Performance of three QRS detection algorithms during sleep: a comparative study

A comparison of the performance of three QRS detectors used in the analysis of electrocardiogram (ECG) during sleep is presented in this paper. Two widely used QRS detection algorithms based on digital filtering (DF) are compared with a newly introduced one, based on higher-order statistics (HOS). The percentage of QRS complexes failed detection along with the number of false positives and false negatives are measured for quantitative performance evaluation. Experimental results, when applying the proposed methods to nocturnal ECG recordings from the Sleep Laboratory of the Philipps University of Marburg, Germany, prove that the HOS-based QRS detector exhibits higher overall QRS detection accuracy (99.95%) than the two DF-based ones (99.75% and 99.59%, respectively). Moreover, it has lower noise susceptibility despite the presence of different noise types, such as smooth or abrupt baseline drift, 50 Hz powerline interference, electromyographic intervention or any arrhythmia effect due to sleep apnea.