QRS complex detection using fractional Stockwell transform and fractional Stockwell Shannon energy

QRS complex present in Electrocardiogram (ECG) is the most vital component which is used as a basis for determining the condition of a human heart. However, due to the non-stationary nature of ECG, QRS detectors are unable to accurately delineate the R-peaks which may result in significant false negatives and false positives. So, in order to improve the detection rate of ECG monitoring system, this paper introduces a novel technique by amalgamating fractional Fourier transform and Stockwell transform i.e., fractional Stockwell transform (FrST) for improving the accuracy and simultaneously suppressing artifacts affecting the ECG. The proposed technique employed in this paper not only assures good detection rate but also provides an effective basis to various front end ECG signal processing measures. It also focuses on accurately identifying the QRS complex of unclassifiable beats which are among the five beat classes of Arrhythmia recommended by the Association for Advancement of Medical Instrumentation (AAMI). The proposed approach follows the five-stage methodology for correctly identifying the occurrence of R-peaks in the presence of noise. Performance is validated against ECG records taken from the MIT-BIH Arrhythmia database. The results prove the superiority of the proposed technique by achieving a sensitivity of 99.99%, positive predictivity of 99.97%, detection accuracy of 99.97%, and error rate of 0.03%. (C) 2019 Elsevier Ltd. All rights reserved.