Open access intrapartum CTG database

Integration of clinical features in a computerized cardiotocography system to predict severe newborn acidemia

BACKGROUND

Cardiotocography (CTG), used during labor to assess fetal wellbeing, is subject to interobserver variability. Computerized CTG is a promising tool to improve fetal hypoxia detection.

OBJECTIVE

To assess if adding clinical features improves the performance of a computerized CTG system to predict severe newborn acidemia (blood cord pH below 7.05).

METHODS

A retrospective multicentric database was built using the data from two sources (the open-source CTU-UHB database and the data from Beaujon university hospital). Four CTG features were extracted from the fetal heart rate (FHR) signal (minimum and maximum value of the baseline, area covered by the accelerations and decelerations). Clinical features were also collected. Severe fetal acidemia was defined by arterial pH < 7.05 on umbilical cord sample. Risk factors for severe acidemia were sought by comparing cases with severe newborn acidemia to the rest of the cohort. We evaluated the accuracy of the model using both CTG and clinical features using area under the curve (AUC) in a cross-center, cross-validation approach.

RESULTS

The datasets contained 1264 cases including 100 cases with severe acidemia. In univariate analysis, hypertensive disorders and other clinical features showed no significant difference, except for meconium-stained amniotic fluid (p = 0.03). Multivariate analysis revealed that a high deceleration area (OR = 1.09 [1.04--1.11]) and apparition of meconium amniotic fluid increased the risk of newborn acidemia (OR = 2.10[1.24-3.49]). In a k-fold cross-validation approach, DeepCTG®1.5 reached an AUC of 0.77, compared to 0.74 when using CTG features only.

CONCLUSION

The CTG features have a good accuracy to predict severe newborn acidemia, confirming existing literature. Integrating clinical features tends to enhance the accuracy. Further research will aim at using more advanced machine learning models to combine the features more efficiently.

Intrapartum electronic fetal heart rate monitoring to predict acidemia at birth with the use of deep learning

BACKGROUND

Electronic fetal monitoring is used in most US hospital births but has significant limitations in achieving its intended goal of preventing intrapartum hypoxic-ischemic injury. Novel deep learning techniques can improve complex data processing and pattern recognition in medicine.

OBJECTIVE

This study aimed to apply deep learning approaches to develop and validate a model to predict fetal acidemia from electronic fetal monitoring data.

STUDY DESIGN

The database was created using intrapartum electronic fetal monitoring data from 2006 to 2020 from a large, multisite academic health system. Data were divided into training and testing sets with equal distribution of acidemic cases. Several different deep learning architectures were explored. The primary outcome was umbilical artery acidemia, which was investigated at 4 clinically meaningful thresholds: 7.20, 7.15, 7.10, and 7.05, along with base excess. The receiver operating characteristic curves were generated with the area under the receiver operating characteristic assessed to determine the performance of the models. External validation was performed using a publicly available Czech database of electronic fetal monitoring data.

RESULTS

A total of 124,777 electronic fetal monitoring files were available, of which 77,132 had <30% missingness in the last 60 minutes of the electronic fetal monitoring tracing. Of these, 21,041 were matched to a corresponding umbilical cord gas result, of which 10,182 were time-stamped within 30 minutes of the last electronic fetal monitoring reading and composed the final dataset. The prevalence rates of the outcomes in the data were 20.9% with a pH of <7.2, 9.1% with a pH of <7.15, 3.3% with a pH of <7.10, and 1.3% with a pH of <7.05. The best performing model achieved an area under the receiver operating characteristic of 0.85 at a pH threshold of <7.05. When predicting the joint outcome of both pH of <7.05 and base excess of less than -10 meq/L, an area under the receiver operating characteristic of 0.89 was achieved. When predicting both pH of <7.20 and base excess of less than -10 meq/L, an area under the receiver operating characteristic of 0.87 was achieved. At a pH of <7.15 and a positive predictive value of 30%, the model achieved a sensitivity of 90% and a specificity of 48%.

CONCLUSION

The application of deep learning methods to intrapartum electronic fetal monitoring analysis achieves promising performance in predicting fetal acidemia. This technology could help improve the accuracy and consistency of electronic fetal monitoring interpretation.

DeepCTG® 2.0: Development and validation of a deep learning model to detect neonatal acidemia from cardiotocography during labor

Cardiotocography (CTG) is the main tool available to detect neonatal acidemia during delivery. Presently, obstetricians and midwives primarily rely on visual interpretation, leading to a significant intra-observer variability. In this paper, we build and evaluate a convolutional neural network to detect neonatal acidemia from the CTG signals during delivery on a multicenter database with 27662 cases in five centers, including 3457 and 464 cases of moderate and severe neonatal acidemia respectively (defined by a fetal pH at birth between 7.05 and 7.20, and lower than 7.05 respectively). To use all the available records, the convolutional layers are pretrained on a task which consists in predicting several features known to be associated with neonatal acidemia from the raw CTG signals. In a cross-center evaluation, the AUC varies from 0.74 to 0.83 between the centers for the detection of severe acidemia, showing the ability of deep learning models to generalize from one dataset to the other and paving the way for more accurate models trained on larger databases. The model can still be significantly improved, by adding clinical variables to account for risk factors of acidemia that may not appear in the CTG signals. Further research will also be led to integrate the model in a tool that could assist humans in the interpretation of CTG.

A Pragmatic Approach to Fetal Monitoring via Cardiotocography Using Feature Elimination and Hyperparameter Optimization

Cardiotocography (CTG) is used to assess the health of the fetus during birth or antenatally in the third trimester. It concurrently detects the maternal uterine contractions (UC) and fetal heart rate (FHR). Fetal distress, which may require therapeutic intervention, can be diagnosed using baseline FHR and its reaction to uterine contractions. Using CTG, a pragmatic machine learning strategy based on feature reduction and hyperparameter optimization was suggested in this study to classify the various fetal states (Normal, Suspect, Pathological). An application of this strategy can be a decision support tool to manage pregnancies. On a public dataset of 2126 CTG recordings, the model was assessed using various standard CTG dataset specific and relevant classifiers. The classifiers' accuracy was improved by the proposed method. The model accuracy was increased to 97.20% while using Random Forest (best classifier). Practically speaking, the model was able to correctly predict 100% of all pathological cases and 98.8% of all normal cases in the dataset. The proposed model was also implemented on another public CTG dataset having 552 CTG signals, resulting in a 97.34% accuracy. If integrated with telemedicine, this proposed model could also be used for long-distance "stay at home" fetal monitoring in high-risk pregnancies.

A multimodal deep learning-based algorithm for specific fetal heart rate events detection

OBJECTIVES

This study aims to develop a multimodal deep learning-based algorithm for detecting specific fetal heart rate (FHR) events, to enhance automatic monitoring and intelligent assessment of fetal well-being.

METHODS

We analyzed FHR and uterine contraction signals by combining various feature extraction techniques, including morphological features, heart rate variability features, and nonlinear domain features, with deep learning algorithms. This approach enabled us to classify four specific FHR events (bradycardia, tachycardia, acceleration, and deceleration) as well as four distinct deceleration patterns (early, late, variable, and prolonged deceleration). We proposed a multi-model deep neural network and a pre-fusion deep learning model to accurately classify the multimodal parameters derived from Cardiotocography signals.

RESULTS

These accuracy metrics were calculated based on expert-labeled data. The algorithm achieved a classification accuracy of 96.2 % for acceleration, 94.4 % for deceleration, 90.9 % for tachycardia, and 85.8 % for bradycardia. Additionally, it achieved 67.0 % accuracy in classifying the four distinct deceleration patterns, with 80.9 % accuracy for late deceleration and 98.9 % for prolonged deceleration.

