K-means algorithm for the detection and delineation of QRS-complexes in Electrocardiogram

Electrocardiogram (ECG) is an important bioelectrical signal used to asses the cardiac state of a patient It consists of a recurrent wave sequence of P-wave, QRS-complex and T-wave associated with each beat The QRS-complex is the prominent feature of the ECG. This paper presents a simple method using K-means clustering algorithm for the detection of QRS-complexes in ECG signal. Digital filters are used to remove the power line interference and baseline wander present in the ECG signal K-means algorithm is used to classify QRS and non-QRS-region in the ECG signal The performance of the algorithm is validated using dataset-3 of the CSE multi-lead measurement library Detection rate of 98.66% is obtained The percentage of false positive and false negative is 1.14% and 1.34% respectively. The mean and standard deviation of the errors between automatic and manual annotations is calculated to validate the delineation performance of the algorithm. The on sets and offsets of the detected QRS-complexes are found well within the tolerance limits its specified by the CSE library. (C) 2009 Elsevier Masson SAS All rights reserved.