The persistent challenge of foetal heart rate monitoring

Non-linear Methods Predominant in Fetal Heart Rate Analysis: A Systematic Review

The analysis of fetal heart rate variability has served as a scientific and diagnostic tool to quantify cardiac activity fluctuations, being good indicators of fetal well-being. Many mathematical analyses were proposed to evaluate fetal heart rate variability. We focused on non-linear analysis based on concepts of chaos, fractality, and complexity: entropies, compression, fractal analysis, and wavelets. These methods have been successfully applied in the signal processing phase and increase knowledge about cardiovascular dynamics in healthy and pathological fetuses. This review summarizes those methods and investigates how non-linear measures are related to each paper's research objectives. Of the 388 articles obtained in the PubMed/Medline database and of the 421 articles in the Web of Science database, 270 articles were included in the review after all exclusion criteria were applied. While approximate entropy is the most used method in classification papers, in signal processing, the most used non-linear method was Daubechies wavelets. The top five primary research objectives covered by the selected papers were detection of signal processing, hypoxia, maturation or gestational age, intrauterine growth restriction, and fetal distress. This review shows that non-linear indices can be used to assess numerous prenatal conditions. However, they are not yet applied in clinical practice due to some critical concerns. Some studies show that the combination of several linear and non-linear indices would be ideal for improving the analysis of the fetus's well-being. Future studies should narrow the research question so a meta-analysis could be performed, probing the indices' performance.

Cardiotocography signal abnormality classification using time-frequency features and Ensemble Cost-sensitive SVM classifier

BACKGROUND

Cardiotocography (CTG) signal abnormality classification plays an important role in the diagnosis of abnormal fetuses. This classification problem is made difficult by the non-stationary nature of CTG and the dataset imbalance. This paper introduces a novel application of Time-frequency (TF) features and Ensemble Cost-sensitive Support Vector Machine (ECSVM) classifier to tackle these problems.

METHODS

Firstly, CTG signals are converted into TF-domain representations by Continuous Wavelet Transform (CWT), Wavelet Coherence (WTC), and Cross-wavelet Transform (XWT). From these representations, a novel image descriptor is used to extract the TF features. Then, the linear feature is derived from the time-domain representation of the CTG signal. The linear and TF features are fed to the ECSVM classifier for prediction and classification of fetal outcome.

RESULTS

The TF features show the significant difference (p-value<0.05) in distinguishing abnormal CTG signals, but not for traditional nonlinear features. In ECSVM abnormality classification, using only linear features, the sensitivity, specificity, and quality index are 59.3%, 78.3%, and 68.1%, respectively, whereas more effective results (sensitivity: 85.2%, specificity: 66.1%, and quality index: 75.0%) are obtained using a combination of linear and TF features, with a performance improvement index of 10.1%. Especially, the area under the receiver operating characteristic curve (0.77 vs. 0.64) is significantly increased with the ECSVM vs. SVM.

CONCLUSION

Our method can greatly improve the classification results, especially for sensitivity. It improves the true positive rate of CTG abnormality classification and reduces the false positive rate, which may help detect and treat abnormal fetuses during labor.

Systematic Review of Intrapartum Fetal Heart Rate Spectral Analysis and an Application in the Detection of Fetal Acidemia

It is fundamental to diagnose fetal acidemia as early as possible, allowing adequate obstetrical interventions to prevent brain damage or perinatal death. The visual analysis of cardiotocography traces has been complemented by computerized methods in order to overcome some of its limitations in the screening of fetal hypoxia/acidemia. Spectral analysis has been proposed by several studies exploring fetal heart rate recordings while referring to a great variety of frequency bands for integrating the power spectrum. In this paper, the main goal was to systematically review the spectral bands reported in intrapartum fetal heart rate studies and to evaluate their performance in detecting fetal acidemia/hypoxia. A total of 176 articles were reviewed, from MEDLINE, and 26 were included for the extraction of frequency bands and other relevant methodological information. An open-access fetal heart rate database was used, with recordings of the last half an hour of labor of 246 fetuses. Four different umbilical artery pH cutoffs were considered for fetuses' classification into acidemic or non-acidemic: 7.05, 7.10, 7.15, and 7.20. The area under the receiver operating characteristic curve (AUROC) was used to quantify the frequency bands' ability to distinguish acidemic fetuses. Bands referring to low frequencies, mainly associated with neural sympathetic activity, were the best at detecting acidemic fetuses, with the more severe definition (pH ≤ 7.05) attaining the highest values for the AUROC. This study shows that the power spectrum analysis of the fetal heart rate is a simple and powerful tool that may become an adjunctive method to CTG, helping healthcare professionals to accurately identify fetuses at risk of intrapartum hypoxia and to implement timely obstetrical interventions to reduce the incidence of related adverse perinatal outcomes.

