Reliable evaluation of fetal condition and early detection of fetal distress is one of the largest challenges in modern obstetrics. Safely protected within the maternal womb, the fetus is rather inaccessible for physiological measurements. One of few physiological phenomena that can be measured antenatal, is fetal heart activity. The heart plays an essential role in the transportation of oxygen to the tissues, but is only one of multiple factors that influence oxygen supply. Consequently, fetal heart activity provides direct, but rather limited information for the evaluation of fetal condition. Cardiotocography, the simultaneous recording of fetal heart rate and uterine activity, has been the standard in fetal monitoring for more than 30 years. Nevertheless, cardiotocography is insufficiently capable of predicting bad fetal outcome and therefore its value in clinical practice is limited. As a result, any additional information that can contribute to reliably evaluating fetal condition, is highly appreciated. The beat-to-beat variability of the fetal heart rate is an expression of cardiovascular control by the autonomic part of the fetal central nervous system. As this cardiovascular control will respond to changes in fetal condition, fetal heart rate variability will indirectly reflect fetal condition. Fetal heart rate activity therefore contains potentially useful information that cardiotocography does not reveal. As modulation by different parts of the autonomic nervous system occurs on characteristic timescales, timefrequency analysis of fetal heart rate variability might provide additional information that can be used to more reliably assess fetal well-being. However, interpretation of this information is complicated by the complexity of the physiological mechanisms for cardiovascular control. Additionally, to obtain accurate spectral information, the beat-to-beat fetal heart rate is required, which can only be obtained in clinical practice by measuring the fetal electrocardiogram directly from the fetal scalp. This currently limits the application of the method to intrapartum measurements. To further explore the potential of time-frequency analysis of fetal heart rate variability for monitoring fetal condition, application of the analysis technique to antepartum measurements is highly appreciated. The first goal of this doctoral dissertation therefore is to: 1. Obtain the beat-to-beat fetal heart rate throughout pregnancy Given the limited successes in literature, it is expected that the obtained fetal heart rate will contain considerably more artifacts than the fetal heart rate obtained from scalp ECG measurements during labor does. Standard techniques for time-frequency analysis, such as the fast Fourier transform, will then fail to provide accurate spectral information. The second goal of this dissertation therefore is to: 2. Obtain accurate spectral information on antepartum fetal heart rate variability To measure fetal heart activity antepartum, a dedicated data-acquisition system has been developed for electrophysiological measurements on the abdomen of a pregnant woman (chapter 2). A novel method developed by a coworker was chosen to remove the dominating maternal electrocardiogram from the recorded signals. An online software implementation of this method has been realized to process the recorded signals real-time. To achieve the first goal of the dissertation, chapter 3 presents an algorithm that uses a priori knowledge on the physiology of the fetal heart to enhance the fetal ECG components in multi-lead abdominal fetal ECG recordings, before QRS-detection. Evaluation of the method on generated fetal ECG recordings with controlled signal-to-noise ratios showed excellent results. However, for actual recordings, evaluation of the results by experts learned that fine-tuning of the algorithm is necessary. In chapter 4, a more theoretical approach has been used to exploit the spatial correlation of multi-channel fetal ECG recordings for increasing the signal-to-noise ratio of the retrieved fetal electrocardiogram. A threedimensional representation of the fetal vectorcardiogram is constructed by erse Dower matrix. An ellipse is fitted to the QRS loop of several overlayed heartbeats and the axes of the ellipse are calculated to determine the source signals of the fetal electrocardiogram. In future work, this technique could be used for calculating the linear combinations that are used in the algorithm of chapter 3, which will increase the accuracy of the heart rate detection. The suitability of non-invasive fetal ECG recordings for fetal monitoring in clinical practice was evaluated by using the developed technology in a longitudinal patient study (chapter 5). Repeated measurements on pregnant patients learned that the performance of the method for removing the maternal electrocardiogram was good and remained more or less constant throughout pregnancy. Between 20 and 25 weeks of gestational age, the quality of the retrieved fetal ECG waveforms generally was very high, and the beat-to-beat fetal heart rate could be accurately detected. For this stage of pregnancy, abdominal measurement of the fetal electrocardiogram offers an opportunity to obtain unique cardiac information on the fetus. However, to increase the performance of the technology throughout pregnancy, the noise in the electrophysiological recordings must be significantly reduced. Still, it remains uncertain whether this will be adequate when isolating sections of the vernix caseosa reduce the amplitude of the fetal electrocardiogram that is measured on the maternal abdomen. For stages of pregnancy in which abdominal recording of the fetal electrocardiogram fails to provide the beat-to-beat heart rate, chapter 6 offers an alternative. By processing Doppler waveforms of ultrasound signals reflected at the fetal heart, the beat-to-beat fetal heart rate can be obtained. However, the measurement requires a skilled operator and is very sensitive to fetal movement. The presence of artifacts in the beat-to-beat fetal heart rate obtained from either abdominal recordings of the fetal electrocardiogram or Doppler ultrasound recordings, is common and cannot be prevented. To obtain the second goal of the dissertation, a continuous wavelet based analysis method has been developed to reliably calculate the power within the scales of interest (chapter 7). This method provides accurate results when up to 20 % of the dataset is missing due to artifacts. In chapter 8, the continuous wavelet based method has been applied for time-scale analysis of the recordings from chapter 5. The results of this analysis correspond with literature on the development of the fetal autonomic nervous system. In addition, the results suggest that functional development of the sympathetic nervous system takes place around 22 weeks of gestational age. The final chapter reflects on the realization of the goals of this dissertation and provides specific directions for future work. Although additional clinical research might contribute to obtaining clinically relevant information from time-scale analysis of fetal heart rate variability, focus should be on solving the technical limitations of the used instrumentation for abdominal recording of the fetal electrocardiogram.
|Qualification||Doctor of Philosophy|
|Award date||30 Mar 2011|
|Place of Publication||Eindhoven|
|Publication status||Published - 2011|