CONCLUSIONS

The proposed multimodal deep learning algorithm serves as a reliable decision support tool for clinicians, significantly improving the detection and assessment of specific FHR events, which are crucial for fetal health monitoring.

Sample Entropy Computation on Signals with Missing Values

Sample entropy embeds time series into m-dimensional spaces and estimates entropy based on the distances between points in these spaces. However, when samples can be considered as missing or invalid, defining distance in the embedding space becomes problematic. Preprocessing techniques, such as deletion or interpolation, can be employed as a solution, producing time series without missing or invalid values. While deletion ignores missing values, interpolation replaces them using approximations based on neighboring points. This paper proposes a novel approach for the computation of sample entropy when values are considered as missing or invalid. The proposed algorithm accommodates points in the m-dimensional space and handles them there. A theoretical and experimental comparison of the proposed algorithm with deletion and interpolation demonstrates several advantages over these other two approaches. Notably, the deviation of the expected sample entropy value for the proposed methodology consistently proves to be lowest one.

A review of deep learning-based information fusion techniques for multimodal medical image classification

Multimodal medical imaging plays a pivotal role in clinical diagnosis and research, as it combines information from various imaging modalities to provide a more comprehensive understanding of the underlying pathology. Recently, deep learning-based multimodal fusion techniques have emerged as powerful tools for improving medical image classification. This review offers a thorough analysis of the developments in deep learning-based multimodal fusion for medical classification tasks. We explore the complementary relationships among prevalent clinical modalities and outline three main fusion schemes for multimodal classification networks: input fusion, intermediate fusion (encompassing single-level fusion, hierarchical fusion, and attention-based fusion), and output fusion. By evaluating the performance of these fusion techniques, we provide insight into the suitability of different network architectures for various multimodal fusion scenarios and application domains. Furthermore, we delve into challenges related to network architecture selection, handling incomplete multimodal data management, and the potential limitations of multimodal fusion. Finally, we spotlight the promising future of Transformer-based multimodal fusion techniques and give recommendations for future research in this rapidly evolving field.

Automatic classification of fetal heart rate based on a multi-scale LSTM network

Introduction

Fetal heart rate monitoring during labor can aid healthcare professionals in identifying alterations in the heart rate pattern. However, discrepancies in guidelines and obstetrician expertise present challenges in interpreting fetal heart rate, including failure to acknowledge findings or misinterpretation. Artificial intelligence has the potential to support obstetricians in diagnosing abnormal fetal heart rates.

Methods

Employ preprocessing techniques to mitigate the effects of missing signals and artifacts on the model, utilize data augmentation methods to address data imbalance. Introduce a multi-scale long short-term memory neural network trained with a variety of time-scale data for automatically classifying fetal heart rate. Carried out experimental on both single and multi-scale models.

Results

The results indicate that multi-scale LSTM models outperform regular LSTM models in various performance metrics. Specifically, in the single models tested, the model with a sampling rate of 10 exhibited the highest classification accuracy. The model achieves an accuracy of 85.73%, a specificity of 85.32%, and a precision of 85.53% on CTU-UHB dataset. Furthermore, the area under the receiver operating curve of 0.918 suggests that our model demonstrates a high level of credibility.

Discussion

Compared to previous research, our methodology exhibits superior performance across various evaluation metrics. By incorporating alternative sampling rates into the model, we observed improvements in all performance indicators, including ACC (85.73% vs. 83.28%), SP (85.32% vs. 82.47%), PR (85.53% vs. 82.84%), recall (86.13% vs. 84.09%), F1-score (85.79% vs. 83.42%), and AUC(0.9180 vs. 0.8667). The limitations of this research include the limited consideration of pregnant women's clinical characteristics and disregard the potential impact of varying gestational weeks.

Construction of a comprehensive fetal monitoring database for the study of perinatal hypoxic ischemic encephalopathy

This article describes the methods used to build a large-scale database of more than 250,000 electronic fetal monitoring (EFM) records linked to a comprehensive set of clinical information about the infant, the mother, the pregnancy, labor, and outcome. The database can be used to investigate how birth outcome is related to clinical and EFM features. The main steps involved in building the database were: (1) Acquiring the raw EFM recording and clinical records for each birth. (2) Assigning each birth to an objectively defined outcome class that included normal, acidosis, and hypoxic-ischemic encephalopathy. (3) Removing all personal health information from the EFM recordings and clinical records. (4) Preprocessing the deidentified EFM records to eliminate duplicates, reformat the signals, combine signals from different sensors, and bridge gaps to generate signals in a format that can be readily analyzed. (5) Post-processing the repaired EFM recordings to extract key features of the fetal heart rate, uterine activity, and their relations. (6) Populating a database that links the clinical information, EFM records, and EFM features to support easy querying and retrieval. •A multi-step process is required to build a comprehensive database linking electronic temporal fetal monitoring signals to a comprehensive set of clinical information about the infant, the mother, the pregnancy, labor, and outcome.•The current database documents more than 250,000 births including almost 4,000 acidosis and 400 HIE cases. This represents more than 80% of the births that occurred in 15 Northern California Kaiser Permanente Hospitals between 2011-2019. This is a valuable resource for studying the factors predictive of outcome.•The signal processing code and schemas for the database are freely available. The database will not be permitted to leave Kaiser firewalls, but a process is in place to allow interested investigators to access it.

Rapid detection of fetal compromise using input length invariant deep learning on fetal heart rate signals

Standard clinical practice to assess fetal well-being during labour utilises monitoring of the fetal heart rate (FHR) using cardiotocography. However, visual evaluation of FHR signals can result in subjective interpretations leading to inter and intra-observer disagreement. Therefore, recent studies have proposed deep-learning-based methods to interpret FHR signals and detect fetal compromise. These methods have typically focused on evaluating fixed-length FHR segments at the conclusion of labour, leaving little time for clinicians to intervene. In this study, we propose a novel FHR evaluation method using an input length invariant deep learning model (FHR-LINet) to progressively evaluate FHR as labour progresses and achieve rapid detection of fetal compromise. Using our FHR-LINet model, we obtained approximately 25% reduction in the time taken to detect fetal compromise compared to the state-of-the-art multimodal convolutional neural network while achieving 27.5%, 45.0%, 56.5% and 65.0% mean true positive rate at 5%, 10%, 15% and 20% false positive rate respectively. A diagnostic system based on our approach could potentially enable earlier intervention for fetal compromise and improve clinical outcomes.

Intelligent classification of cardiotocography based on a support vector machine and convolutional neural network: Multiscene research

OBJECTIVE

To propose a computerized system utilizing multiscene analysis based on a support vector machine (SVM) and convolutional neural network (CNN) to assess cardiotocography (CTG) intelligently.

METHODS

We retrospectively collected 2542 CTG records of singleton pregnancies delivered at the maternity ward of the First Affiliated Hospital of Xi'an Jiaotong University from October 10, 2020, to August 7, 2021. CTG records were divided into five categories (baseline, variability, acceleration, deceleration, and normality). Apart from the category of normality, the other four different categories of abnormal data correspond to four scenes. Each scene was divided into training and testing sets at 9:1 or 7:3. We used three computer algorithms (dynamic threshold, SVM, and CNN) to learn and optimize the system. Accuracy, sensitivity, and specificity were performed to evaluate performance.

RESULTS

The global accuracy, sensitivity, and specificity of the system were 93.88%, 93.06%, and 94.33%, respectively. In acceleration and deceleration scenes, when the convolution kernel was 3, the test data set reached the highest performance.

