17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IET Signal Processing

Research Article

Automated QRS complex detection using ISSN 1751-9675

Received on 15th November 2017

MFO-based DFOD
Revised 13th June 2018
Accepted on 16th July 2018
E-First on 4th October 2018
doi: 10.1049/iet-spr.2018.5230
www.ietdl.org

Chandan Nayak1 , Suman Kumar Saha1, Rajib Kar2, Durbadal Mandal2

1Department of Electronics and Telecommunication Engineering, NIT Raipur, Raipur, Chhattisgarh 492010, India
2Department of Electronics and Communication Engineering, NIT Durgapur, Durgapur, West Bengal 713209, India
E-mail: chandanayak234@gmail.com

Abstract: This study proposes a heuristic approach for designing highly efficient, infinite impulse response (IIR) type Digital
First-Order Differentiator (DFOD) by employing a nature-inspired evolutionary algorithm called Moth-Flame Optimisation (MFO)
for the detection of the QRS complexes in the electrocardiogram (ECG) signal. The designed DFOD is used in the pre-
processing stage of the proposed QRS complex detector, to generate feature signals corresponding to each R-peak by
efficiently differentiating the ECG signal. The generated feature signal is employed to detect the precise instants of the R-peaks
by using a Hilbert transform-based R-peak detection logic. The performance efficiency of the proposed QRS complex detector is
evaluated by using all the first channel records of the MIT/BIH arrhythmia database (MBDB), regarding the standard
performance evaluation metrics. The proposed approach has resulted in Sensitivity (Se) of 99.93%, Positive Predictivity (PP) of
99.92%, Detection Error Rate (DER) of 0.15%, and QRS Detection Rate (QDR) of 99.92%. Performance comparison with the
recent works justifies the superiority of the proposed approach.

1 Introduction mother wavelet, there is no universal rule, and the performance of

An ECG is a composite recording of all the action potentials
generated by the nodes and the cells of the myocardium and
represents the electrical activity related to the function of the
human heart. The conventional heart rhythm called normal sinus
rhythm comprises a sequence of P, Q, R, S, and T wave. The
combination of Q, R, and S waves together forms the QRS
complex and carries the precious clinical information regarding the
cardiac health of the human heart. Any deviation in shape and the
time of occurrence of the QRS complex from its normal behaviour
reflects a particular cardiac disease and needs an immediate
intervention by the physician to eliminate any chance of fatality.
Also, the reliability of the QRS-based applications like heart rate
variability (HRV) analysis, ECG beat classification and
compression, foetal heart rate monitoring (FHRM), and biometric,
depends upon the accurate estimation of the time of occurrence of
the R-peaks (the peak of the QRS complex) in the ECG signal.
However, usually the ECG signals are affected by the power line
interference (PLI) [1], baseline wanders, and muscle noise [2] etc.
The presence of this unwanted noise and interference along with
large amplitude P/T wave, patient-specific QRS complex, time-
varying ECG morphology, and patient movement artefacts due to
poor electrode contact unreasonably complicate the R-peak
detection process [3].
QRS complex detection is a two-stage process comprising the
ECG pre-processing stage and the heartbeat detection stage [4].
The job of the pre-processing step is to accentuate the QRS
complex by reducing the noise and artefacts. The role of the
heartbeat detection stage is to locate the occurrence of the R-peaks
in the ECG signal.
Many R-peak detection approaches have been reported in the
literature [4–31]. Out of this, several wavelet transform (WT)-
based R-peak detection methods are published [5–9]. A wavelet
filter bank-based R-peak detector for low power cardiac pacemaker
is reported in [5]. A quadratic spline WT-based arrhythmia monitor
is proposed in [6]. A discrete WT (Daubechies db10 wavelet)-
based R-peak detector is reported in [7]. A multi-resolution WT
(Daubechies db6 wavelet)-based R-peak detector is reported in [8].
A Haar WT-based R-peak detector is reported in [9]. Though WT-
based peak finding logics are efficient, but, for the selection of the
the WT-based algorithm is entirely application dependent [7].
Several other efficient R-peak detection techniques are also
reported in various literature. These include adaptive linear
predictor [10–11], empirical mode decomposition [12], peak
enhancement and adaptive thresholding [13], filter banks [14],
quadratic filter [15], mathematical morphology [16–17], neural
network [18], integrate and fire sampling [19], level crossing
sampling [20], sparse derivative [21], dynamic plosion index (DPI)
[22], max-min difference algorithm [23], weighted total variation
denoising [24], MaMeMi (highpass) filter [25], and derivative-
based methods [26–31].
The derivative-based algorithms use the slope information to
detect the occurrence of the R-peaks [26]. The derivative-based R-
peak detection techniques are easy to implement and are efficient
[26–27]. A real-time derivative-based 24 h Holter monitoring
system is reported in [26]. For the analysis of enormous data sets,
and for real-time study, the first derivative-based R-peak detection
method is frequently used [27]. Also, the algorithm training,
patient-related parameter changes and the manual segmentation of
data are not essential in case of the derivative-based methods [27].
The derivative-based R-peak detection techniques are employed by
many researchers like Manikandan et al. [28], Hamilton and
Tompkins [29], and Sabherwal et al. [30]. A five-point
differentiator, whose frequency response is linear from (0–30) Hz,
with a delay of two samples, is reported in [26]. In [27], the
derivative is implemented by using a centre differentiation with a
unit sample delay (method-1), five-point differentiation
(method-2), both centre and forward differentiation (method-3). A
forward differentiation is used in [28] with unit sample delay. In
[29], a five-point differentiation is implemented. In [30], both first-
and second-order derivatives are performed by using three-point
differentiation operation. In [31], the differentiator is designed by
using the fractional operators. In all these cases, the differentiators
are developed by using sub-optimal methods. Though the
differentiator-based R-peak detection techniques are frequently
used in real time R-peak detection applications, the author is
insisted on designing highly accurate, computationally efficient
first-order digital differentiator by using an evolutionary
optimisation technique called Moth-flame optimisation (MFO) for
the R-peak detection application.

IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184 1172

© The Institution of Engineering and Technology 2018

In the proposed R-peak detector, the differentiator stage is
designed by using an efficient metaheuristic optimisation algorithm
called MFO algorithm. The optimally designed infinite impulse
response (IIR) type digital first-order differentiator (DFOD) for the
R-peak detection application is the novelty of the proposed
research work. Finite impulse response (FIR) filter-based design is
more suitable than IIR filter-based design due to the linear phase
characteristics, stability, and ease to model [32]. However, for the
design of DFOD the IIR type filter structure is suitable over FIR
counterpart, as linear phase response is not essential for many
applications [33], and the IIR filters are more computationally
efficient than the FIR filter counterpart [34]. The magnitude
response of the proposed DFOD approximates the ideal
differentiator over the entire Nyquist frequency range. The
designed DFOD is both stable and minimum phase, hence provides
minimum and constant group delay. Due to the advantages
mentioned above, when the ECG signal is differentiated by the
proposed DFOD, maximum slope information is obtained. This
slope information is used as the feature signal for the subsequent
signal processing stages for reliable R-peak detection in the ECG
signal.
The noteworthy contributions made in this paper are described
below.
(a) This paper justifies the applicability of the MFO algorithm, for
the design of the wideband, stable, and accurate IIR type digital
first-order differentiator. The magnitude response of the proposed
differentiator almost replicates to that of the ideal integer order
differentiator counterpart, and affirms the superiority of the MFO-
based DFOD design problem.
(b) The proposed DFOD, an integral part of the pre-processing
stage, is used to generate the feature signal corresponding to each
R-peak by efficiently differentiating the ECG signal. The generated
feature signal is employed to detect the exact instants of the R-
peaks by using a Hilbert transform (HT)-based R-peak finding
logic.
(c) The performance of the proposed R-peak detector is evaluated
against all the first channel records of the MBDB by using the
standard R-peak detection statistical metrics such as Detection
Error Rate (DER), QRS Detection Rate (QDR), Positive
Predictivity (PP), and Sensitivity (Se).
(d) The superiority of the proposed R-peak detector over all the
recently published R-peak detection techniques is also justified in
terms of all the above mentioned standard R-peak detection
metrics.
The rest part of the proposed research work is organised as
follows: after the introduction, Section 2 describes the proposed
MFO-based DFOD, an integral part of the pre-processor. Section 3
presents the HT-based R-peak detection technique. Section 4 shows
the simulation results and discussions. Section 5 concludes the
proposed work.
2 Proposed DFOD
The proposed DFOD is analogous to the ideal integer order
differentiator (IOD) in the discrete time domain.
2.1 Problem formulation
The frequency response of the ideal integer order differentiator is
given by (1).
Dideal( jω) = jω (1)
where ω ∈ [0, 1] is the normalised frequency and j = −1.
The Transfer Function (TF) of an N th order IIR-type DFOD is
given by (2).
N
∑i = 0 aiZ − i
HDFOD(Z) = N (2)
∑i = 0 biZ − i
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
where ai denotes the numerator coefficient (Num. Coeff.) and bi
signifies the denominator coefficients (Denum. Coeff.) of the IIR-
type DFOD, for i = 0, 1, 2, 3, …, N.
Hence, the frequency response of the N th order IIR-type DFOD
is given by (3).
a0 + a1e− j1ω + a2e− j2ω + ⋯ + aN e− jNω
HDFOD(ω) = (3)
b0 + b1e− j1ω + b2e− j2ω + ⋯ + bN e− jNω
The proposed MFO-based DFOD design scheme is shown in
Fig. 1a. Here f(n), y(n), and d(n) denote the input to the
differentiator, the output of the ideal differentiator, and output of
the proposed DFOD, respectively.
The primary target is to design the optimal coefficient values of
HDFOD(ω) by minimising the cost function as given in (4), in such a
way that the input–output relationships between the proposed
DFOD and the ideal IOD exactly match with each other. A total
number of 512 frequency sample points is used for the calculation
of JDFOD.
JDFOD = ∑ | | Dideal(ω) | − | HDFOD(ω) | | 2
(4)
The performance of the proposed DFOD is demonstrated by using
the maximum absolute magnitude error (MAME) metric as given
by (5) [35].
MAMEDFOD = max ( | | Dideal(ω) | − | HDFOD(ω) | | ) (5)
Small group delay value is essential for the practical signal
processing applications. The group delay of the DFOD is given by
(6).
dθDFOD(ω)
τDFOD(ω) = − (6)
dω
where θDFOD(ω) = ∠HDFOD(ω) is the phase response of the
HDFOD(ω).
The phase response of the proposed DFOD is demonstrated by
evaluating the average group delay, given by (7).
π
1
τDFOD =
π ∫
0
τDFOD(ω)d(ω) (7)
2.2 Moth-flame optimisation algorithm
The design of digital differentiator is a very old research area. An
IIR-type digital differentiator was designed by using Simpson's
rule for integration [36]. However, in the recent trend, highly
efficient digital integrators and differentiators are developed by
using evolutionary optimisation techniques [37–40]. Since lower
order integrators and differentiators are suitable for the real-time
applications [40], in this paper, an IIR type, first-order digital
differentiator is designed by using MFO algorithm, for the
detection of the QRS complexes in the ECG signal.
MFO algorithm is inspired by the natural navigation mechanism
of moths (search agents), called transverse orientation [41]. During
the night time, the moth traverses a very long distance in the
straight line path, by maintaining a fixed angle with respect to the
moon. However, when the moth navigates for artificial light
(flame) by retaining the same transverse orientation, due to the
short distance between moth and light source, moths traverse in a
shrinking spiral path towards the artificial light. This spiral
movement of moth concerning the artificial lights is modelled to
perform optimisation.
The initial positions of the moths are distributed randomly in
the search space and the random position of the ith moth in the jth
dimension is stored in a (n × d) matrix called M. Here, n represents
the total number of moths and d represents the total number of
design variables.
The cost function of each moth is evaluated, and its
corresponding fitness values are stored in a (n × 1) matrix called
1173

Fig. 1 (a) Block diagram of the proposed DFOD. (b) The proposed R-peak detection algorithm
OM. Best-so-far positions of the moths called flames are stored in
a (n × d) matrix called F. The fitness values of the flames are stored
in a (n × 1) matrix called OF.
The next position of the moths is evaluated by using (8).
M(i, j) = S(Mi, F j) = Di ⋅ ebt ⋅ cos(2πt) + F j (8)
where S is the spiral function; Di = | F j − Mi| is the separation
distance between the ith moth (Mi) and the jth flame (F j); b is a
constant denoting the shape of the logarithmic spiral; and t is the
random number t ∈ [r, + 1]. Here, the value of r (adaptive
convergence constant) is varied from −1 to −2 over the number of
iterations and ensures faster convergence and intensification of the
MFO algorithm.
In order to avoid the local optima and to diversify the problem
search space, after each iteration, the flames are sorted according to
their fitness values. The sorted flames are assigned to their
respective moths, i.e. the first moth modifies its position with
respect to the best flame, and the last most adjusts its position with
respect to the worst flame.
To ensure a proper balance between the exploration and the
exploitation phases, the number of flames is varied adaptively over
the number of iterations as given by (9).
P−1
flame_no = round (P − l) × (9)
T
where l represents the current iteration number, T represents the
maximum iteration number, and P represents the maximum
number of flames.
Steps of MFO algorithm for evaluating the optimal set of
coefficients are described below.
Step 1: Randomise the initial position of the moths (search agents)
in the problem search space. For the N th order DFOD design
problem, each candidate solution consists of equal number of Num.
Coeff. (ai) and Denum. Coeff. (bi) for i = 0, 1, 2, 3, …, N, with
dimension D = 2 × (N + 1). Initialise n = 50 and b = 1.
Step 2: Update the flame_no by using (9).
Step 3: Calculate OM = JDFOD(M) by using (4).
1174
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Step 4: If iteration_no = 1, then go to Step 5, otherwise, go to Step
6.
Step 5: Evaluate F = sort(M) and OF = sort(OM).
Step 6:Evaluate F = sort(Mt − 1, Mt)
and OF = sort(Mt − 1, Mt).
Step 7: For each moth (mi) repeat Step 8 to Step 10.
Step 8: For each dimension (d j) of the design variable repeat Step 9
to Step 10.
Step 9: Update r and t.
Step 10: Find Di = | F j − Mi|, and update M(i, j) by using (8).
Step 11: Until the stopping criteria are not satisfied repeat Step 2 to
Step 10.
Step 12: Return the position of the best moth, consisting of the
optimal sets of Num. Coeff. (ai) and Denum. Coeff. (bi) of the
proposed designed DFOD.
3 Proposed R-peak detection scheme
The proposed DFOD-based R-peak detection scheme is shown in
Fig. 1b. It consists of ECG pre-processor design and R-wave
detector design, as discussed in the following sub-sections.
3.1 ECG pre-processor design (stage-1)
The proposed ECG pre-processor design scheme consists of a
bandpass filter (BPF), an optimally designed DFOD, an amplitude
normaliser (AN), and a smooth waveform generator as discussed in
the following sub-sections.
3.1.1 Enhancement of QRS complex by noise reduction and
differentiation: Practical ECG signal is corrupted by several types
of noise and artefacts as shown in Fig. 2a. The presence of this
noise and artefacts hinders the R-peak detection process and
decreases the R-peak detection accuracy. Therefore, to combat
these noise and artefacts, the ECG signals are filtered by using a
third order Chebyshev type-1 bandpass filter (5–22 Hz), so that
these unwanted components are automatically discarded from the
raw ECG signal. To accentuate the QRS complex, the filtered ECG
signal is differentiated by the proposed DFOD. The differentiated
ECG signal is shown in Fig. 2b. The DFOD stage generates feature
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018
 17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 2 Illustration of the performance of the different stages of the pre-processor;
(a) Raw ECG; (b) Generated feature signal by the proposed DFOD; (c) Smooth waveform. (d) Illustration of HT-based R-peak detection technique

