R-peaks detection based on stationary wavelet transform

• This work makes a R-peaks detection based on stationary wavelet transform (SWT).
• It has the same sampling time of the signal.
• It inherently leads to more accurate estimation of the QRS fiducial points.
• It detect R peaks over different types of arrhythmias and types of noise.
• Only one scale of SWT is applied, in order to reduce the computational complexity.

Automatic detection of the QRS complexes/R-peaks in an electrocardiogram (ECG) signal is the most important step preceding any kind of ECG processing and analysis. The performance of these systems heavily relies on the accuracy of the QRS detector. The objective of present work is to drive a new robust method based on stationary wavelet transform (SWT) for R-peaks detection. The decimation of the coefficients at each level of the transformation algorithm is omitted, more samples in the coefficient sequences are available and hence a better outlier detection can be performed. Using the information of local maxima, minima and zero crossings of the fourth SWT coefficient detail, the proposed algorithm identifies the significant points for detection and delineation of the QRS complexes, as well as detection and identification of the QRS individual waves peaks of the pre-processed ECG signal.Various experimental results show that the proposed algorithm exhibits reliable QRS detection as well as accurate ECG delineation, achieving excellent performance on different databases, on the MIT-BIH database (Se = 99.84%, P = 99.88%), on the QT Database (Se = 99.94%, P = 99.89%) and on MIT-BIH Noise Stress Test Database, (Se = 95.30%, P = 93.98%). Reliability and accuracy are close to the highest among the ones obtained in other studies. Experiments results being satisfactory, the SWT may represent a novel QRS detection tool, for a robust ECG signal analysis.