Extrakce abdominálního plodového vektorkardiogramu
Loading...
Downloads
6
Date issued
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Vysoká škola báňská – Technická univerzita Ostrava
Location
Signature
Abstract
In the theoretical part, the diploma thesis deals with the study of signal processing, non-invasive methods of fetal electrocardiogram scanning with a focus on single-channel methods. The aim was to select a suitable method for fetal ECG extraction, which can extract fetal ECG in sufficient quality to plot the fetal vector cardiogram. After the search, the template substraction method was chosen for a very good extraction evaluation. The aim of the practical part is to verify whether the electrocardiograms of the fetus in three axes can be obtained in a non-invasive way so that they can be used to plot the fetal vector cardiogram. Furthermore, the practical part is focused on the generation of synthetic abdominal signals, preprocessing and processing of abdominal signals with subsequent extraction of the fetal ECG by five methods of template subtraction. The experiment was performed on artificial data generated from the FECGSYNDB database, which allows to set a large number of signal parameters, maternal and fetal ECG ratios and different types of artifacts. A part of the experiment is the evaluation of the success of each extraction method by comparison with the original fetal ECG, using the mean square deviation (MSE), the standard deviation (RMSE) and the correlation in each scanning axis. This comparison was allowed by using synthetic data that uses the original, undistorted fetal ECG signal to generate the abdominal signal. This original signal was stored during generation and later, after extraction of the fetal ECG from the abdominal signal, it was used for comparison. According to MSE, the SUZANNA method achieves the best results; the best correlation between the original and extracted signal is by the CERUTTI method, but it cannot be claimed that both methods are significantly better than the others. There was a significant difference in time. The PCA method requires about 25 to 40 times longer extraction time than other methods. The LP method showed a singularity at certain points at the end of the signal, which was also reflected in the extracted signal.
Description
Subject(s)
Vectorcardiography, fetal electrocardiogram extraction, non - invasive abdominal sensing, template substraction, heart rate, single channel methods.