Detection of R-peaks using fractional Fourier transform and principal component analysis

• Published in: Journal of ambient intelligence and humanized computing Vol. 13; no. 2; pp. 961 - 972
• Main Authors: Gupta, Varun, Mittal, Monika, Mittal, Vikas, Chaturvedi, Yatender
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.02.2022; Springer Nature B.V
• Subjects: Algorithms; Amplitudes; Artificial Intelligence; Casualties; Computational Intelligence; Datasets; Electrocardiography; Engineering; Fourier transforms; Heart; Morphology; Noise; Original Research; Physiology; Principal components analysis; Robotics and Automation; S waves; Signal processing; Support vector machines; User Interfaces and Human Computer Interaction; Wavelet transforms; Electrocardiogram (ECG); Fractional Fourier transform (FrFT); MIT-BIH Arrhythmia database (M/B Ar DB); Principal component analysis (PCA)
• ISSN: 1868-5137; 1868-5145
• DOI: 10.1007/s12652-021-03484-3

An electrocardiogram (ECG) is world’s most recognized, widely accepted and essential primitive diagnostic tool to assess health status of heart of a subject (patient) by analyzing its constituent P, QRS and T waves. QRS wave is further consists of three waves namely; Q-wave, R-wave, and S-wave, where R-wave has highest amplitude (about 1 mVolt known as R-peaks). Despite of higher amplitudes, their detection by visual inspection is still challenging due to physiological variability and presence of various types of noise/distortion in acquired ECG signal. Pre-processing of raw ECG datasets can help in tackling these two problems to some extent but that incurs an appreciable amount of computational effort. Therefore in this paper, the need of pre-processing is made redundant by using fractional Fourier transform (FrFT) for extracting features i.e. directly using the raw ECG datasets alongwith using well-known principal component analysis (PCA) for detecting R-peaks effectively in the presence of varying morphologies of ECG signal and various types of noise/distortions. Obviating the need of pre-processing altogether results in faster computations and use of PCA results in higher detection accuracies. The proposed technique has been evaluated on the basis of sensitivity (Se), positive predictive value (PPV), & accuracy (Acc) with 99.93% of Se, 99.95% of PPV, & 99.88% of Acc on MIT-BIH Arrhythmia database (M/B Ar DB). The proposed methodology will a long way in assisting the cardiologists in efficient, effective and timely computer-aided diagnosis of irregularities in heart rhythms of a subject (patient).