An automatic R-peak detection method based on hierarchical clustering

The detection of R peaks in electrocardiogram (ECG) is an important task because R peaks can be used to identify the heart rate in order to detect different types of cardiac abnormalities including arrhythmias. This paper proposes a novel R peak detection algorithm from ECG based on a machine learning algorithm named hierarchical clustering. We evaluate the algorithm by using the 48 half-hour ECG records of MIT-BIT arrhythmias database and compare with different techniques. Our R peak detector achieves average detection accuracy of 99.83%, a sensitivity of 99.89% and a positive predictive value of 99.94% over the validation database and the results also show the proposed algorithm significantly reduces the false detection of the R-peaks