A noise robust QRS delineation method based on path simplification

In this work we present a new method for the delineation of QRS complexes in ECG signals. The main objective is to provide a robust method that gives acceptable results even under severe noise conditions in the input signal, as often happens in continuous bedside or home monitoring. Our method is based on relevant point selection using a path simplification algorithm, and then a clustering strategy combined with a qualitative description of the waveform is applied in order to select the most promising signal segments to include inside the QRS limits. The validation was performed by adding different levels of noise to the records of a standard database, and then comparing our proposal with a state-of-the art approach. Results show a high sensitivity and stable error evolution even at the highest noise levels.