A Robust Algorithm for R-peak Detection in an ECG Waveform Using Local Threshold Computed over a Sliding Window

摘要

Detection of the R peak in an ECG signal is central to detection of arrhythmia. Fast reliable methods to detect R peaks in a ECG signal thus assume great importance, particularly in emerging applications such as automatic mobile based monitoring of a patient's cardiac condition. In this paper, we propose a fast R-peak detection algorithm that uses adaptive thresholding to compute the locations of the R-peaks. Thresholds for detecting peaks are computed based on differentiated ECG signal over a sliding window. This ensures that the computed threshold takes into account local characteristics of the signal thus resulting in higher peak detection accuracy. The proposed algorithm was tested over the MIT-BIH arrhythmia database and accuracy of the peak detection was verified with the doctors annotation. Its performance was compared with Pan-Tompkins (PT) [1| and Difference Operation Method (DOM) |2| algorithm. The proposed algorithm shows better accuracy in peak detection compared to them.