Complexity of Cardiotocographic Signals as A Predictor of Labor

Prediction of labor is of extreme importance in obstetric care to allow for preventive measures, assuring that both baby and mother have the best possible care. In this work, the authors studied how important nonlinear parameters (entropy and compression) can be as labor predictors. Linear features retrieved from the SisPorto system for cardiotocogram analysis and nonlinear measures were used to predict labor in a dataset of 1072 antepartum tracings, at between 30 and 35 weeks of gestation. Two groups were defined: Group A—fetuses whose traces date was less than one or two weeks before labor, and Group B—fetuses whose traces date was at least one or two weeks before labor. Results suggest that, compared with linear features such as decelerations and variability indices, compression improves labor prediction both within one (C-Statistics of 0.728) and two weeks (C-Statistics of 0.704). Moreover, the correlation between compression and long-term variability was significantly different in groups A and B, denoting that compression and heart rate variability look at different information associated with whether the fetus is closer to or further from labor onset. Nonlinear measures, compression in particular, may be useful in improving labor prediction as a complement to other fetal heart rate features.

On the prediction of foetal acidaemia: A spectral analysis-based approach

A computational analysis of physiological systems has been used to support the understanding of how these systems work, and in the case of foetal heart rate, many different approaches have been developed in the last decades. Our objective was to apply a new method of classification, which is based on spectral analysis, in foetal heart rate (FHR) traces to predict foetal acidosis diagnosed with umbilical arterial blood pH ≤ 7.05. Fast Fourier transform was applied to a real database for the classification approach. To evaluate the models, sensitivity, specificity, and area under the receiver operating characteristic (ROC) curve were used. Sensitivity equal to 1, specificity equal to 0.85 and an area under the ROC curve of 0.94 were found. In addition, when the definition of metabolic acidosis of umbilical arterial blood pH ≤ 7.05 and base excess ≤ -10 mmol/L was used, the proposed methodology obtained sensitivity = 1, specificity = 0.97 and area under the ROC curve = 0.98. The proposed methodology relies exclusively on the spectral frequency decomposition of the FHR signal. After further successful validation in more datasets, this approach can be incorporated easily in clinical practice due to its simple implementation. Likewise, the incorporation of this novel technique in an intrapartum monitoring station should be straightforward, thus enabling the assistance of labour professionals in the anticipated detection of acidaemia.

Evolution of linear and nonlinear fetal heart rate indices throughout pregnancy in appropriate, small for gestational age and preterm fetuses: A cohort study

BACKGROUND AND OBJECTIVES

To assess the evolution of linear and nonlinear fetal heart rate (FHR) analysis throughout pregnancy in appropriate (AGA), small for gestational age (SGA) and preterm (PTB) fetuses.

METHODS

A prospective cohort study was carried out in 171 singleton pregnancies divided in three groups: AGA (n = 147), SGA (n = 13) fetuses and spontaneous PTB (n = 11). FHR was recorded with an external sensor from the 24th to the 40th week of gestation. Linear time- and frequency-domain and nonlinear FHR indices were computed on 10-min segments. Longitudinal analysis of indices throughout pregnancy was performed with generalized estimating equations, and receiver operating characteristic (ROC) curves were calculated for the prediction of SGA and PTB fetuses.

RESULTS

Increasing gestational age significantly affected most FHR indices, with a general increase in variability and entropy indices, and a decrease in mean FHR. The PTB group exhibited a significantly lower short-term variation, and no monotonic increase in the sympatho-vagal balance as observed in the AGA group. The SGA group exhibited higher long-term irregularity and lower short-term irregularity than the AGA group throughout gestation. In prediction of SGA and PTB, the largest areas under the ROC curves obtained were 0.76 and 0.78, respectively.

CONCLUSIONS

Linear and nonlinear FHR analysis provides useful information on the evolution of fetal autonomic nervous and complexity control systems throughout pregnancy, in relation with AGA, SGA and PTB fetuses, which may be helpful in clinical practice.