signal (slope information) corresponding to each R-peak in the by considering the amplitude of the few previously detected R-
ECG signal. From Fig. 2b it can be observed that the out-of-band peaks. If an R-peak is not detected for a specific duration, search
noise and artefacts are diminished, and suitable feature signals back mechanism is implemented by reducing the present threshold
corresponding to each R-peak are obtained. This feature signal is value. If more than one R-peak is detected within a particular
used as a reference point (marker or feature signal) for the period, based upon their current amplitudes, any of the R-peak is
detection of the R-peaks in the subsequent signal processing stages. accepted.
The differentiated ECG signal (d[n]) is normalised by using (10). There are a wide variety of thresholding techniques [28]. These
thresholding techniques work well for regular heartbeats [28].
d[n] However, the difficulty arises for the irregular beats.
a[n] = (10) The R-peak detection accuracy of the threshold-based R-peak
max ( | d[n] | )
detection logic is dependent on the initial threshold value. To avoid
the complexity associated with thresholding-based R-peak
3.1.2 Smooth waveform generator: The differentiated output is detection logic an HT-based simple R-peak finding logic [7, 28] is
further enhanced by calculating the Shannon energy and the implemented in this paper and is explained in the following sub-
moving sum of the differentiated signal. The Shannon energy is section.
calculated by using (11) [7, 28] to amplify the differentiated ECG
signal nonlinearly, which further assists to accentuate the QRS 3.2.1 Simple peak finding logic based on Hilbert transform
complex compared to other components. Further, to obtain the and zero crossing detection: Let g(t) is a real-time signal, and
smooth waveform moving window integration is implemented. h(t) = (1/πt) is the impulse response of the HT. Then the HT of
g(t) is denoted by g^ (t).
s(n) = − ((a[n])2 ⋅ log(a[n])2) (11)
∞
1 1
For moving window integration, the number of sample points L is g^ (t) = HT[g(t)] = g(t) ∗
πt
=
π ∫−∞
g(τ)
t−τ
dτ (12)
chosen in such a way that it corresponds to the maximum duration
of the QRS complex. For the ECG signal with 360 Hz sampling By using the convolution property of the Fourier transform one can
frequency the length, L = 55 is considered to maintain a proper get
balance between the design complexity and the detection accuracy
[28]. The output of the smooth waveform generator is shown in F g^ (t) = F g(t) × F h(t) (13)
Fig. 2c.
^
G( f ) = G( f )H( f ) (14)
3.2 R-peak detector design (stage-2)
From Figs. 2a and c it can be observed that the maximum positions ^
G( f ) = F g(t) × F
1
= − jsgn( f )G( f ) (15)
of the smoothed waveform (Fig. 2c) are uniquely related to the R- πt
peaks in the raw ECG signal (Fig. 2a). Hence, the R-peaks of the
ECG signal can be detected by determining the maximum positions where
of the corresponding smooth waveforms. The maximum positions
of the smooth waveform can be obtained either by using an +1 f >0
amplitude threshold-based peak finding logic or by using a sgn( f ) = 0 f =0
threshold independent peak finding logic. In case of the threshold-
−1 f <0
based logic, the smoothed waveform is continuously compared
with a fixed or adaptive threshold. One R-peak is declared each
time the smoothed waveform crosses the threshold. The amplitude Hence, the HT of the real-time signal g(t) can be evaluated
of the ECG signal is patient specific and varies with respect to the by using (16).
external conditions [10]. Hence, the adaptive threshold is more
preferred than a fixed threshold. ^ ^ jG( f ) if f < 0
g^ (t)(t) = IFT G( f ) where G( f ) = (16)
In case of adaptive thresholding logic, initially, a default − jG( f ) if f > 0
threshold value is assigned. The next threshold value is calculated

IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184 1175

© The Institution of Engineering and Technology 2018

^
where G( f ) is the Fourier transform of the signal g(t) and IFT
represents the inverse Fourier transform.
The smoothed waveform is applied to the HT block. When an
inflection point appears in the input signal, its corresponding HT
will cross the zero axes, due to the odd symmetry nature of the HT.
This zero crossing positions (ZCPs) will be used as the reference
points for the detection of actual instants of the R-peaks in the
subsequent signal processing stages. The HT-based peak detection
logic is justified by generating a smooth waveform from the
extracted ECG signal from Record-100. The signal is derived for a
time duration of 1 s, which corresponds to the time interval of a
single heartbeat. Fig. 2d justifies the applicability of the HT-based
technique for accurately identifying the maximum position in the
smooth waveform, which approximately corresponds to the R-peak
position of the ECG signal.
3.2.2 Low-frequency drift eliminator: Due to an extensive peak
amplitude variation and the baseline drift, the HT sequence is
strayed from the reference line, called low-frequency drift (LFD).
Due to the LFD in the HT sequence, the probability of missing true
R-peaks increases. This unfavourable situation is evaded by
subtracting the output of the moving average filter from the Hilbert
transform output. The selection of the length of the moving average
filter plays a vital role in realigning the HT signal with the
reference line. For the ECG signal with a sampling frequency of
360 Hz, the length of the moving average filter K is considered as
400 in this paper.
3.2.3 True R-peak detector: Practically there is a minute
deviation between the ZCPs and the true instants of the R-peaks.
Hence, the true R-peak instants are determined by searching the
maximum positions in the raw ECG signal within ±25 samples
around each ZCP [28].
3.3 ECG database
To evaluate the performance of the proposed R-peak detector, a
universally accepted database called MIT/BIH arrhythmia database
(MBDB) [42, 43] is used in this work. The MBDB consists of dual
channel, 48 ECG records, sampled at a rate of 360 Hz, with 11-bit
resolution over ±5 mv range. All the reported QRS detectors are
validated against all the 48 numbers of the first channel (i.e. Lead
II) ECG records of the MBDB except [31] which is validated
against 47 ECG records of MBDB. To have a fair comparison, the
performance of the proposed R-peak detector is also evaluated
against all the 48 numbers of first channel ECG records of the
MBDB, except the duration of ventricular flutter region of
Record-207. This ventricular flutter region is also not evaluated by
several researchers [18, 19, 22, 28, 31] as this period is not
annotated.
3.4 Performance measurement
For the sake of comparison of the proposed R-peak detector with
the recently published R-peak detectors, standard R-peak detection
performance metrics are calculated by using (17)–(20).
TP
Sensitivity, Se = (17)
TP + FN
FP + FN
Detection Error Rate, DER = (18)
Total no of QRS compex
Table 1 Best control parameter values of MFO for DFOD design
Size of population (n) Maximum number of epochs Logarithmic spiral shape constant (b) Adaptive convergence constant (r)
50 400
Table 2 Performance of the proposed MFO-based DFOD
N Algorithm Num. coeff. [a0,a1,…,aN] Denum. coeff. [b0,b1,…,bN]
1 MFO [1.8792–1.7354]
1176
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
TP
Positive Predictivity, PP = (19)
TP + FP
QRS Detection Rate, QDR = min (Se, PP) (20)
where TP (true positive) indicates the detection of the correct
heartbeat, the FP (false positive) indicates the detection of the
incorrect heartbeat, and the FN (false negative) indicates the
detection failure of the true heartbeat.
4 Results and discussions
The proposed R-peak detector is designed in the following
environment-OS: Windows XP; programming language: MATLAB
version 7.10 (R2010a); CPU: 2 GHz Intel(R) Core(TM) i3-6006U
processor; and RAM: 8GB.
4.1 Results of the proposed MFO-based DFOD
The best-chosen control parameter values of the MFO algorithm
for the optimal design of the DFOD are shown in Table 1.
4.1.1 Performance of the proposed MFO-based DFOD: The
magnitude and the phase response performance of the proposed
MFO-based DFOD are examined by evaluating MAME (dB) and
τDFOD (samples), respectively. The optimal sets of Num. Coeff. and
Denum. Coeff., obtained from the MFO algorithm for the design of
the proposed DFOD, and the values of MAME (dB) and τDFOD
(samples) are presented in Table 2. The magnitude response, AME,
and phase response plots of the proposed MFO-based DFOD are
presented in Figs. 3a–c, respectively. From Fig. 3a, it can be
noticed that the magnitude response of the proposed MFO-based
DFOD almost replicates to that of the ideal differentiator. From
Fig. 3b, it can be seen that the MAME (dB) value is −13.12 dB.
The calculated value of the average group delay τDFOD is only 0.02
samples. The average group delay for the real-time signal
processing applications should be small [40]. Also, lower order
IIR-type digital differentiator is suitable for real-time applications
[40]. The proposed differentiator satisfies above two criterions,
along with the magnitude response of the proposed DFOD is
almost identical to the ideal differentiator. Hence, the proposed
approach is suitable for real-time signal processing applications,
like QRS complex detection in the ECG signal.
4.1.2 Average performance analysis of the MFO-based
DFOD with respect to MAME metric: The performance of the
MFO-based DFOD regarding the MAME metric is evaluated by
running the algorithm for 50 independent times, to calculate the
statistical parameters like best, worst, mean, and standard deviation
(SD) values. For the MAME (dB) metric the best, worst, mean, and
SD values are −13.12, −2.87, −12.27 dB, and 1.63, respectively, as
presented in Table 3. The variation of the MAME (dB) metric for
the MFO-based designed DFOD with respect to 50 independent
runs is depicted in Fig. 3d.
4.1.3 Convergence speed analysis of the MFO-based DFOD
design: The convergence plot of the MFO-based DFOD design is
depicted in Fig. 4a. It can be observed that within few numbers of
iterations MFO reaches near the global optima. Faster convergence
is achieved by the MFO algorithm due to the adaptive decrease of
the convergence constant (r) over the number of iterations. The
1000 −1 to −2
MAME, dB τDFOD (samples)
[1.6021–0.3646] −13.12 0.02
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018

Fig. 3 (a) The magnitude response of the proposed MFO-based DFOD. (b) AME response of the proposed MFO-based DFOD. (c) The phase response of the
proposed-MFO-based DFOD. (d) Variation of MAME (dB) with respect to the number of independent trials runs
Table 3 Average performance of the proposed MFO-based DFOD with respect to MAME metric
Error metrics Best
MAME, dB −13.12
best, worst, average, and standard deviation values are estimated
by running the algorithm for 50 independent times, and the outputs
are shown in Table 4.
4.1.4 Execution time analysis of the MFO-based DFOD
design: The computational time also decides the efficiency of the
algorithm. Hence, in order to know the execution time required for
400 function evaluations by the MFO algorithm for the
implementation of the DFOD, the MFO algorithm is run for 50
independent times, to calculate the statistical parameters namely
SD, worst, mean, and best values, and the corresponding results are
shown in Table 4. The mean time taken for the evaluation by the
algorithm is only 45.485 s, for the DFOD type design problem.
4.1.5 Stability analysis of the proposed MFO-based
DFOD: The pole-zero plot of the designed MFO-based DFOD is
depicted in Fig. 4b. It can be observed that both the pole and zero
reside within the unit circle, which guarantees the system to be a
stable and minimum phase.
4.2 Sensitivity analysis
The impact of MFO algorithm due to the change in initial control
parameter values like logarithmic spiral shape constant (b),
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Worst Mean SD
−2.87 −12.27 1.63
adaptive convergence constant (r), and size of population (n) on the
final solution quality of the designed DFOD, in terms of the
MAME metric are examined in the following sub-sections. The
algorithm is run for 50 independent times for each case to calculate
the best, worst, mean, and standard deviation (SD) values of the
MAME metric. While varying a particular control parameter value,
the other control parameter values are kept constant, as per Table 1.
4.2.1 MFO sensitivity analysis due to the change in the
logarithmic spiral shape constant (b): The logarithmic spiral
shape constant, b controls the convergence speed and the direction
of the spiral movement. For a very low value of b, the spiral
becomes a circle, and for a very large value of b the spiral tends
towards a straight half-line. The effect of the MFO algorithm on
the resulted solution quality for different values of b is examined,
and the results are presented in Table 5. It can be observed that the
better MAME (dB) metric value is obtained corresponding to b = 1,
due to a proper balance between the diversification and the
intensification phases.
4.2.2 MFO sensitivity analysis by changing the value of
adaptive convergence constant (r): r is used to accelerate the
convergence speed of the moths towards the flames. Also, it is used
to enhance the intensification phase of the MFO algorithm. Table 6
1177
 17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 4 (a) Convergence characteristics of the proposed MFO-based DFOD. (b) Pole-zero diagram of the proposed MFO-based DFOD

Table 4 Statistical analysis of the convergence speed for the MFO algorithm employed for DFOD design
Number of iterations required for convergence to the optimal solution Computational time (s) for 400 iterations
best worst average SD best worst average SD
112 163 130 2.548 45.269 45.742 45.485 0.249

Table 5 Comparison of MFO-based DFODs for different values b

N Logarithmic spiral shape constant (b) Num. coeff. [a0,a1,…,aN] Denum. coeff. [b0,b1,…,bN] MAME, dB
Best Worst Mean SD
1 0.01 [1.8789–1.5535] [1.5203–0.3057] −10.02 −1.21 −8.65 3.12
1 0.1 [1.8789–1.7368] [1.6001–0.3604] −12.96 −2.79 −11.85 1.98
1 1 [refer Table 1] [refer Table 1] −13.12 −2.87 −12.27 1.63
1 10 [1.8770–1.5548] [1.5176–0.3003] −9.83 −1.15 −8.56 3.65

Table 6 Comparison of MFO-based DFOD for different values of r

N Adaptive convergence constant (r) Num. coeff. [a0,a1,…,aN] Denum. coeff. [b0,b1,…,bN] MAME, dB
Best Worst Mean SD
1 −1 to −2 [refer Table 1] [refer Table 1] −13.12 −2.87 −12.27 1.63
1 −1 to −4 [1.8785–1.7365] [1.6018–0.3638] −13.08 −2.81 −12.01 1.78
1 −1 to −8 [1.8797–1.7345] [1.5990–0.3597] −12.95 −2.68 −11.89 2.0