CONCLUSION

The multiscene research model using SVM and CNN is a potential effective tool to assist obstetricians in classifying CTG intelligently.

Fetal Heart Rate Preprocessing Techniques: A Scoping Review

Monitoring fetal heart rate (FHR) through cardiotocography is crucial for the early diagnosis of fetal distress situations, necessitating prompt obstetrical intervention. However, FHR signals are often marred by various contaminants, making preprocessing techniques essential for accurate analysis. This scoping review, following PRISMA-ScR guidelines, describes the preprocessing methods in original research articles on human FHR (or beat-to-beat intervals) signal preprocessing from PubMed and Web of Science, published from their inception up to May 2021. From the 322 unique articles identified, 54 were included, from which prevalent preprocessing approaches were identified, primarily focusing on the detection and correction of poor signal quality events. Detection usually entailed analyzing deviations from neighboring samples, whereas correction often relied on interpolation techniques. It was also noted that there is a lack of consensus regarding the definition of missing samples, outliers, and artifacts. Trends indicate a surge in research interest in the decade 2011-2021. This review underscores the need for standardizing FHR signal preprocessing techniques to enhance diagnostic accuracy. Future work should focus on applying and evaluating these methods across FHR databases aiming to assess their effectiveness and propose improvements.

Machine learning on cardiotocography data to classify fetal outcomes: A scoping review

INTRODUCTION

Uterine contractions during labour constrict maternal blood flow and oxygen delivery to the developing baby, causing transient hypoxia. While most babies are physiologically adapted to withstand such intrapartum hypoxia, those exposed to severe hypoxia or with poor physiological reserves may experience neurological injury or death during labour. Cardiotocography (CTG) monitoring was developed to identify babies at risk of hypoxia by detecting changes in fetal heart rate (FHR) patterns. CTG monitoring is in widespread use in intrapartum care for the detection of fetal hypoxia, but the clinical utility is limited by a relatively poor positive predictive value (PPV) of an abnormal CTG and significant inter and intra observer variability in CTG interpretation. Clinical risk and human factors may impact the quality of CTG interpretation. Misclassification of CTG traces may lead to both under-treatment (with the risk of fetal injury or death) or over-treatment (which may include unnecessary operative interventions that put both mother and baby at risk of complications). Machine learning (ML) has been applied to this problem since early 2000 and has shown potential to predict fetal hypoxia more accurately than visual interpretation of CTG alone. To consider how these tools might be translated for clinical practice, we conducted a review of ML techniques already applied to CTG classification and identified research gaps requiring investigation in order to progress towards clinical implementation.

MATERIALS AND METHOD

We used identified keywords to search databases for relevant publications on PubMed, EMBASE and IEEE Xplore. We used Preferred Reporting Items for Systematic Review and Meta-Analysis for Scoping Reviews (PRISMA-ScR). Title, abstract and full text were screened according to the inclusion criteria.

RESULTS

We included 36 studies that used signal processing and ML techniques to classify CTG. Most studies used an open-access CTG database and predominantly used fetal metabolic acidosis as the benchmark for hypoxia with varying pH levels. Various methods were used to process and extract CTG signals and several ML algorithms were used to classify CTG. We identified significant concerns over the practicality of using varying pH levels as the CTG classification benchmark. Furthermore, studies needed to be more generalised as most used the same database with a low number of subjects for an ML study.

CONCLUSION

ML studies demonstrate potential in predicting fetal hypoxia from CTG. However, more diverse datasets, standardisation of hypoxia benchmarks and enhancement of algorithms and features are needed for future clinical implementation.

Large-scale analysis of interobserver agreement and reliability in cardiotocography interpretation during labor using an online tool

BACKGROUND

While the effectiveness of cardiotocography in reducing neonatal morbidity is still debated, it remains the primary method for assessing fetal well-being during labor. Evaluating how accurately professionals interpret cardiotocography signals is essential for its effective use. The objective was to evaluate the accuracy of fetal hypoxia prediction by practitioners through the interpretation of cardiotocography signals and clinical variables during labor.

MATERIAL AND METHODS

We conducted a cross-sectional online survey, involving 120 obstetric healthcare providers from several countries. One hundred cases, including fifty cases of fetal hypoxia, were randomly assigned to participants who were invited to predict the fetal outcome (binary criterion of pH with a threshold of 7.15) based on the cardiotocography signals and clinical variables. After describing the participants, we calculated (with a 95% confidence interval) the success rate, sensitivity and specificity to predict the fetal outcome for the whole population and according to pH ranges, professional groups and number of years of experience. Interobserver agreement and reliability were evaluated using the proportion of agreement and Cohen's kappa respectively.

RESULTS

The overall ability to predict a pH level below 7.15 yielded a success rate of 0.58 (95% CI 0.56-0.60), a sensitivity of 0.58 (95% CI 0.56-0.60) and a specificity of 0.63 (95% CI 0.61-0.65). No significant difference in the success rates was observed with respect to profession and number of years of experience. The success rate was higher for the cases with a pH level below 7.05 (0.69) and above 7.20 (0.66) compared to those falling between 7.05 and 7.20 (0.48). The proportion of agreement between participants was good (0.82), with an overall kappa coefficient indicating substantial reliability (0.63).

CONCLUSIONS

The use of an online tool enabled us to collect a large amount of data to analyze how practitioners interpret cardiotocography data during labor. Despite a good level of agreement and reliability among practitioners, the overall accuracy is poor, particularly for cases with a neonatal pH between 7.05 and 7.20. Factors such as profession and experience level do not present notable impact on the accuracy of the annotations. The implementation and use of a computerized cardiotocography analysis software has the potential to enhance the accuracy to detect fetal hypoxia, especially for ambiguous cardiotocography tracings.

Hybrid-FHR: a multi-modal AI approach for automated fetal acidosis diagnosis

BACKGROUND

In clinical medicine, fetal heart rate (FHR) monitoring using cardiotocography (CTG) is one of the most commonly used methods for assessing fetal acidosis. However, as the visual interpretation of CTG depends on the subjective judgment of the clinician, this has led to high inter-observer and intra-observer variability, making it necessary to introduce automated diagnostic techniques.

METHODS

In this study, we propose a computer-aided diagnostic algorithm (Hybrid-FHR) for fetal acidosis to assist physicians in making objective decisions and taking timely interventions. Hybrid-FHR uses multi-modal features, including one-dimensional FHR signals and three types of expert features designed based on prior knowledge (morphological time domain, frequency domain, and nonlinear). To extract the spatiotemporal feature representation of one-dimensional FHR signals, we designed a multi-scale squeeze and excitation temporal convolutional network (SE-TCN) backbone model based on dilated causal convolution, which can effectively capture the long-term dependence of FHR signals by expanding the receptive field of each layer's convolution kernel while maintaining a relatively small parameter size. In addition, we proposed a cross-modal feature fusion (CMFF) method that uses multi-head attention mechanisms to explore the relationships between different modalities, obtaining more informative feature representations and improving diagnostic accuracy.

RESULTS

Our ablation experiments show that the Hybrid-FHR outperforms traditional previous methods, with average accuracy, specificity, sensitivity, precision, and F1 score of 96.8, 97.5, 96, 97.5, and 96.7%, respectively.

CONCLUSIONS

Our algorithm enables automated CTG analysis, assisting healthcare professionals in the early identification of fetal acidosis and the prompt implementation of interventions.

