A FULLY AUTOMATIC METHOD FOR ACCURATE PARAMETRIZATION AND RECONSTRUCTION OF ECG WAVEFORMS

摘要

Parameterization and synthesis of electrocardiogram(ECG) recordings are some of the most challengingproblems in biomedical signal processing owing to thefact that ECG signals commonly exhibit complextemporal morphology and contain numerous artifacts ofdata collection process. In this paper, we present a fullyautomatic framework for accurate and robustparameterization and reconstruction of ECG waveforms.The method uses the observed signal to ascertain a nondeterministicmodel for the ECG signal and employs theDynamic Time warping (DTW) algorithm to determine anon-linear temporal relationship between the establishedECG model and the individual pulses in the ECG signal.The results of parameterization provide a set of data thataccurately describe the morphology of the ECG pulses.The proposed signal synthesis algorithm is able toindependently account for the temporal and spatialdynamics of consecutive ECG pulses and provide afaithful reconstruction of ECG signals. Performanceevaluation experiments are conducted on a database of135 one-minute ECG recordings. The percentage rootmean-square difference measure is employed to evaluatethe quality of signal reconstruction and also validate theresults of signal parameterization.