Gender-specific evolution of fetal heart rate variability throughout gestation: A study of 8823 cases

BACKGROUND

Fetal heart rate (FHR) variability throughout gestation reflects aspects of fetal development, and its analysis has been used for the assessment of fetal well-being.

AIM

The objective of this study was to provide a gender-specific analysis of the evolution of FHR variability indices throughout gestation, using linear time-domain, spectral and nonlinear FHR indices.

STUDY DESIGN

A large retrospective study was carried out using antepartum FHR recordings obtained from 4713 male and 4110 female fetuses, with normal pregnancy outcome, between 2004 and 2013, with gestational ages ranging between 25 and 40weeks.

OUTCOME MEASURES

FHR variability was analysed through linear time-domain methods, as well as using spectral analysis and entropy indices. Evolution of FHR indices throughout gestation was analysed through Spearman correlation coefficient. Comparison between male and female fetuses was performed using nonparametric bootstrap 95% confidence intervals for the median.

RESULTS

Mean FHR decreased significantly throughout gestation, whereas most variability indices increased. Sympatho-vagal balance measured by spectral analysis exhibited two local maxima at 29-30 and 34-35weeks and decreased afterwards. Entropy indices increased until around the 34th week, slightly decreasing after the 37th week. Female fetuses presented higher mean FHR and entropy from the 34th week afterwards, and lower short-term variability and sympatho-vagal balance in the same period.

CONCLUSIONS

Spectral and entropy analysis should be considered as a complement to conventional FHR variability analysis, aiming at a better characterization and follow-up of fetal development/maturation throughout gestation. Additionally, gestational age needs to be considered when defining reference ranges for FHR indices in systems of computerized analysis.

Agreement on intrapartum cardiotocogram recordings between expert obstetricians

RATIONALE, AIMS AND OBJECTIVES

To evaluate obstetricians' inter- and intra-observer agreement on intrapartum cardiotocogram (CTG) recordings and to examine obstetricians' evaluations with respect to umbilical artery pH and base deficit.

METHODS

Nine experienced obstetricians annotated 634 intrapartum CTG recordings. The evaluation of each recording was divided into four steps: evaluation of two 30-minute windows in the first stage of labour, evaluation of one window in the second stage of labour and labour outcome prediction. The complete set of evaluations used for this experiment is available online. The inter- and intra-observer agreement was evaluated using proportion of agreement and kappa coefficient. Clinicians' sensitivity and specificity was computed with respect to umbilical artery pH, base deficit and to Apgar score at the fifth minute.

RESULTS

The overall proportion of agreement between clinicians reached 48% with 95% confidence intervals (CI) (CI: 47-50). Regarding the different classes, proportion of agreement ranged from 57% (CI: 54-60) for normal to 41% (CI: 36-46) for pathological class. The sensitivity of clinicians' majority vote to objective outcome was 39% (CI: 16-63) for the umbilical artery base deficit and 27% (CI: 16-42) for pH. The specificity was 89% (CI: 86-92) for both types of objective outcome.

CONCLUSIONS

The reported inter-/intra-observer variability is large and this holds irrespective of clinicians' experience or work place. The results support the need of modernized guidelines for CTG evaluation and/or objectivization and repeatability by introduction of a computerized approach that could standardize the process of CTG evaluation within the delivery ward.

Computer analysis of foetal monitoring signals

Five systems for computer analysis of foetal monitoring signals are currently available, incorporating the evaluation of cardiotocographic (CTG) or combined CTG with electrocardiographic ST data. All systems have been integrated with central monitoring stations, allowing the simultaneous monitoring of several tracings on the same computer screen in multiple hospital locations. Computer analysis elicits real-time visual and sound alerts for health care professionals when abnormal patterns are detected, with the aim of prompting a re-evaluation and subsequent clinical action, if considered necessary. Comparison between the CTG analyses provided by the computer and clinical experts has been carried out in all systems, and in three of them, the accuracy of computer alerts in predicting newborn outcomes was evaluated. Comparisons between these studies are hampered by the differences in selection criteria and outcomes. Two of these systems have just completed multicentre randomised clinical trials comparing them with conventional CTG monitoring, and their results are awaited shortly. For the time being, there is limited evidence regarding the impact of computer analysis of foetal monitoring signals on perinatal indicators and on health care professionals' behaviour.