Introducing Artificial Intelligence in Interpretation of Foetal Cardiotocography: Medical Dataset Curation and Preliminary Coding-An Interdisciplinary Project

Artificial intelligence (AI) is gaining increasing interest in the field of medicine because of its capacity to process big data and pattern recognition. Cardiotocography (CTG) is widely used for the assessment of foetal well-being and uterine contractions during pregnancy and labour. It is characterised by inter- and intraobserver variability in interpretation, which depends on the observers' experience. Artificial intelligence (AI)-assisted interpretation could improve its quality and, thus, intrapartal care. Cardiotocography (CTG) raw signals from labouring women were extracted from the database at the University Hospital of Bern between 2006 and 2019. Later, they were matched with the corresponding foetal outcomes, namely arterial umbilical cord pH and 5-min APGAR score. Excluded were deliveries where data were incomplete, as well as multiple births. Clinical data were grouped regarding foetal pH and APGAR score at 5 min after delivery. Physiological foetal pH was defined as 7.15 and above, and a 5-min APGAR score was considered physiologic when reaching ≥7. With these groups, the algorithm was trained to predict foetal hypoxia. Raw data from 19,399 CTG recordings could be exported. This was accomplished by manually searching the patient's identification numbers (PIDs) and extracting the corresponding raw data from each episode. For some patients, only one episode per pregnancy could be found, whereas for others, up to ten episodes were available. Initially, 3400 corresponding clinical outcomes were found for the 19,399 CTGs (17.52%). Due to the small size, this dataset was rejected, and a new search strategy was elaborated. After further matching and curation, 6141 (31.65%) paired data samples could be extracted (cardiotocography raw data and corresponding maternal and foetal outcomes). Of these, half will be used to train artificial intelligence (AI) algorithms, whereas the other half will be used for analysis of efficacy. Complete data could only be found for one-third of the available population. Yet, to our knowledge, this is the most exhaustive and second-largest cardiotocography database worldwide, which can be used for computer analysis and programming. A further enrichment of the database is planned.

Computerised Cardiotocography Analysis for the Automated Detection of Fetal Compromise during Labour: A Review

The measurement and analysis of fetal heart rate (FHR) and uterine contraction (UC) patterns, known as cardiotocography (CTG), is a key technology for detecting fetal compromise during labour. This technology is commonly used by clinicians to make decisions on the mode of delivery to minimise adverse outcomes. A range of computerised CTG analysis techniques have been proposed to overcome the limitations of manual clinician interpretation. While these automated techniques can potentially improve patient outcomes, their adoption into clinical practice remains limited. This review provides an overview of current FHR and UC monitoring technologies, public and private CTG datasets, pre-processing steps, and classification algorithms used in automated approaches for fetal compromise detection. It aims to highlight challenges inhibiting the translation of automated CTG analysis methods from research to clinical application and provide recommendations to overcome them.

Comparison of fetal heart rate baseline estimation by the cardiotocograph network and clinicians: a multidatabase retrospective assessment study

Background

This study aims to compare the fetal heart rate (FHR) baseline predicted by the cardiotocograph network (CTGNet) with that estimated by clinicians.

Material and methods

A total of 1,267 FHR recordings acquired with different electrical fetal monitors (EFM) were collected from five datasets: 84 FHR recordings acquired with F15 EFM (Edan, Shenzhen, China) from the Guangzhou Women and Children's Medical Center, 331 FHR recordings acquired with SRF618B5 EFM (Sanrui, Guangzhou, China), 234 FHR recordings acquired with F3 EFM (Lian-Med, Guangzhou, China) from the NanFang Hospital of Southen Medical University, 552 cardiotocographys (CTG) recorded using STAN S21 and S31 (Neoventa Medical, Mölndal, Sweden) and Avalon FM40 and FM50 (Philips Healthcare, Amsterdam, The Netherlands) from the University Hospital in Brno, Czech Republic, and 66 FHR recordings acquired using Avalon FM50 fetal monitor (Philips Healthcare, Amsterdam, The Netherlands) at St Vincent de Paul Hospital (Lille, France). Each FHR baseline was estimated by clinicians and CTGNet, respectively. And agreement between CTGNet and clinicians was evaluated using the kappa statistics, intra-class correlation coefficient, and the limits of agreement.

Results

The number of differences <3 beats per minute (bpm), 3-5 bpm, 5-10 bpm and ≥10 bpm, is 64.88%, 15.94%, 14.44% and 4.74%, respectively. Kappa statistics and intra-class correlation coefficient are 0.873 and 0.969, respectively. Limits of agreement are -6.81 and 7.48 (mean difference: 0.36 and standard deviation: 3.64).

Conclusion

An excellent agreement was found between CTGNet and clinicians in the baseline estimation from FHR recordings with different signal loss rates.

The Effect of Fetal Heart Rate Segment Selection on Deep Learning Models for Fetal Compromise Detection

Monitoring the fetal heart rate (FHR) is common practice in obstetric care to assess the risk of fetal compromise. Unfortunately, human interpretation of FHR recordings is subject to inter-observer variability with high false positive rates. To improve the performance of fetal compromise detection, deep learning methods have been proposed to automatically interpret FHR recordings. However, existing deep learning methods typically analyse a fixed-length segment of the FHR recording after removing signal gaps, where the influence of this segment selection process has not been comprehensively assessed. In this work, we develop a novel input length invariant deep learning model to determine the effect of FHR segment selection for detecting fetal compromise. Using this model, we perform five times repeated five-fold cross-validation on an open-access database of 552 FHR recordings and assess model performance for FHR segment lengths between 15 and 60 minutes. We show that the performance after removing signal gaps improves with increasing segment length from 15 minutes (AUC = 0.50) to 60 minutes (AUC = 0.74). Additionally, we demonstrate that using FHR segments without removing signal gaps achieves superior performance across signal lengths from 15 minutes (AUC = 0.68) to 60 minutes (AUC = 0.76). These results show that future works should carefully consider FHR segment selection and that removing signal gaps might contribute to the loss of valuable information.

Machine Learning to Classify Cardiotocography for Fetal Hypoxia Detection

Fetal hypoxia can cause damaging consequences on babies' such as stillbirth and cerebral palsy. Cardiotocography (CTG) has been used to detect intrapartum fetal hypoxia during labor. It is a non-invasive machine that measures the fetal heart rate and uterine contractions. Visual CTG suffers inconsistencies in interpretations among clinicians that can delay interventions. Machine learning (ML) showed potential in classifying abnormal CTG, allowing automatic interpretation. In the absence of a gold standard, researchers used various surrogate biomarkers to classify CTG, where some were clinically irrelevant. We proposed using Apgar scores as the surrogate benchmark of babies' ability to recover from birth. Apgar scores measure newborns' ability to recover from active uterine contraction, which measures appearance, pulse, grimace, activity and respiration. The higher the Apgar score, the healthier the baby is.We employ signal processing methods to pre-process and extract validated features of 552 raw CTG. We also included CTG-specific characteristics as outlined in the NICE guidelines. We employed ML techniques using 22 features and measured performances between ML classifiers. While we found that ML can distinguish CTG with low Apgar scores, results for the lowest Apgar scores, which are rare in the dataset we used, would benefit from more CTG data for better performance. We need an external dataset to validate our model for generalizability to ensure that it does not overfit a specific population.Clinical Relevance- This study demonstrated the potential of using a clinically relevant benchmark for classifying CTG to allow automatic early detection of hypoxia to reduce decision-making time in maternity units.

Trustworthy artificial intelligence and ethical design: public perceptions of trustworthiness of an AI-based decision-support tool in the context of intrapartum care

BACKGROUND

Despite the recognition that developing artificial intelligence (AI) that is trustworthy is necessary for public acceptability and the successful implementation of AI in healthcare contexts, perspectives from key stakeholders are often absent from discourse on the ethical design, development, and deployment of AI. This study explores the perspectives of birth parents and mothers on the introduction of AI-based cardiotocography (CTG) in the context of intrapartum care, focusing on issues pertaining to trust and trustworthiness.