Table 7 Comparison of MFO-based DFOD for different population size

N Population size Num. coeff. [a0,a1,…,aN] Denum. coeff. [b0,b1,…,bN] MAME, dB
Best Worst Mean SD
1 25 [1.8779–1.7362] [1.5984–0.3529] −12.40 −1.53 −11.22 1.90
1 50 [refer Table 1] [refer Table 1] −13.12 −2.87 −12.27 1.63
1 100 [1.8825–1.7564] [1.6153–0.3712] −13.28 −2.64 −11.92 1.71

presents the performance of the MFO algorithm in terms of
MAME metric for different ranges of r. It can be observed that
when the range of r varies from −1 to −2, over the number of
iterations, better solution quality is obtained due to the better
intensification of the search space.
4.2.3 MFO sensitivity analysis due to the change in
population size (n): The optimal sets of Num. Coeff. and Denum.
Coeff. and the corresponding MAME error metrics resulted from
the design of the MFO-based DFOD with n of 25, 50, and 100
numbers of moths are presented in Table 7. It can be observed that
by considering a lesser number of moths (n = 25), poor solution
quality is obtained. Also, by considering a large number of moths
(n = 100), minute improvement in the solution quality is achieved
with higher design time overhead. Hence, by considering a suitable
trade-off between the design time and the solution quality, 50
number of search agents are considered as the most suitable for the
MFO-based DFOD type design problem.
4.3 QRS detection
An R-peak detection example illustrating the operation of the
proposed DFOD-based R-peak detection technique is depicted in
Figs. 5–10. In each figure, the subplot (a) portrays the raw ECG
signal. The output of the proposed DFOD is portrayed in subplot
(b). It can be observed that at this stage a suitable feature signal is
generated corresponding to each R-peak in the differentiated ECG
signal. Also, it can be observed that across the output of the DFOD,
the effects of the P/T-waves and the baseline shift are eliminated.
The true instants of the R-peaks are detected by using this suitable
feature signal in subsequent signal processing stages. The subplot
(c) represents the output of the smooth waveform (portrayed in
blue), the HT output (portrayed in green), and the zero crossing

1178 IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184

© The Institution of Engineering and Technology 2018

Fig. 5 R-peak detection of ECG signal (Record-112 of the MBDB)
(a) ECG signal, (b) DFOD output, (c) Smooth waveform (blue), HT output (green),
ZCPs (red dots) (d) Detected R-peaks (black upward arrowhead), annotations (red
vertical lines)
Fig. 6 R-peak detection of ECG signal (Record-105 of the MBDB)
(a) ECG signal, (b) DFOD output, (c) Smooth waveform (blue), HT output (green),
ZCPs (red dots) (d) Detected R-peaks (black upward arrowhead), annotations (red
vertical lines)
points (portrayed in red dots). All the zero crossing points represent
the local maxima of the smooth waveforms. The subplot (d)
represents the instants of the detected R-peaks (portrayed by black
upward arrowhead), and the corresponding annotations (portrayed
in red vertical lines). Fig. 5 depicts the performance of the
proposed QRS detector for a normal ECG rhythm in the presence
of small noise. Figs. 6–10 depict the performance of the proposed
QRS detector for noisy ECG segments which are commonly found
in ECG records 105, 108, 203, 207, and 208.
The overall R-peak detection performance against the MBDB is
presented in Table 8. The performance of the proposed DFOD-
based R-peak detector is evaluated by using a total number of
109,494 heartbeats which yields 109,414 TP beats, 85 FP beats,
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 7 R-peak detection of ECG signal (Record-108 of the MBDB)
(a) ECG signal, (b) DFOD output, (c) Smooth waveform (blue), HT output (green),
ZCPs (red dots) (d) Detected R-peaks (black upward arrowhead), annotations (red
vertical lines)
Fig. 8 R-peak detection of ECG signal (Record-203 of the MBDB)
(a) ECG signal, (b) DFOD output, (c) Smooth waveform (blue), HT output (green),
ZCPs (red dots) (d) Detected R-peaks (black upward arrowhead), annotations (red
vertical lines)
and 80 FN beats. The calculated values of Se (%), PP (%), DER
(%), and QDR (%) are 99.93%, 99.92%, 0.1507%, and 99.92%,
respectively. Also, the effectiveness of the proposed DFOD-based
R-peak detector is further justified in terms of the visual
inspections of different noisy and critical ECG morphology,
extracted from different ECG records, and are portrayed in Figs. 11
and 12.
Figs. 11a–c substantiate the applicability of the proposed
DFOD-based R-peak detector, to determine the true instants of the
R-peaks in the noise and artefacts contaminated ECG records such
1179

Fig. 9 R-peak detection of ECG signal (Record-207 of the MBDB)
(a) ECG signal, (b) DFOD output, (c) Smooth waveform (blue), HT output (green),
ZCPs (red dots) (d) Detected R-peaks (black upward arrowhead), annotations (red
vertical lines)
Fig. 10 R-peak detection of ECG signal (Record-208 of the MBDB)
(a) ECG signal, (b) DFOD output, (c) Smooth waveform (blue), HT output (green),
ZCPs (red dots) (d) Detected R-peaks (black upward arrowhead), annotations (red
vertical lines)
as 104, 105, 108, 116, 203, 210, and 228. Fig. 11a presents an ECG
segment extracted from Record-105, which is affected by severe
baseline drift. Fig. 11b presents an ECG signal extracted from
Record-108. It consists of high-grade noise and baseline drift.
Fig. 11c depicts an ECG segment extracted from MBDB
Record-203. It is corrupted by high-frequency muscle noise.
Figs. 12a–d justify the efficacy of the proposed DFOD-based
R-peak detector for faithful detection of the true instants of the R-
peaks, for different critical ECG morphologies. Fig. 12a presents a
segment of the ECG extracted from Record-208. It consists of
fusion of ventricular and normal beat, premature ventricular
contraction (PVC) beats, normal beats, and R-on-T PVC beat. It
1180
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
can be observed that in this extracted ECG segment every normal
heartbeat follows the PVC beats. Fig. 12b illustrates an ECG
segment with high amplitude T-waves, which have similar
characteristics to that of the QRS complex and are very often
misinterpreted as QRS complex. Fig. 12c shows a segment of ECG
extracted from MBDB Record-217. It consists of fusion of paced
and normal beat, normal beats, PVC beats, and paced beat. An
abrupt baseline drift just after the time instant 6 s in Fig. 12c can be
noticed. In this short period, the amplitude of R-peak is very small.
Fig. 12d shows a segment of ECG extracted from MBDB
Record-200. It represents an irregular ECG segment.
Despite the presence of noise and irregularities in the ECG
signal depicted in Figs. 11 and 12, all the heart beats are detected
successfully by the proposed QRS complex detector.
To justify the efficacy of the proposed MFO-based DFOD for
identifying the true time instants of the R-peaks in the ECG signal,
the proposed differentiator is replaced by the reported non-optimal
method-based [26, 28] and optimal method-based [33]
differentiators in the proposed R-peak detection model as shown in
Fig. 1b. In [26, 28], the differentiators are designed by employing
difference equations, whereas, in [33], the differentiator is designed
by employing an evolutionary optimisation algorithm called
Harmony Search (HS). Table 9 presents the differentiators'
performance along with the corresponding QRS detection results
obtained by incorporating the proposed and the reported
differentiators of [26, 28, 33] in the proposed QRS detection
model. In Table 9, it can be observed that the lowest MAME value
of −13.12 dB and τDFOD value of 0.02 samples are achieved by the
proposed MFO-based DFOD. In terms of QRS detection
performance by incorporating the differentiators of [26, 28, 33] in
the proposed R-peak detection model produced 260 (= 143 + 117),
208 (= 136 + 72), and 179 (= 92 + 87) false beat detections,
respectively. Whereas, the proposed approach yields the lowest
false beat detection value of 165 (= 85 + 80) beats, and thereby
reduces the false beat detection by 95 (= 260–165), 43 (= 208–
165), and 14 (= 179–165) beats, respectively, while incorporating
the differentiators of [26, 28, 33] in the proposed R-peak detection
model. Due to this a sensible improvement in QRS detection
performance with over 37% (= (260–165)/260), 21% (= (208–165)/
208), and 8% = ((179–165)/179) have been achieved in comparison
to the incorporation of reported differentiators of [26, 28, 33].
4.4 Performance comparison
The performance comparison between the proposed and the
reported approaches is presented in Table 10. In Table 10, the
values of all the performance indices are taken from the
corresponding reported literature. The comparative results confirm
the superiority of the proposed R-peak detection technique
compared to all the recently published literature, in terms of the
standard statistical parameters like Se, PP, DER, and QDR. The
ECG records of the MBDB are severely affected by noise and
artefacts. Also, the morphology of these signals is very much
complicated. Due to which, it is a very tough job for simultaneous
QRS augmentation and noise elimination. Hence, the researchers
try to design improved R-peak detectors based on the pre-
processed results [28]. But, in case of the proposed R-peak
detection technique, improved detection accuracy is achieved due
to the better pre-processing of the ECG signal by using the DFOD.
The significant advantages of the proposed DFOD are that the
magnitude response of the proposed DFOD almost replicates to
that of the ideal differentiator counterpart. Hence, when the DFOD
differentiates the ECG signals, maximum slope information is
obtained, due to its near ideal characteristics. Most of the R-peak
detection techniques produce higher numbers of FP and FN beats,
in case of the QRS complex with decreased signal slope
information [27]. The slope information is less in case of
attenuated QRS complex and wide duration QRS complex. Though
the proposed DFOD has almost ideal differentiator characteristics,
it is very much capable of detecting the minimal amplitude and
wide duration QRS complexes. Also, the designed DFOD is a
minimum phase system, hence provides a minimum and constant
group delay to every frequency component of the input signal. So,
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018