Low dimensional manifold embedding for scattering coefficients of intrapartum fetale heart rate variability

Intrapartum fetal surveillance for early detection of fetal acidosis in clinical practice focuses on reducing neonatal morbidity via early detection. It is the subject of on going research studies attempting notably to improve detection performance by reducing false positive rate. In that context, the present contribution tailors to fetal heart rate variability analysis a graph-based dimensionality reduction procedure performed on scattering coefficients. Applied to a high quality and well-documented database constituted by obstetricians from a French academic hospital, the low dimensional embedding enables to distinguish between the temporal dynamics of healthy and acidotic fetuses, as well as to achieve satisfactory detection performance detection compared to those obtained by the clinical-benchmark FIGO criteria.

How an extended perinatal audit may improve perinatal policy

Abstract Objective: A perinatal audit has the intention of quality of care improvement based on analysis of perinatal death, with our without analysis of maternal morbidity and/or mortality. Additional analysis of cases of intrapartum asphyxia could provide more insight into ways to improve quality of perinatal care. Methods: Analysis of cases of perinatal death and asphyxia in Jan Yperman Hospital, Ieper, Belgium, in 2012. Results: Three perinatal deaths occurred, none were preventable. Nineteen cases of proven metabolic acidosis have been identified. Three cases are considered possibly preventable, four cases are considered preventable. In three (possibly) preventable cases, foetal monitoring was absent during the active second stage of labour. In two preventable cases, intervention following a significant ST event in the second stage of labour was delayed. In one case intervention was delayed in the first stage of labour, while in another, indicated operative delivery in the second stage was not conducted. Conclusions: Integrating intrapartum asphyxia in the perinatal audit gives an opportunity to identify and eliminate weak points in the perinatal care chain, thereby optimizing quality of care. Lessons learned from our internal audit are the value of foetal monitoring and adequate action on significant ST events during second stage of labour.

Scattering transform for intrapartum fetal heart rate variability fractal analysis: a case-control study

Intrapartum fetal heart rate monitoring, aiming at early acidosis detection, constitutes an important public health stake. Scattering transform is proposed here as a new tool to analyze intrapartum fetal heart rate (FHR) variability. It consists of a nonlinear extension of the underlying wavelet transform, that thus preserves its multiscale nature. Applied to an FHR signal database constructed in a French academic hospital, the scattering transform is shown to permit to efficiently measure scaling exponents characterizing the fractal properties of intrapartum FHR temporal dynamics, that relate not only to the sole covariance (correlation scaling exponent), but also to the full dependence structure of data (intermittency scaling exponent). Such exponents are found to satisfactorily discriminate temporal dynamics of healthy subjects (from that of nonhealthy ones) and to emphasize the role of the highest frequencies (around and above 1 Hz) in intrapartum FHR variability. This permits us to achieve satisfactory classification performance that improves on those obtained from the analysis of International Federation of Gynecology and Obstetrics (FIGO) criteria, notably by classifying as healthy a number of subjects that were incorrectly classified as nonhealthy by classical clinically used FIGO criteria. Combined to obstetrician annotations, these scaling exponents enable us to sketch a typology of these FIGO-false positive subjects. Also, they permit us to monitor the evolution along time of the intrapartum health status of the fetuses and to estimate an optimal detection time-frame.

Entropy and compression: two measures of complexity

RATIONALE, AIMS AND OBJECTIVES

Traditional complexity measures are used to capture the amount of structured information present in a certain phenomenon. Several approaches developed to facilitate the characterization of complexity have been described in the related literature. Fetal heart rate (FHR) monitoring has been used and improved during the last decades. The importance of these studies lies on an attempt to predict the fetus outcome, but complexity measures are not yet established in clinical practice. In this study, we have focused on two conceptually different measures: Shannon entropy, a probabilistic approach, and Kolmogorov complexity, an algorithmic approach. The main aim of the current investigation was to show that approximation to Kolmogorov complexity through different compressors, although applied to a lesser extent, may be as useful as Shannon entropy calculated by approximation through different entropies, which has been successfully applied to different scientific areas.

METHODS

To illustrate the applicability of both approaches, two entropy measures, approximate and sample entropy, and two compressors, paq8l and bzip2, were considered. These indices were applied to FHR tracings pertaining to a dataset composed of 48 delivered fetuses with umbilical artery blood (UAB) pH in the normal range (pH ≥ 7.20), 10 delivered mildly acidemic fetuses and 10 moderate-to-severe acidemic fetuses. The complexity indices were computed on the initial and final segments of the last hour of labour, considering 5- and 10-minute segments.

