Biomedical Signal Processing and Control 49 (2019) 440–464 Contents lists available at ScienceDirect Biomedical Signal Processing and Control journal homepage: www.elsevier.com/locate/bspc An optimally designed digital differentiator based preprocessor for R-peak detection in electrocardiogram signal Chandan Nayak a,∗ , Suman Kumar Saha a , Rajib Kar b , Durbadal Mandal b a Department of Electronics and Telecommunication Engineering, NIT Raipur, Raipur, Chhattisgarh, 492010, India b Department of Electronics and Communication Engineering, NIT Durgapur, Durgapur, West Bengal, 713209, India a r t i c l e i n f o Article history: Received 30 August 2017 Received in revised form 26 July 2018 Accepted 3 September 2018 Keywords: Electrocardiogram (ECG) Gravitational search algorithm Integer order digital differentiator Hilbert transform QRS complex detection a b s t r a c t Globally the human death rate is accelerating day by day due to the cardiovascular diseases (CVDs), and it will be elevated shortly. In this scenario, the QRS complex detection of electrocardiogram (ECG) signal is considered as a simple, non-invasive, inexpensive, and preliminary diagnosis method used to assess the cardiac health of a patient. In this paper, an optimally designed Integer Order Digital Differentiator (IODD) based preprocessor is proposed for the accurate estimation of R-peak locations in the ECG signal. IODD, one of the major constituents of the preprocessor, is designed most proficiently by using a metaheuristic evolutionary optimization method called Gravitational Search Algorithm (GSA). In GSA, as the number of iteration increases the exploration capability fades out, and the exploitation capability fades in, which help it to avoid the local optima stagnation problem and results in faster convergence. The IODD based preprocessor accentuates the QRS complexes of the ECG signal irrespective of its abnormal morphology. The employed detector is a simple threshold independent R-peak decision logic designed by utilizing the properties of the Hilbert transform. In order to emphasize the superiority of the proposed research work the proposed IODD based QRS detection approach is validated on the first channel records of MIT/BIH Arrhythmia database (MBAD), QT database (QTDB), MIT/BIH noise stress test database (NSTDB), atrial fibrillation termination challenge database (AFTDB), and MIT/BIH ST change database (STDB). The sensi- tivity (Se) and positive Predictivity (+P) values for MBAD, QTDB, NSTDB, AFTDB, and STDB are Se=99.92% and +P=99.92%, Se=99.98% and +P=99.96%, Se=95.23% and +P=94.41%, Se=99.03% and +P=99.76%, and Se=99.93% and +P=99.90%, respectively. These performance metrics ensure the accuracy of the proposed R-peak detection technique for a wide variety of QRS morphologies and thereby affirm the applicability of the proposed IODD for the efficient detection of R-peak locations. The performance of the proposed R-peak detector significantly outperforms the reported methods in terms of all the performance met- rics. The enhanced QRS detection accuracy of the proposed approach is due to the better feature signal generating capability of the proposed IODD. © 2018 Elsevier Ltd. All rights reserved. 1. Introduction Cardiovascular diseases (CVDs) disrupt the normal function of the heart and blood vessels. In this situation, an electrocardiogram (ECG) is a straightforward, non-invasive, cost-effective, and prelim- inary diagnosis method used to detect the risk factors of the human heart, which facilitates the physician for early diagnosis and timely treatment of a patient. An ECG is a quasi-periodical, rhythmically repeating signal synchronized with the function of the heart, which acts as a generator of bioelectric events. The ECG signal comprises ∗ Corresponding author. E-mail address: chandanayak234@gmail.com (C. Nayak). several segments where each segment represents a particular event of the whole cardiac cycle. Out of all the segments, the QRS com- plex which denotes the ventricular depolarization state is the most notable. From the position and the shape of the QRS complex in the ECG signal, the health of the human heart can be studied. Hence, the precise detection of R-peak (highest peak within the QRS complex) location is the prime job for any autonomous ECG signal analysis system. The occurrence of R-peaks can easily be estimated for nor- mal noise free ECG signal. However, the challenges arise when the ECG signal is corrupted by noise and artefacts. The QRS complex detection technique is a three-decade matured topic. However, due to noisy ECG and its complex mor- phology, to date, it is not possible to detect all the types of QRS complexes by any individual algorithm. The QRS complex detection method is a two-stage process which consists of the preprocessing https://doi.org/10.1016/j.bspc.2018.09.005 1746-8094/© 2018 Elsevier Ltd. All rights reserved.