a fixed relationship exists between the signal maximum points and
the zero crossing points. Also, besides bandpass filtering, the
proposed DFOD further eliminates the effect of P/T waves, and out
of the band noise and artefacts. Due to the above-mentioned
advantages of the proposed DFOD-based pre-processor, the true
instants of the R-peaks are detected most effectively in the
subsequent signal processing stages. Rakshit et al. in [7] have
reported a total number of 183 (= 99 + 84) false beats. This is the
lowest reported false beat detection value in the QRS detection
literature. Whereas, the proposed approach yields a total number of
165 (= 85 + 80) false beats. Hence, it can be inferred that as
compared to [7] the proposed approach reduces the false beat
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
detection by 18 (= 183–165) beats, and thereby improves the QRS
detection performance by 10% (= (183–165)/183). The percentage
of improvements resulted by the proposed approach over the
reported state-of-the-art QRS detection techniques are presented in
Fig. 13a. Also, it should be noted that the proposed method does
not employ any complex search back and blanking technique.
Hence, the detection accuracy of the proposed detector can be
further enhanced by using the sophisticated search back and
blanking procedure, to reduce the total number of FN and FP beat
counts, respectively. Also, the proposed DFOD is computationally
efficient due to the lowest order implementation. The average time
required by the proposed and the reported QRS detection

Table 8 Performance of the proposed DFOD-based R-peak detector over the MBDB
Record Total beats TP beats FP beats FN beats Se, % PP, % DER, % QDR, %
100 2273 2273 0 0 100 100 0 100
101 1865 1864 2 1 99.95 99.89 0.1609 99.89
102 2187 2187 0 0 100 100 0 100
103 2084 2084 0 0 100 100 0 100
104 2229 2226 13 3 99.87 99.42 0.7178 99.42
105 2572 2566 19 6 99.77 99.26 0.9720 99.26
106 2027 2026 2 1 99.95 99.90 0.1480 99.90
107 2137 2137 0 0 100 100 0 100
108 1763 1763 4 0 100 99.77 0.2269 99.77
109 2532 2532 0 0 100 100 0 100
111 2124 2124 0 0 100 100 0 100
112 2539 2538 0 1 99.96 100 0.0394 99.96
113 1795 1795 0 0 100 100 0 100
114 1879 1879 1 0 100 99.95 0.0532 99.95
115 1953 1953 0 0 100 100 0 100
116 2412 2400 5 12 99.50 99.79 0.7048 99.50
117 1535 1535 0 0 100 100 0 100
118 2278 2278 2 0 100 99.91 0.0878 99.91
119 1987 1987 1 0 100 99.95 0.0503 99.95
121 1863 1863 1 0 100 99.95 0.0537 99.95
122 2476 2476 0 0 100 100 0 100
123 1518 1518 0 0 100 100 0 100
124 1619 1619 0 0 100 100 0 100
200 2601 2597 1 4 99.85 99.96 0.1922 99.85
201 1963 1963 0 0 100 100 0 100
202 2136 2134 0 2 99.91 100 0.0936 99.91
203 2980 2966 4 14 99.53 99.87 0.6040 99.53
205 2656 2655 0 1 99.96 100 0.0377 99.96
207 1860 1854 2 6 99.68 99.89 0.4301 99.68
208 2955 2952 0 3 99.90 100 0.1015 99.90
209 3005 3005 1 0 100 99.97 0.0333 99.97
210 2650 2641 0 9 99.66 100 0.3396 99.66
212 2748 2748 1 0 100 99.96 0.0364 99.96
213 3251 3251 0 0 100 100 0 100
214 2262 2262 0 0 100 100 0 100
215 3363 3363 0 0 100 100 0 100
217 2208 2207 0 1 99.95 100 0.0453 99.95
219 2154 2154 0 0 100 100 0 100
220 2048 2048 0 0 100 100 0 100
221 2427 2427 0 0 100 100 0 100
222 2483 2483 1 0 100 99.96 0.0403 99.96
223 2605 2605 0 0 100 100 0 100
228 2053 2043 16 10 99.51 99.22 1.2664 99.22
230 2256 2256 0 0 100 100 0 100
231 1571 1571 0 0 100 100 0 100
232 1780 1776 8 4 99.78 99.55 0.6742 99.55
233 3079 3077 0 2 99.94 100 0.0650 99.94
234 2753 2753 1 0 100 99.96 0.0363 99.96
overall 109,494 109,414 85 80 99.93 99.92 0.1507 99.92

IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184 1181

© The Institution of Engineering and Technology 2018
 17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Fig. 11 R-peak detection using the MBDB (in each figure the detected R-peak is portrayed in black upward arrowhead, and the annotations are represented
in red vertical lines)
(a) Record-105: detection of R-peaks in presence of severe baseline drift, (b) Record-108: detection of R-peaks in presence of high grade noise and baseline drift, (c) Record-203:
detection of QRS complex in presence of severe noise

