Hybridní metody pro extrakci plodového elektrokardiogramu

Abstract

The aim of this thesis is non-invasive processing of fetal electrocardiogram (fECG) using hybrid methods, which combine two or more transabdominal methods. The theoretical part is dedicated to the problems of fetal electrocardiography, complex overview of already existing hybrid methods and mathematical description of implemented methods. The experimental part is primarily focused on testing and analysis combinations of independent component analysis (ICA), wavelet transform (WT), wavelet shrinkage (WS), empirical mode decomposition (EMD), ensemble empirical mode decomposition (EEMD) and principal component analysis (PCA). The evaluation of extraction quality is based on the determination of fetal heart rate variability (fHRV). The performence of hybrid methods in this thesis is better than usage of individual ICA method. The output of the thesis is also implementation of methods in the graphical user interface in Matlab.