Wavelet Packet Transform Based Adaptive R-Peak Detection in ECG Signals Using CSE Database
Main Article Content
Abstract
Accurate detection of P, QRS, and T waves is essential for automatic electrocardiogram (ECG) analysis and cardiac diagnosis. This paper presents a robust ECG delineation method based on Wavelet Packet Transform (WPT) combined with adaptive peak detection. The ECG signal from the CSE database sampled at 500 Hz is first preprocessed using a Kaiser window-based FIR bandpass filter (0.5–40 Hz). A level 4 WPT decomposition is applied to enhance the QRS complex. R peaks are detected using energy-based thresholding, and P and T waves are identified using physiological search windows around each R peak. The proposed method demonstrates reliable detection performance and accurate counting of P, R, and T waves, making it suitable for clinical ECG analysis. For this methos the result is for detection of QRS complexes with a accuracy of 98.57.
Introduction: Cardiovascular diseases remain a major global health concern, making accurate ECG analysis essential for early diagnosis. The electrocardiogram contains clinically significant components such as P, QRS, and T waves. Among these, reliable detection of the QRS complex is fundamental for heart rate estimation and rhythm analysis. Conventional wavelet methods provide limited frequency resolution for ECG processing. Wavelet Packet Transform (WPT) offers improved time–frequency decomposition, making it suitable for accurate ECG feature extraction.
Objective: The objective of this study is to develop a robust and adaptive R-peak detection algorithm using Wavelet Packet Transform. The method aims to enhance QRS complexes while suppressing noise and baseline disturbances. It also seeks to improve detection accuracy using adaptive thresholding and physiological constraints. The proposed approach is evaluated using the CSE ECG database sampled at 500 Hz.
Methods: The ECG signals are preprocessed using a Kaiser window-based FIR bandpass filter (0.5–40 Hz) to remove noise and baseline drift. A level-4 Wavelet Packet Transform with db6 wavelet is applied for detailed frequency decomposition. Selected nodes corresponding to the QRS frequency band are reconstructed to enhance QRS energy. First derivative, squaring, and moving average integration are performed for energy enhancement. Adaptive thresholding with a refractory period is used for accurate R-peak detection.
Results: The proposed algorithm was tested on the complete CSE ECG dataset containing 17,988 QRS complexes. The method achieved 17,625 true positives, 363 false negatives, and 106 false positives. The sensitivity obtained is 97.8%, and positive predictivity is 98.57%. The detection error rate is limited to 2.60%. These results demonstrate reliable and consistent R-peak detection performance.
Conclusion: A Wavelet Packet Transform-based adaptive R-peak detection method has been successfully implemented and validated. The level 4 db6 decomposition effectively enhances QRS components while suppressing noise. Adaptive thresholding combined with physiological constraints improves detection reliability. The method achieves high sensitivity and positive predictivity on the CSE database. The proposed approach is suitable for automated ECG analysis applications.
Article Details
References
J. Pan, and W.J. Thompkins, “Real time QRS Detection Algorithm,” IEEE Trans. BME., Vo.32, No.3, pp. 230-236, Mar. 1985.
C. Li, Zheng C., and Changfeng T., “Detection of ECG Characteristic Points Using Wavelet Transforms,” IEEE Trans. BME., Vol.42, No.1, pp. 22-28, Jan. 1995.
A Novel Approach of P and T Wave Detection using Wavelets Nilesh Parihar, Chetan Gupta, Ms. Megha Desai, August 2025 | IJIRT | Volume 12 Issue 3 | ISSN: 2349-6002.
Wickerhauser Mladen Victor. Lectures on wavelet packet algorithms in Lecture notes, INRIACiteseer 1991.
Wavelet Packet and Matched Filter inspired QRS Detector, Carlos M. Pais1 and Hugo Leonardo Rufiner, Science Center for Signals, Systems and Computational Intelligence (fich.unl.edu.ar/sinc) Anales del VI Congreso Latinoamericano de Ingeniería Biomedical (CLAIB 2014), pp. 403, oct, 2014.
V. S. Chouhan, Dr. S. S. Mehta, “Delineation of QRS-complex, P and T - wave in 12 - lead ECG”, international journal of computer science and network security, vol. 8 no.4, April 2008.
Y. T. Chan, “Wavelet packet analysis of ECG signals,” Proceedings of IEEE EMBS, 2002
A. Subasi, “ECG signal classification using wavelet feature extraction,” Expert Systems with Applications, vol. 34, no. 2, pp. 264–270, 2008.
U. R. Acharya et al., “Automated ECG analysis using wavelet packet decomposition,” Computers in Biology and Medicine, vol. 42, no. 9, pp. 912–920, 2012.
S. Banerjee and M. Mitra, “Wavelet packet-based QRS detection in noisy ECG,” IEEE Sensors Journal, vol. 21, no. 8, pp. 10234–10242, 2021.
M. Alkhodari et al., “Comparison of DWT and WPT for ECG feature extraction,” IEEE Access, vol. 10, pp. 11844–11856, 2022.
S. Kaur and R. K. Sharma, “Energy-based wavelet packet node selection for QRS detection,” IEEE Access, vol. 11, 2023.
Yun et al. (2022) – Robust R-peak detection in an electrocardiogram with stationary wavelet transformation and separable convolution
Aqil & Jbari (2025) – Electrocardiogram features detection using stationary wavelet transform.
Adam et al. (2025) – Comparative evaluation of filtration techniques for ECG signal denoising with emphasis on stationary wavelet transform.