Fig. 12 R-peak detection using the MBDB (in each figure the detected R-peak is portrayed in black upward arrowhead and the annotation is portrayed in red
vertical lines)
(a) Record-208: detection of premature ventricular contraction (PVC) beats, (b) Record-117: detection of R-peak in the presence of high amplitude T-waves, (c)Record-217:
detection of R-peaks of a very critical ECG morphology, (d) Record-200: detection of irregular heartbeat

algorithms for processing each sample is shown in Table 10. As
illustrated in Table 10 the average execution time required for
performing the proposed method on each 30 min ECG record in the
MBDB is about 2.41 s, which corresponds to around 3.7 µs for
each sample. The execution time required by the proposed
approach is the least compared to all the reported methods except
the algorithm reported in [28]. It is noteworthy to mention that the
proposed QRS detection algorithm is run on a slower processor
than that used in [28].
Altogether the performance comparison between the proposed
and the reported R-peak detection techniques is demonstrated in
terms of three statistical parameters called Se (%), PP(%), and
DER (%), as portrayed in Fig. 13b. It can be observed that the
proposed DFOD-based R-peak detection technique achieves the
best values of Se, PP, and DER. Though the Se (%) value of the

1182 IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184

© The Institution of Engineering and Technology 2018
 17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Table 9 QRS detection performance comparison by employing the proposed MFO and reported differentiators of [26, 28, 33]
in the proposed QRS detection model
Publication Differentiator performance QRS detection performance
MAME, dB τDFOD (Samples) TP FP FN Se, % PP, % DER, % QDR, %
Pan et al. [26] 9.94 2 109,377 143 117 99.89 99.87 0.24 99.87
Manikandan et al. [28] 1.15 0.5 109,422 136 72 99.93 99.88 0.19 99.88
Mahata et al. [33] −11.1 1.01 109,407 92 87 99.92 99.92 0.16 99.92
proposed −13.12 0.02 109,414 85 80 99.93 99.92 0.15 99.92

Table 10 Comparison of performance between the proposed and the recently published R-peak detectors
Publication Total TP FP FN Se, % PP, % DER, % QDR, % Execution time, µs/sample
Min et al. [5] Δ 109,496 109,383 102 113 99.90 99.91 0.19 99.90 N/R
Leong et al. [6] 109,134 N/R 330 753 99.31 99.70 0.99 99.31 N/R
Rakshit et al. [7] N/R 109410 99 84 99.93 99.91 0.16 99.91 N/R
Bouaziz et al. [8] 109,494 109,354 232 140 99.87 99.79 0.33 99.79 N/R
Deepu et al. [10] 109,508 N/R 201 395 99.64 99.81 0.54 99.64 5.2*
Jain et al. [11] Δ 109,494 N/R 184 272 99.75 99.83 0.41 99.75 13
Phukpattaranont et al. [15] 109,483 109,281 210 202 99.82 99.81 0.37 99.81 N/R
Yazdani et al. [16] 109,494 N/R 108 137 99.87 99.90 0.22 99.87 N/R
Zhang et al. [17] 109,510 N/R 204 213 99.81 99.80 0.38 99.80 N/R
Arbateni et al. [18] 109,483 N/R 109 210 99.82 99.91 0.29 99.82 N/R
Nallathambi et al. [19] 109,494 N/R 495 462 99.58 99.55 0.87 99.55 N/R
Pandit et al. [23] 109,809 109,432 369 389 99.65 99.66 0.69 99.65 15.74
Sharma et al. [24] 109,494 109,381 136 113 99.90 99.88 0.22 99.88 4
Pan et al. [26] 116,137 115,860 507 277 99.76 99.56 0.67 99.56 13.56*
Manikandan et al. [28] 109,496 N/R 140 79 99.93 99.86 0.20 99.86 3.45
Sabherwal et al. [30] 109,498 109,368 98 93 99.90 99.90 0.17 99.90 N/R
Benmalek et al. [31] N/R 107,632 153 156 99.86 99.86 0.28 99.86 10
proposed method 109,494 109,414 85 80 99.93 99.92 0.15 99.92 3.7
N/R: Not reported; ‘Δ’: Best value is reported; *: Reported in [11].

Fig. 13 (a) Percentage improvement of the proposed QRS detector over the reported QRS detection techniques in terms of a total number of false beat
detection. (b) Performance comparison in terms of Se, PP, and DER

proposed method and the method reported in [7] are the same, it
should be noted that the proposed method achieves the lowest DER
value and the highest values of PP. Also, the process reported in [7]
used a refractory period of 200 ms whereas, the proposed approach
does not consider any refractory period. Hence, the detection
accuracy can be further enhanced by finding a suitable refractory
period.
5 Conclusion
An optimally designed, highly accurate, and computationally
efficient Digital First-order Differentiator (DFOD) for the R-peak
detection application has been proposed by employing a highly
efficient optimisation algorithm, called Moth-flame optimisation
(MFO). The performance of the proposed DFOD-based R-peak
detector is verified against the MIT/BIH arrhythmia database
(MBDB) by using the standard statistical parameters like
Sensitivity (Se), positive Predictivity (PP), detection error rate
(DER), and QRS detection rate (QDR), which yields Se (%) of
99.93%, PP(%) of 99.92%, DER (%) of 0.1507%, and QDR (%) of
99.92%. The supremacy of the proposed R-peak detector in
comparison with the other recently published techniques is
confirmed from the calculated statistical parameters. Finally, this
paper ensures the MFO-based DFOD is an efficient pre-processing

IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184 1183

© The Institution of Engineering and Technology 2018

tool used for generating feature signal efficiently related to the
QRS complex region in the ECG signal. The optimally designed
digital differentiator for real-time R-peak detection application is
the novelty of the proposed research work.
6 Acknowledgments
This project is financially supported by Science and Engineering
Research Board (SERB), Department of Science and Technology,
Government of India (Grant No: EEQ/2016/000215).
7 References
[1] Meidani, M., Mashoufi, B.: ‘Introducing new algorithms for realizing an FIR
filter with less hardware to eliminate power line interference from the ECG
signal’, IET Signal Process., 2016, 10, (7), pp. 709–716
[2] Yadav, S.K., Sinha, R., Bora, P.K.: ‘Electrocardiogram signal denoising using
non-local wavelet transform domain filtering’, IET Signal Process., 2015, 9,
(1), pp. 88–96
[3] Friesen, G.M., Jannett, T.C., Jadallah, M.A., et al.: ‘A comparison of the noise
sensitivity of nine QRS detection algorithms’, IEEE Trans. Biomed. Eng.,
1990, 37, (1), pp. 85–98
[4] Kohler, B.U., Hennig, C., Orglmeister, R.: ‘The principles of software QRS
detection’, IEEE Eng. Med. Biol. Mag., 2002, 21, (1), pp. 42–57
[5] Min, Y. J., Kim, H. K., Kang, Y.R., et al.: ‘Design of wavelet-based ECG
detector for implantable cardiac pacemakers’, IEEE Trans. Biomed. Circuits
Syst., 2013, 7, (4), pp. 426–436
[6] Leong, C.I., Mak, P.I., Lam, C.P., et al.: ‘A 0.83 - µW QRS detection
processor using quadratic spline wavelet transform for wireless ECG
acquisition in 0.35 - µm CMOS’, IEEE Trans. Biomed. Circuits Syst., 2012, 6,
(6), pp. 586–595
[7] Rakshit, M., Das, S.: ‘An efficient wavelet-based automated R-peaks
detection method using Hilbert transform’, Biocybern. Biomed. Eng., 2017,
37, (3), pp. 566–577
[8] Bouaziz, F., Boutana, D., Benidir, M.: ‘Multiresolution wavelet-based QRS
complex detection algorithm suited to several abnormal morphologies’, IET
Signal Process., 2014, 8, (7), pp. 774–782
[9] Zidelmal, Z., Amirou, A., Adnane, M., et al.: ‘QRS detection based on
wavelet coefficients’, Comput. Methods Programs Biomed., 2012, 107, (3),
pp. 490–496
[10] Deepu, C.J., Lian, Y.: ‘A joint QRS detection and data compression scheme
for wearable sensors’, IEEE Trans. Biomed. Eng., 2015, 62, (1), pp. 165–175
[11] Jain, S., Kumar, A., Bajaj, V.: ‘Techniques for QRS complex detection using
particle swarm optimization’, IET Sci. Meas. Technol., 2016, 10, (6), pp. 626–
636
[12] Slimane, Z.E.H., Nait-Ali, A.: ‘QRS complex detection using empirical mode
decomposition’, Digit. Signal Process., 2010, 20, pp. 1221–1228
[13] Dohare, A.K., Kumar, V., Kumar, R.: ‘An efficient new method for the
detection of QRS in electrocardiogram’, Comput. Electr. Eng., 2014, 40, (5),
pp. 1717–1730
[14] Afonso, V.X., Tompkins, W.J., Nguyen, T.Q., et al.: ‘ECG beat detection
using filter banks’, IEEE Trans. Biomed. Eng., 1999, 46, (2), pp. 192–201
[15] Phukpattaranont, P.: ‘QRS detection algorithm based on the quadratic filter’,
Expert Syst. Appl., 2015, 42, (11), pp. 4867–4877
[16] Yazdani, S., Vesin, J.M.: ‘Extraction of QRS fiducial points from the ECG
using adaptive mathematical morphology’, Digit. Signal Process., 2016, 56,
pp. 100–109
[17] Zhang, F., Lian, Y.: ‘QRS detection based on multiscale mathematical
morphology for wearable ECG devices in body area networks’, IEEE Trans.
Biomed. Circuits Syst., 2009, 3, (4), pp. 220–228
[18] Arbateni, K., Bennia, A.: ‘Sigmoidal radial basis function ANN for QRS
complex detection’, Neurocomputing, 2014, 145, (5), pp. 438–450
[19] Nallathambi, G., Príncipe, J.C.: ‘Integrate and fire pulse train automaton for
QRS detection’, IEEE Trans. Biomed. Eng., 2014, 61, (2), pp. 317–326
1184
17519683, 2018, 9, Downloaded from https://ietresearch.onlinelibrary.wiley.com/doi/10.1049/iet-spr.2018.5230 by National Medical Library The Director, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[20] Ravanshad, N., Dehsorkh, H.R., Lotfi, R., et al.: ‘A level-crossing based
QRS-detection algorithm for wearable ECG sensors’, IEEE. J. Biomed.
Health. Inform., 2014, 18, (1), pp. 183–192
[21] Ning, X., Selesnick, I.W.: ‘ECG enhancement and QRS detection based on
sparse derivative’, Biomed. Signal Proc. Control, 2013, 8, pp. 713–723
[22] Ramakrishnan, A.G., Prathosh, A.P., Ananthapadmanabha, T.V.: ‘Threshold-
independent QRS detection using the dynamic plosion index’, IEEE Signal
Process. Lett., 2014, 21, (5), pp. 554–558
[23] Pandit, D., Zhang, L., Liu, C., et al.: ‘A lightweight QRS detector for single
lead ECG signals using a max-min difference algorithm’, Comput. Methods
Programs Biomed., 2017, 144, pp. 61–75
[24] Sharma, T., Sharma, K.K.: ‘QRS complex detection in ECG signals using
locally adaptive weighted total variation denoising’, Comput. Biol. Med.,
2017, 87, pp. 187–199
[25] Castells-Rufas, D., Carrabina, J.: ‘Simple and real-time QRS detector with the
MaMeMi filter’, Biomed. Signal Proc. Control, 2015, 21, pp. 137–145
[26] Pan, J., Tompkins, W.J.: ‘A real-time QRS detection algorithm’, IEEE Trans.
Biomed. Eng., 1985, 32, (3), pp. 230–236
[27] Arzeno, N.M., Deng, Z.D., Poon, C.S.: ‘Analysis of first-derivative based
QRS detection algorithms’, IEEE Trans. Biomed. Eng., 2008, 55, (2), pp.
478–484
[28] Manikandan, M.S., Soman, K.P.: ‘A novel method for detecting R-peaks in
electrocardiogram (ECG) signal’, Biomed. Signal. Process. Control., 2012, 7,
(2), pp. 118–128
[29] Hamilton, P.S., Tompkins, W.J.: ‘Quantitative investigation of QRS detection
rules using the MIT/BIH arrhythmia database’, IEEE Trans. Biomed. Eng.,
1986, 33, (12), pp. 1157–1164
[30] Sabherwal, P., Agrawal, M., Singh, L.: ‘Automatic detection of the R peaks in
single-lead ECG signal’, Circuits Syst. Signal Process., 2017, 36, (11), pp.
4637–4652. doi: 10.1007/s00034-017-0537-2
[31] Benmalek, M., Charef, A.: ‘Digital fractional order operators for R-wave
detection in electrocardiogram signal’, IET Signal Process., 2008, 3, (5), pp.
381–391
[32] Saha, S.K., Kar, R., Mandal, D., et al.: ‘Seeker optimization algorithm:
application to the design of linear phase finite impulse response filter’, IET
Signal Process., 2012, 6, (8), pp. 763–771
[33] Mahata, S., Saha, S.K., Kar, R., et al.: ‘Optimal design of wideband digital
integrators and differentiators using harmony search algorithm’, Int. J. Numer.
Model., Electron. Netw. Devices Fields, 2017, 30, p. e2203. doi: 10.1002/
jnm.2203
[34] Mitra, S.K.: ‘Digital signal processing’ (McGraw-Hill, New York, 2006, 3rd
edn.)
[35] Mahata, S., Saha, S.K., Kar, R., et al.: ‘Optimal and accurate design of
fractional order digital differentiator – an evolutionary approach’, IET Signal
Process., 2017, 11, (2), pp. 181–196
[36] Al-Alaoui, M.A.: ‘Novel IIR differentiator from the Simpson integration
rule’, IEEE Trans. Circuits Syst. I, Fundam. Theory Appl., 1994, 41, (2), pp.
186–187
[37] Al-Alaoui, M.A.: ‘Class of digital integrators and differentiators’, IET Signal
Process., 2010, 5, (2), pp. 251–260
[38] Mahata, S., Saha, S.K., Kar, R., et al.: ‘Optimal design of wideband IIR
fractional order digital integrators using colliding bodies optimization
algorithm’, IET Signal Process., 2016, 10, (9), pp. 1135–1156
[39] Gupta, M., Relan, B., Yadav, R., et al.: ‘Wideband digital integrators and
differentiators designed using particle swarm optimization’, IET Signal
Process., 2014, 8, (6), pp. 668–679
[40] Mahata, S., Saha, S.K., Kar, R., et al.: ‘Optimal design of wideband digital
integrators and differentiators using hybrid flower pollination algorithm’, Soft
Comput., 2017, 22, (11), pp. 3757–3783. doi: 10.1007/s00500-017-2595-6
[41] Mirjalili, S.: ‘Moth-flame optimization algorithm: a novel nature-inspired
heuristic paradigm’, Knowl.-Based Syst., 2015, 89, pp. 228–249
[42] Goldberger, A.L., Amaral, L.A.N., Glass, L., et al.: ‘Physiobank,
PhysioToolkit, and PhysioNet: components of a new research resource for
complex physiologic signals’, Circulation, 2000, 101, (23), pp. e215–e220
[43] Moody, G.B., Mark, R.G.: ‘The impact of MIT-BIH arrhythmia database’,
IEEE Eng. Med. Biol. Mag., 2001, 20, (3), pp. 45–50
IET Signal Process., 2018, Vol. 12 Iss. 9, pp. 1172-1184
© The Institution of Engineering and Technology 2018