METHODS

Seventeen semi-structured interviews were conducted with birth parents and mothers based on a speculative case study. Interviewees were based in England and were pregnant and/or had given birth in the last two years. Thematic analysis was used to analyze transcribed interviews with the use of NVivo. Major recurring themes acted as the basis for identifying the values most important to this population group for evaluating the trustworthiness of AI.

RESULTS

Three themes pertaining to the perceived trustworthiness of AI emerged from interviews: (1) trustworthy AI-developing institutions, (2) trustworthy data from which AI is built, and (3) trustworthy decisions made with the assistance of AI. We found that birth parents and mothers trusted public institutions over private companies to develop AI, that they evaluated the trustworthiness of data by how representative it is of all population groups, and that they perceived trustworthy decisions as being mediated by humans even when supported by AI.

CONCLUSIONS

The ethical values that underscore birth parents and mothers' perceptions of trustworthy AI include fairness and reliability, as well as practices like patient-centered care, the promotion of publicly funded healthcare, holistic care, and personalized medicine. Ultimately, these are also the ethical values that people want to protect in the healthcare system. Therefore, trustworthy AI is best understood not as a list of design features but in relation to how it undermines or promotes the ethical values that matter most to its end users. An ethical commitment to these values when creating AI in healthcare contexts opens up new challenges and possibilities for the design and deployment of AI.

Multimodal Deep Learning for Predicting Adverse Birth Outcomes Based on Early Labour Data

Cardiotocography (CTG) is a widely used technique to monitor fetal heart rate (FHR) during labour and assess the health of the baby. However, visual interpretation of CTG signals is subjective and prone to error. Automated methods that mimic clinical guidelines have been developed, but they failed to improve detection of abnormal traces. This study aims to classify CTGs with and without severe compromise at birth using routinely collected CTGs from 51,449 births at term from the first 20 min of FHR recordings. Three 1D-CNN and LSTM based architectures are compared. We also transform the FHR signal into 2D images using time-frequency representation with a spectrogram and scalogram analysis, and subsequently, the 2D images are analysed using a 2D-CNNs. In the proposed multi-modal architecture, the 2D-CNN and the 1D-CNN-LSTM are connected in parallel. The models are evaluated in terms of partial area under the curve (PAUC) between 0-10% false-positive rate; and sensitivity at 95% specificity. The 1D-CNN-LSTM parallel architecture outperformed the other models, achieving a PAUC of 0.20 and sensitivity of 20% at 95% specificity. Our future work will focus on improving the classification performance by employing a larger dataset, analysing longer FHR traces, and incorporating clinical risk factors.

DeepCTG® 1.0: an interpretable model to detect fetal hypoxia from cardiotocography data during labor and delivery

Introduction

Cardiotocography, which consists in monitoring the fetal heart rate as well as uterine activity, is widely used in clinical practice to assess fetal wellbeing during labor and delivery in order to detect fetal hypoxia and intervene before permanent damage to the fetus. We present DeepCTG® 1.0, a model able to predict fetal acidosis from the cardiotocography signals.

Materials and methods

DeepCTG® 1.0 is based on a logistic regression model fed with four features extracted from the last available 30 min segment of cardiotocography signals: the minimum and maximum values of the fetal heart rate baseline, and the area covered by accelerations and decelerations. Those four features have been selected among a larger set of 25 features. The model has been trained and evaluated on three datasets: the open CTU-UHB dataset, the SPaM dataset and a dataset built in hospital Beaujon (Clichy, France). Its performance has been compared with other published models and with nine obstetricians who have annotated the CTU-UHB cases. We have also evaluated the impact of two key factors on the performance of the model: the inclusion of cesareans in the datasets and the length of the cardiotocography segment used to compute the features fed to the model.

Results

The AUC of the model is 0.74 on the CTU-UHB and Beaujon datasets, and between 0.77 and 0.87 on the SPaM dataset. It achieves a much lower false positive rate (12% vs. 25%) than the most frequent annotation among the nine obstetricians for the same sensitivity (45%). The performance of the model is slightly lower on the cesarean cases only (AUC = 0.74 vs. 0.76) and feeding the model with shorter CTG segments leads to a significant decrease in its performance (AUC = 0.68 with 10 min segments).

Discussion

Although being relatively simple, DeepCTG® 1.0 reaches a good performance: it compares very favorably to clinical practice and performs slightly better than other published models based on similar approaches. It has the important characteristic of being interpretable, as the four features it is based on are known and understood by practitioners. The model could be improved further by integrating maternofetal clinical factors, using more advanced machine learning or deep learning approaches and having a more robust evaluation of the model based on a larger dataset with more pathological cases and covering more maternity centers.

Machine learning-based clinical decision support systems for pregnancy care: A systematic review

BACKGROUND

Clinical decision support systems (CDSSs) can provide various functions and advantages to healthcare delivery. Quality healthcare during pregnancy and childbirth is of vital importance, and machine learning-based CDSSs have shown positive impact on pregnancy care.

OBJECTIVE

This paper aims to investigate what has been done in CDSSs in the context of pregnancy care using machine learning, and what aspects require attention from future researchers.

METHODS

We conducted a systematic review of existing literature following a structured process of literature search, paper selection and filtering, and data extraction and synthesis.

RESULTS

17 research papers were identified on the topic of CDSS development for different aspects of pregnancy care using various machine learning algorithms. We discovered an overall lack of explainability in the proposed models. We also observed a lack of experimentation, external validation and discussion around culture, ethnicity and race from the source data, with most studies using data from a single centre or country, and an overall lack of awareness of applicability and generalisability of the CDSSs regarding different populations. Finally, we found a gap between machine learning practices and CDSS implementation, and an overall lack of user testing.

CONCLUSION

Machine learning-based CDSSs are still under-explored in the context of pregnancy care. Despite the open problems that remain, the few studies that tested a CDSS for pregnancy care reported positive effects, reinforcing the potential of such systems to improve clinical practice. We encourage future researchers to take into consideration the aspects we identified in order for their work to translate into clinical use.

Identifying fetal status with fetal heart rate: Deep learning approach based on long convolution

CTG (Cardiotocography) is an effective tool for fetal status assessment. Clinically, doctors mainly evaluate the health of fetus by observing FHR (fetal heart rate). The rapid development of Artificial Intelligence has led realization of computer-aided CTG technology, Intelligent CTG classification based on FHR is a fundamental component of these technologies. Its implementation can provide doctors with auxiliary decisions. Most of existing FHR classification methods are based on combing different deep learning models, such as CNN (Convolutional Neural Network), LSTM (Long short-term memory) and Transformer. However, these studies ignore the balance of positive and negative samples in dataset and the matching degree between model and FHR classification task, which reduces the classification accuracy. In this paper, we mainly discuss two major problems in previous FHR classification studies: reduce class imbalance and select appropriate convolution kernel. To address above two problems, we propose a data augmentation method based on ECMN (Edge Clipping and Multiscale Noise) to resolve class imbalance. Subsequently, we introduce a one-dimensional long convolutional layer, which use trend area to calculate the appropriate convolution kernel. Based on appropriate convolution kernel, an improved residual structure with attention mechanism named TGLCN (Trend-Guided Long Convolution Network) is proposed to improve FHR classification accuracy. Finally, horizontal and longitudinal experiments show that the TGLCN obtains high classification accuracy and speed of parameter adjustment.

