Date of Award

12-23-2014

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Applied Science

First Advisor

Mitchell Hudson

Abstract

Intrauterine growth restriction (IUGR) is a fetal condition characterized by growth-rate reduction. Afflicted fetuses tend to display abnormalities in heart rate. Hence, studying its heart rate variability (HRV) along with comparisons against normal fetal HRV might give some insight into the heart rate dynamics between the two groups. Heart rate variability (HRV) is one of the prominent markers that could help in identifying fetal distress. Four fetal heart rate patterns were previously defined based on HRV, which can provide clinicians with valuable information regarding fetus health during pregnancy. Past studies that differentiate the HRV with or without considering patterns, have shown varied results depending on the condition's severity. The data sets chosen for this study were recorded using the SQUID (Superconducting Quantum Interference Device) technology installed at University of Arkansas for Medical Sciences (UAMS). Recording of fetal heart and brain has been possible for almost a decade using this non-invasive technology. Signal processing techniques such as orthogonal projection (OP) and Independent Component Analysis (ICA) have been applied to extract the fetal magnetocardiogram and attenuate interference from other biological sources such as maternal heart. But successful application of such techniques among other factors depends on the non-stationary characteristics of the signals. Non-stationarity can be due to maternal and/or fetal movement in long duration datasets. Previous to this work, some data sets for studying the differences in the HRV of IUGR and normal fetuses have been eliminated because of inadequate removal of maternal signals using current methods. Hence, the objective of this dissertation is to first identify the fetal heart rate pattern (fHRP) automatically. For this purpose a novel non-linear measurement was developed which was able to successfully identify fHRPs. Next, using that method the differences in the HRV of IUGR and low-risk fetuses based on fHRPs was identified successfully. To compute the fetal HRV, the maternal magnetocardiogram (MCG) must first be removed. Non-stationarity in the maternal MCG may cause inadequate suppression of the maternal MCG, thus corrupting the fetal HRV analysis. The final objective is to automate the orthogonal projection and to reduce the effect of maternal MCG. This can be achieved by applying OP to shorter segments of data allowing the method to adapt to the changing maternal MCG. A generalized method that can be applied to algorithms similar to the OP method was developed to study the effectiveness of the OP method as a function data length. Using the new evaluation tool, it was determined that the OP should not be used on data segments less than four minutes in length.

Download

Share

COinS