Robust identification of QRS-complexes in electrocardiogram signals using a combination of interval and trigonometric threshold values. (August 2020)

• Record Type: Journal Article
• Authors: Mukhopadhyay, Sourav Kumar; Krishnan, Sridhar
• Is Part Of: Biomedical signal processing and control. Volume 61(2020)
• Publication Date: 2020-08
• Subjects: Histogram analysis -- Hilbert transform -- Shannon energy transform -- Trigonometric feature -- QRS-complex detection; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2020.102007
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: QRS-complex detection from ECG signals using a unique trigonometric-trait. Histogram-based signal-specific amplitude-thresholding. High detection-rate and fast, and the performance of the proposed algorithm is tested on ECG signals of 15.91 days long. The trigonometric-trait can be used to extract the breathing rate and pattern. The trigonometric-trait can also be used for the identification of the abnormal QRS-complexes. Abstract: Combination of interval and amplitude threshold-based algorithms have been tested rigorously over the last few decades for detecting the QRS-complexes in electrocardiogram (ECG) signals. However, the very eccentric nature of appearance of the QRS-complexes often make an amplitude threshold-based algorithm fail. This paper presents a unique combination of trigonometric feature and interval threshold-based algorithm for efficient detection of the QRS-complexes. First, the ECG signal is denoised, and the first-difference (FD) of the denoised ECG is calculated. The FD signal is then sequentially passed through a Hilbert and Shannon energy (SE) transformers. Next, a histogram-based analysis technique is applied on the SE-transformed data to filter out the contributions of the low frequency components of the ECG signal. Next, each of the modified SE-transformed data having a non-zero amplitude value is mapped onto the denoised ECG signal, and a unique trigonometric feature is extracted from that corresponding location of the denoised ECG signal. Finally, the QRS-complexes are identified using a combination of trigonometric and interval-threshold values. Wired and wireless ECG signals are collected from seven databases, and are used as the evaluation test-beds of the proposed algorithm. The performance of the algorithm is found highly-competent compared to that of the state-of-the-art methods. The significance of the proposed algorithm is that, not only the detection-accuracy of the algorithm is high, it is also fast; the trigonometric feature can be used to extract the breathing rate and pattern; the trigonometric feature can also be used for the identification of the abnormal QRS-complexes.