Clinical practice guidelines for intrapartum cardiotocography interpretation: A systematic review

BACKGROUND

Clinical practice guidelines on intrapartum cardiotocography (CTG) interpretation provide structured tools to detect fetal hypoxia. Despite frequent use of different guidelines, little is known about their comparable consistency. We sought to appraise guidelines relevant to intrapartum CTG interpretation and summarise consensus and non-consensus recommendations.

AIMS

To compare existing intrapartum CTG interpretation guidelines.

MATERIALS AND METHODS

We searched PubMed, CINAHL, Cochrane, Embase, guideline databases and websites of guideline development institutions using terms 'cardiotocography', 'electronic fetal/foetal monitoring', and 'guideline' or equivalent term. The search was restricted to English-language articles published between January 1980 and January 2023 and excluded animal studies. The initial search yielded 2128 articles with 1253 unique citations. Guidelines were included if they: used English as the reporting language; included CTG interpretation criteria or guidelines as a primary objective; were published or updated since 1980; and were the most recently updated publications when multiple versions were identified.

RESULTS

Nineteen studies were considered for full review, and 13 met inclusion criteria. Two reviewers independently assessed guideline quality using the AGREE II instrument, and synthesised consensus and non-consensus recommendations using the content analysis approach. Most guidelines employed a three-tier interpretation framework. There were significant differences between the guidelines for relative importance of key CTG features such as accelerations, decelerations and variability, with respect to the outcome of fetal hypoxia.

CONCLUSIONS

There are significant differences between key intrapartum CTG interpretation guidelines currently being used. Greater consistency is needed across CTG interpretation guidelines to improve the quality of data, clinical governance, monitoring of outcomes, and to support future developments.

A machine learning pipeline to classify foetal heart rate deceleration with optimal feature set

Deceleration is considered a commonly practised means to assess Foetal Heart Rate (FHR) through visual inspection and interpretation of patterns in Cardiotocography (CTG). The precision of deceleration classification relies on the accurate estimation of corresponding event points (EP) from the FHR and the Uterine Contraction Pressure (UCP). This work proposes a deceleration classification pipeline by comparing four machine learning (ML) models, namely, Multilayer Perceptron (MLP), Random Forest (RF), Naïve Bayes (NB), and Simple Logistics Regression. Towards an automated classification of deceleration from EP using the pipeline, it systematically compares three approaches to create feature sets from the detected EP: (1) a novel fuzzy logic (FL)-based approach, (2) expert annotation by clinicians, and (3) calculated using National Institute of Child Health and Human Development guidelines. The classification results were validated using different popular statistical metrics, including receiver operating characteristic curve, intra-class correlation coefficient, Deming regression, and Bland-Altman Plot. The highest classification accuracy (97.94%) was obtained with MLP when the EP was annotated with the proposed FL approach compared to RF, which obtained 63.92% with the clinician-annotated EP. The results indicate that the FL annotated feature set is the optimal one for classifying deceleration from FHR.

Use of artificial intelligence in obstetrics: not quite ready for prime time

Artificial intelligence is finding several applications in healthcare settings. This study aimed to report evidence on the effectiveness of artificial intelligence application in obstetrics. Through a narrative review of literature, we described artificial intelligence use in different obstetrical areas as follows: prenatal diagnosis, fetal heart monitoring, prediction and management of pregnancy-related complications (preeclampsia, preterm birth, gestational diabetes mellitus, and placenta accreta spectrum), and labor. Artificial intelligence seems to be a promising tool to help clinicians in daily clinical activity. The main advantages that emerged from this review are related to the reduction of inter- and intraoperator variability, time reduction of procedures, and improvement of overall diagnostic performance. However, nowadays, the diffusion of these systems in routine clinical practice raises several issues. Reported evidence is still very limited, and further studies are needed to confirm the clinical applicability of artificial intelligence. Moreover, better training of clinicians designed to use these systems should be ensured, and evidence-based guidelines regarding this topic should be produced to enhance the strengths of artificial systems and minimize their limits.

Computerized cardiotocography analysis during labor - A state-of-the-art review

Cardiotocography is defined as the recording of fetal heart rate and uterine contractions and is widely used during labor as a screening tool to determine fetal wellbeing. The visual interpretation of the cardiotocography signals by the practitioners, following common guidelines, is subject to a high interobserver variability, and the efficiency of cardiotocography monitoring is still debated. Since the 1990s, researchers and practitioners work on designing reliable computer-aided systems to assist practitioners in cardiotocography interpretation during labor. Several systems are integrated in the monitoring devices, mostly based on the guidelines, but they have not clearly demonstrated yet their usefulness. In the last decade, the availability of large clinical databases as well as the emergence of machine learning and deep learning methods in healthcare has led to a surge of studies applying those methods to cardiotocography signals analysis. The state-of-the-art systems perform well to detect fetal hypoxia when evaluated on retrospective cohorts, but several challenges remain to be tackled before they can be used in clinical practice. First, the development and sharing of large, open and anonymized multicentric databases of perinatal and cardiotocography data during labor is required to build more accurate systems. Also, the systems must produce interpretable indicators along with the prediction of the risk of fetal hypoxia in order to be appropriated and trusted by practitioners. Finally, common standards should be built and agreed on to evaluate and compare those systems on retrospective cohorts and to validate their use in clinical practice.

Computerized analysis of cardiotocograms in clinical practice and the SisPorto® system thirty-two years after: technological, physiopathological and clinical studies

OBJECTIVES

The objective of this study is to present the why, what and how about computerized analysis of cardiotocograms (cCTG) and the SisPorto system for cCTG.

CONTENT

A narrative review about cCTG and the SisPorto system for cCTG is presented. The meta-analysis of randomized controlled trials (RCT) performed so far have evidenced that cCGT compared to traditional CTG analysis may save time spent in hospital for women, in the antepartum period, and is objective with at least equivalent results in maternal and perinatal outcomes, both in the ante and intrapartum periods. The SisPorto system for cCTG closely follows the FIGO guidelines for fetal monitoring. It may be used both in the ante and intrapartum periods, alone or connected to a central monitoring station, with simultaneous monitoring of fetal and maternal signals, not only in singletons but also in twins. It has been assessed in technical, physiopathological and clinical studies, namely in one large multicentric international RCT during labor and two meta-analysis.

SUMMARY AND OUTLOOK

There is evidence that cCTG may be useful in clinical practice with advantages compared to traditional CTG analysis, although without clear impact on the decrease of preventable maternal and perinatal mortality and morbidity. More studies are warranted, namely on technical improvements and assessment in larger studies in a wider range of clinical settings.

Cardiotocography analysis by empirical dynamic modeling and Gaussian processes

Introduction: During labor, fetal heart rate (FHR) and uterine activity (UA) can be continuously monitored using Cardiotocography (CTG). This is the most widely adopted approach for electronic fetal monitoring in hospitals. Both FHR and UA recordings are evaluated by obstetricians for assessing fetal well-being. Due to the complex and noisy nature of these recordings, the evaluation by obstetricians suffers from high interobserver and intraobserver variability. Machine learning is a field that has seen unprecedented advances in the past two decades and many efforts have been made in computerized analysis of CTG using machine learning methods. However, in the literature, the focus is often only on FHR signals unlike in evaluations performed by obstetricians where the UA signals are also taken into account. Methods: Machine learning is a field that has seen unprecedented advances in the past two decades and many efforts have been made in computerized analysis of CTG using machine learning methods. However, in the literature, the focus is often only on FHR signals unlike in evaluations performed by obstetricians where the UA signals are also taken into account. In this paper, we propose to model intrapartum CTG recordings from a dynamical system perspective using empirical dynamic modeling with Gaussian processes, which is a Bayesian nonparametric approach for estimation of functions. Results and Discussion: In the context of our paper, Gaussian processes are capable for simultaneous estimation of the dimensionality of attractor manifolds and reconstructing of attractor manifolds from time series data. This capacity of Gaussian processes allows for revealing causal relationships between the studied time series. Experimental results on real CTG recordings show that FHR and UA signals are causally related. More importantly, this causal relationship and estimated attractor manifolds can be exploited for several important applications in computerized analysis of CTG recordings including estimating missing FHR samples, recovering burst errors in FHR tracings and characterizing the interactions between FHR and UA signals.

