Application of PNN for the Detection of QRS-complexes in Electrocardiogram using Combined Entropy

This paper presents a new classification scheme for the automatic detection of QRS-boundaries in Electrocardiogram (ECG) using Probabilistic Neural Networks (PNN). Digital filtering techniques are used to remove power line interference and baseline wander present in the ECG signal. PNN is used as a classifier to delineate QRS and non-QRSregions in single-lead ECG signal. The algorithm is implemented using MATLAB. The performance of the algorithm is validated using each lead of the 12-lead simultaneously recorded ECGs from the dataset-3 of the CSE multi-lead measurement library. Significant detection rate of 99.50% is achieved. The percentage of false positive and false negative is 0.63% and 0.50% respectively. The overall results obtained show the capability of PNN in terms of the detection rate performance in comparison to the other methods reported in literature.