the R peaks is 200, 491 and 808. Fig. 5Open in figure viewerPowerPoint R-peak detection using Record 111 (a) Location of R-peak for Record 111, (b) Placement of the distinguished R-peak for Record 111 The detection of R peak location and the situation of detecting R peak of the ECG signal from 202 of the MIT-BIH databases are appearing in Figs. 6 a and b. Here, K [15:0] is the input ECG signal and R peak [15:0] is the peak of the ECG signal, R location [15:0] is the location of the ECG signal and the value of R location [15:0] is 745, in Fig. 6 a. Fig. 6 b shows two actual R peaks, the location of the true R peaks is 351 and 745. Fig. 6Open in figure viewerPowerPoint R-peak detection using Record 202 (a) Location of R-peak for Record 202, (b) Placement of the distinguished