A lightweight fetal distress-assisted diagnosis model based on a cross-channel interactive attention mechanism

Fetal distress is a symptom of fetal intrauterine hypoxia, which is seriously harmful to both the fetus and the pregnant woman. The current primary clinical tool for the assessment of fetal distress is Cardiotocography (CTG). Due to subjective variability, physicians often interpret CTG results inconsistently, hence the need to develop an auxiliary diagnostic system for fetal distress. Although the deep learning-based fetal distress-assisted diagnosis model has a high classification accuracy, the model not only has a large number of parameters but also requires a large number of computational resources, which is difficult to deploy to practical end-use scenarios. Therefore, this paper proposes a lightweight fetal distress-assisted diagnosis network, LW-FHRNet, based on a cross-channel interactive attention mechanism. The wavelet packet decomposition technique is used to convert the one-dimensional fetal heart rate (FHR) signal into a two-dimensional wavelet packet coefficient matrix map as the network input layer to fully obtain the feature information of the FHR signal. With ShuffleNet-v2 as the core, a local cross-channel interactive attention mechanism is introduced to enhance the model's ability to extract features and achieve effective fusion of multichannel features without dimensionality reduction. In this paper, the publicly available database CTU-UHB is used for the network performance evaluation. LW-FHRNet achieves 95.24% accuracy, which meets or exceeds the classification results of deep learning-based models. Additionally, the number of model parameters is reduced many times compared with the deep learning model, and the size of the model parameters is only 0.33 M. The results show that the lightweight model proposed in this paper can effectively aid in fetal distress diagnosis.

Deep learning based fetal distress detection from time frequency representation of cardiotocogram signal using Morse wavelet: research study

Background

Clinically cardiotocography is a technique which is used to monitor and evaluate the level of fetal distress. Even though, CTG is the most widely used device to monitor determine the fetus health, existence of high false positive result from the visual interpretation has a significant contribution to unnecessary surgical delivery or delayed intervention.

Objective

In the current study an innovative computer aided fetal distress diagnosing model is developed by using time frequency representation of FHR signal using generalized Morse wavelet and the concept of transfer learning of pre-trained ResNet 50 deep neural network model.

Method

From the CTG data that is obtained from the only open access CTU-UHB data base only FHR signal is extracted and preprocessed to remove noises and spikes. After preprocessing the time frequency information of FHR signal is extracted by using generalized Morse wavelet and fed to a pre-trained ResNet 50 model which is fine tuned and configured according to the dataset.

Main outcome measures

Sensitivity (Se), specificity (Sp) and accuracy (Acc) of the model adopted from binary confusion matrix is used as outcome measures.

Result

After successfully training the model, a comprehensive experimentation of testing is conducted for FHR data for which a recording is made during early stage of labor and last stage of labor. Thus, a promising classification result which is accuracy of 98.7%, sensitivity of 97.0% and specificity 100% are achieved for FHR signal of 1st stage of labor. For FHR recorded in last stage of labor, accuracy of 96.1%, sensitivity of 94.1% and specificity 97.7% are achieved.

Conclusion

The developed model can be used as a decision-making aid system for obstetrician and gynecologist.

A deep feature fusion network for fetal state assessment

CTG (cardiotocography) has consistently been used to diagnose fetal hypoxia. It is susceptible to identifying the average fetal acid-base balance but lacks specificity in recognizing prenatal acidosis and neurological impairment. CTG plays a vital role in intrapartum fetal state assessment, which can prevent severe organ damage if fetal hypoxia is detected earlier. In this paper, we propose a novel deep feature fusion network (DFFN) for fetal state assessment. First, we extract spatial and temporal information from the fetal heart rate (FHR) signal using a multiscale CNN-BiLSTM network, increasing the features' diversity. Second, the multiscale CNN-BiLSM network and frequently used features are integrated into the deep learning model. The proposed DFFN model combines different features to improve classification accuracy. The multiscale convolutional kernels can identify specific essential information and consider signal's temporal information. The proposed method achieves 61.97%, 73.82%, and 66.93% of sensitivity, specificity, and quality index, respectively, on the public CTU-UHB database. The proposed method achieves the highest QI on the private database, verifying the proposed method's effectiveness and generalization. The proposed DFFN combines the advantages of feature engineering and deep learning models and achieves competitive accuracy in fetal state assessment compared with related works.

Nature inspired method for noninvasive fetal ECG extraction

This paper introduces a novel algorithm for effective and accurate extraction of non-invasive fetal electrocardiogram (NI-fECG). In NI-fECG based monitoring, the useful signal is measured along with other signals generated by the pregnant women's body, especially maternal electrocardiogram (mECG). These signals are more distinct in magnitude and overlap in time and frequency domains, making the fECG extraction extremely challenging. The proposed extraction method combines the Grey wolf algorithm (GWO) with sequential analysis (SA). This innovative combination, forming the GWO-SA method, optimises the parameters required to create a template that matches the mECG, which leads to an accurate elimination of the said signal from the input composite signal. The extraction system was tested on two databases consisting of real signals, namely, Labour and Pregnancy. The databases used to test the algorithms are available on a server at the generalist repositories (figshare) integrated with Matonia et al. (Sci Data 7(1):1-14, 2020). The results show that the proposed method extracts the fetal ECG signal with an outstanding efficacy. The efficacy of the results was evaluated based on accurate detection of the fQRS complexes. The parameters used to evaluate are as follows: accuracy (ACC), sensitivity (SE), positive predictive value (PPV), and F1 score. Due to the stochastic nature of the GWO algorithm, ten individual runs were performed for each record in the two databases to assure stability as well as repeatability. Using these parameters, for the Labour dataset, we achieved an average ACC of 94.60%, F1 of 96.82%, SE of 97.49%, and PPV of 98.96%. For the Pregnancy database, we achieved an average ACC of 95.66%, F1 of 97.44%, SE of 98.07%, and PPV of 97.44%. The obtained results show that the fHR related parameters were determined accurately for most of the records, outperforming the other state-of-the-art approaches. The poorer quality of certain signals have caused deviation from the estimated fHR for certain records in the databases. The proposed algorithm is compared with certain well established algorithms, and has proven to be accurate in its fECG extractions.