RESULTS

In our sample set, both entropies and compressors were successfully utilized to distinguish fetuses at risk of hypoxia from healthy ones. Fetuses with lower UAB pH presented significantly lower entropy and compression indices, more markedly in the final segments.

CONCLUSIONS

The combination of these conceptually different measures appeared to present an improved approach in the characterization of different pathophysiological states, reinforcing the theory that entropies and compressors measure different complexity features. In view of these findings, we recommend a combination of the two approaches.

Linear and nonlinear measures of fetal heart rate patterns evaluated on very short fetal magnetocardiograms

We analyzed the effectiveness of linear short- and long-term variability time domain parameters, an index of sympatho-vagal balance (SDNN/RMSSD) and entropy in differentiating fetal heart rate patterns (fHRPs) on the fetal heart rate (fHR) series of 5, 3 and 2 min duration reconstructed from 46 fetal magnetocardiograms. Gestational age (GA) varied from 21 to 38 weeks. FHRPs were classified based on the fHR standard deviation. In sleep states, we observed that vagal influence increased with GA, and entropy significantly increased (decreased) with GA (SDNN/RMSSD), demonstrating that a prevalence of vagal activity with autonomous nervous system maturation may be associated with increased sleep state complexity. In active wakefulness, we observed a significant negative (positive) correlation of short-term (long-term) variability parameters with SDNN/RMSSD. ANOVA statistics demonstrated that long-term irregularity and standard deviation of normal-to-normal beat intervals (SDNN) best differentiated among fHRPs. Our results confirm that short- and long-term variability parameters are useful to differentiate between quiet and active states, and that entropy improves the characterization of sleep states. All measures differentiated fHRPs more effectively on very short HR series, as a result of the fMCG high temporal resolution and of the intrinsic timescales of the events that originate the different fHRPs.

Fractal and nonlinear changes in the long-term baseline fluctuations of fetal heart rate

The interpretation of heart rate patterns obtained by fetal monitoring relies on the definition of a baseline, which is considered as the running average heart rate in the absence of external stimuli during periods of fetal rest. We present a study along gestation of the baseline's fluctuations, in relation to fractal and nonlinear properties, to assess these fluctuations according with time-varying attracting levels introduced by maturing regulatory mechanisms. A low-risk pregnancy was studied weekly from the 17th to 38th week of gestation during long-term recording sessions at night (>6 h). Fetal averaged pulse rate samples and corresponding baseline series were obtained from raw abdominal ECG ambulatory data. The fractal properties of these series were evaluated by applying detrended fluctuation analysis. The baseline series were also explored to evaluate nonlinear properties and time ordering by applying the scaling magnitude and sign analyses. Our main findings are that the baseline shows fractal and even nonlinear anticorrelated fluctuations. This condition was specially the case before mid-gestation, as revealed by α values near to unit, yet becoming significantly more complex after 30 weeks of gestation as indicated by α(mag) values >0.5. The structured (i.e. not random) fluctuations and particular nonlinear changes that we found thus suggest that the baseline provides on itself information concerning the functional integration of cardiac regulatory mechanisms.

ST analysis of fetal electrocardiography in labor

Since its introduction more than 40 years ago, electronic fetal monitoring has become widely used for intrapartum surveillance to determine fetal wellbeing in labor. Although fetal hypoxia and acidosis are reflected in changes in fetal heart rate, there is no evidence that cardiotocography has been effective in reducing neonatal morbidity related to fetal distress occurring during labor. Indeed the specificity of this tool is poor and in many instances the incorporation of electronic fetal monitoring into intrapartum care has merely led to an increase in medical intervention rather than an improvement in neonatal outcome. Fetal electrocardiography (ECG) analysis provides an additional method for assessing the response of the fetus to hypoxia and in particular to the development of metabolic acidosis. ST changes in the fetal ECG can be quantified with computational analysis, reducing subjective interpretation that has been problematic with traditional electronic fetal monitoring. Formal algorithms indicating appropriate points for intervention in labor have been designed. The fetal ECG has been shown to be a useful adjunct to traditional electronic fetal monitoring in several randomized controlled trials with evidence of reduced rates of neonatal encephalopathy and reduced rates of obstetric intervention.