Extraction of the Fetal Electrocardiogram Using Dynamic Neural Networks

Abstract

This article presents a nonlinear dynamics model for discriminating the sources of the maternal abdominal elec-trocardiogram (aECG). The coefficients of the separating matrix were determined by training a neural network. This method provides efficient extraction of the fetal electrocardiogram (fECG) independently of the choice of recording point, input signal duration, or number of independent leads.