R-Peak Identification in ECG Signals using Pattern-Adapted Wavelet Technique

The electrocardiogram (ECG) signal consists of vital information that can be used in detecting various heart diseases. R-peaks in ECG signal play a major role in the diagnosis of the heart disorder. While numerous methods exist for the purpose, this research work aims at improving the efficiency of R-peak detection through a novel pattern-adapted wavelet designed to reduce the rate of false positives and the detection error. The experimental results show that the proposed pattern-adapted wavelet method achieves better performance when compared with the Symlet4 and other published methods. The new wavelet was designed using the least square optimisation method such that it not only approximates the given R-peak pattern of the ECG signal but also is admissible according to the constraints prescribed by Continuous Wavelet Transform (CWT). The algorithm uses the wavelet-specific property that CWT coefficients of a given signal are computed where local maximum and minimum pair appear around the signal peak location. When applied to the signals available through the standard MIT-BIH (Massachusetts Institute of Technology, Beth Israel Hospital) Arrhythmia database, Symlet4 detects R-peaks with an average of 98.73% accuracy, 99.99% sensitivity, 98.74% positive predictive value, 1.3336% error rate and overall F-score of 0.9937, while the proposed pattern-adapted wavelet detects the same with an average of 99.83% accuracy, 99.91% sensitivity, 99.92% positive predictive value, 0.16% error rate and overall F-score of 0.999.

Automated summary

The R-peaks in the electrocardiogram (ECG) signal are crucial for diagnosing heart disorders. This research proposes a novel pattern-adapted wavelet that improves the efficiency of R-peak detection by reducing false positives and detection errors. Experimental results demonstrate that this method outperforms existing methods like Symlet4.