Metody pro měření a zpracovávání děložních kontrakcí pomocí elektrohysterografie

Abstract

The subject of this bachelor thesis is the study of the measurement method of electrohysterography, which is used to monitor the electrical activity of the uterus and testing of selected methods for EHG signal processing. The introductory part of the thesis explains important topics that are essential for understanding the subject such as topography, anatomy and histology of the uterus, pregnancy and childbirth, available techniques for monitoring uterine contractions, properties of the EHG signal, publicly available databases and signal processing. Discrete wavelet transform and empirical modal decomposition methods were applied to signals from the TPEHGT database. Envelopes of the decomposed signal were obtained using the Hilbert transform. Contractions were detected in the smoothed signal envelopes and the overall accuracy and error of contraction detection were calculated. Finally, the methods were compared with each other and the results were evaluated. Using a discrete wavelet transform for EHG signal processing seems to be better option. The overall accuracy of contraction detection in signals decomposed using the discrete wavelet transform was 82 % compared to signals decomposed using empirical modal decomposition, where the overall accuracy of contraction detection was 71 %.