Multimodal learning for fetal distress diagnosis using a multimodal medical information fusion framework

Cardiotocography (CTG) monitoring is an important medical diagnostic tool for fetal well-being evaluation in late pregnancy. In this regard, intelligent CTG classification based on Fetal Heart Rate (FHR) signals is a challenging research area that can assist obstetricians in making clinical decisions, thereby improving the efficiency and accuracy of pregnancy management. Most existing methods focus on one specific modality, that is, they only detect one type of modality and inevitably have limitations such as incomplete or redundant source domain feature extraction, and poor repeatability. This study focuses on modeling multimodal learning for Fetal Distress Diagnosis (FDD); however, exists three major challenges: unaligned multimodalities; failure to learn and fuse the causality and inclusion between multimodal biomedical data; modality sensitivity, that is, difficulty in implementing a task in the absence of modalities. To address these three issues, we propose a Multimodal Medical Information Fusion framework named MMIF, where the Category Constrained-Parallel ViT model (CCPViT) was first proposed to explore multimodal learning tasks and address the misalignment between multimodalities. Based on CCPViT, a cross-attention-based image-text joint component is introduced to establish a Multimodal Representation Alignment Network model (MRAN), explore the deep-level interactive representation between cross-modal data, and assist multimodal learning. Furthermore, we designed a simple-structured FDD test model based on the highly modal alignment MMIF, realizing task delegation from multimodal model training (image and text) to unimodal pathological diagnosis (image). Extensive experiments, including model parameter sensitivity analysis, cross-modal alignment assessment, and pathological diagnostic accuracy evaluation, were conducted to show our models' superior performance and effectiveness.

Digital signal extraction approach for cardiotocography image

BACKGROUND AND OBJECTIVE

Cardiotocography, commonly called CTG, has become an indispensable auxiliary examination in obstetrics. Generally, CTG is provided in the form of a report, so the fetal heart rate and uterine contraction signals have to be extracted from the CTG images. However, most studies focused on reading data for a single curve, and the influence of complex backgrounds was usually not considered.

METHODS

An efficient signal extraction method was proposed for the binary CTG images with complex backgrounds. Firstly, the images' background grids and symbol noise were removed by templates. Then a morphological method was used to fill breakpoints of curves. Moreover, the projection map was utilized to localize the area and the starting and ending positions of curves. Subsequently, data of the curves were extracted by column scanning. Finally, the amplitude of the extracted signal was calibrated.

RESULTS

This study had tested 552 CTG images simulated using the CTU-UHB database. The correlation coefficient between the extracted and original signals was 0.9991 ± 0.0030 for fetal heart rate and 0.9904 ± 0.0208 for uterine contraction, and the mean absolute error of fetal heart rate and uterine contraction were 2.4658 ± 1.8446 and 1.8025 ± 0.6155, and the root mean square error of fetal heart rate and uterine contraction were 4.2930 ± 2.9771 and 2.5214 ± 0.9640, respectively. After being validated using 293 clinical authentic CTG images, the extracted signals were remarkably similar to the original counterparts, and no significant differences were observed.

CONCLUSIONS

The proposed method could effectively extract the fetal heart rate and uterine contraction signals from the binary CTG images with complex backgrounds.

Use of Deep Learning to Detect the Maternal Heart Rate and False Signals on Fetal Heart Rate Recordings

We have developed deep learning models for automatic identification of the maternal heart rate (MHR) and, more generally, false signals (FSs) on fetal heart rate (FHR) recordings. The models can be used to preprocess FHR data prior to automated analysis or as a clinical alert system to assist the practitioner. Three models were developed and used to detect (i) FSs on the MHR channel (the FSMHR model), (ii) the MHR and FSs on the Doppler FHR sensor (the FSDop model), and (iii) FSs on the scalp ECG channel (the FSScalp model). The FSDop model was the most useful because FSs are far more frequent on the Doppler FHR channel. All three models were based on a multilayer, symmetric, GRU, and were trained on data recorded during the first and second stages of delivery. The FSMHR and FSDop models were also trained on antepartum recordings. The training dataset contained 1030 expert-annotated periods (mean duration: 36 min) from 635 recordings. In an initial evaluation of routine clinical practice, 30 fully annotated recordings for each sensor type (mean duration: 5 h for MHR and Doppler sensors, and 3 h for the scalp ECG sensor) were analyzed. The sensitivity, positive predictive value (PPV) and accuracy were respectively 62.20%, 87.1% and 99.90% for the FSMHR model, 93.1%, 95.6% and 99.68% for the FSDop model, and 44.6%, 87.2% and 99.93% for the FSScalp model. We built a second test dataset with a more solid ground truth by selecting 45 periods (lasting 20 min, on average) on which the Doppler FHR and scalp ECG signals were recorded simultaneously. Using scalp ECG data, the experts estimated the true FHR value more reliably and thus annotated the Doppler FHR channel more precisely. The models achieved a sensitivity of 53.3%, a PPV of 62.4%, and an accuracy of 97.29%. In comparison, two experts (blinded to the scalp ECG data) respectively achieved a sensitivity of 15.7%, a PPV of 74.3%, and an accuracy of 96.91% and a sensitivity of 60.7%, a PPV of 83.5% and an accuracy of 98.24%. Hence, the models performed at expert level (better than one expert and worse than the other), although a well-trained expert with good knowledge of FSs could probably do better in some cases. The models and datasets have been included in the Fetal Heart Rate Morphological Analysis open-source MATLAB toolbox and can be used freely for research purposes.

Development of a comprehensive database for research on foetal acidosis

OBJECTIVE

To develop a research database for mother-and-child clinical and laboratory data and digital foetal heart rate (FHR) recordings.

METHODS

The Base Bien Naître (BBN) database was derived from a single-centre health data warehouse. It contains exhaustive data on all parturients with a singleton pregnancy, a vaginal or caesarean delivery in labour with a cephalic presentation after at least 37 weeks of amenorrhea, and a live birth between February 1st, 2011, and December 31st, 2018. On arrival in the delivery room, the FHR was recorded digitally for at least 30 min. A cord blood sample was always taken in order to obtain arterial pH (pHa). More than 6,000 recordings were analyzed visually for the risk of foetal acidosis and classified into five groups (according to the French College of Gynaecologists and Obstetricians (CNGOF) classification) or three groups (according to the International Federation of Gynaecology and Obstetrics (FIGO) classification).

RESULTS

Of the 16,089 files in the health data warehouse, 11,026 were complete and met the BBN's inclusion criteria. The FHR digital recordings were of good quality, with low signal loss (median [interquartile range]: 7.0% [4.3;10.9]) and a median recording time of 304 min [190;438]). In 3.7% of the children, the pHa was below 7.10. We selected a subset of 6115 records with good-quality FHR recordings over 120 min and reliable cord blood gas data: 692 (11.3%) had at least a significant risk of acidosis (according to the CNGOF classification), and 1638 (26.8%) were at least suspicious (according to the FIGO classification).

CONCLUSION

The BBN database has been designed as a searchable tool with data reuse. It currently contains over 11,000 records with comprehensive data.

Fully Automatic Classification of Cardiotocographic Signals with 1D-CNN and Bi-directional GRU

Prenatal fetal monitoring, which can monitor the growth and health of the fetus, is vital for pregnant women before delivery. During pregnancy, it is essential to classify whether the fetus is abnormal, which helps physicians carry out early intervention to avoid fetal heart hypoxia and even death. Fetal heart rate and uterine contraction signals obtained by fetal heart monitoring equipment are essential to estimate fetal health status. In this paper, we pre-process the obtained data set and enhance them using Hermite interpolation on the abnormal classification in the samples. We use the 1D-CNN and GRU hybrid models to extract the abstract features of fetal heart rate and uterine contraction signals. Several evaluation metrics are used for evaluation, and the accuracy is 96 %, while the sensitivity is 95 %, and the specificity is 96 %. The experiments show the effectiveness of the proposed method, which can provide physicians and users with more stable, efficient, and convenient diagnosis and decision support.

Heart rate markers for prediction of fetal acidosis in an experimental study on fetal sheep

To overcome the difficulties in interpreting fetal heart rate (FHR), several tools based on the autonomic nervous system and heart rate variability (HRV) have been developed. The objective of this study was to use FHR and HRV parameters for the prediction of fetal hypoxia. It was an experimental study in the instrumented fetal sheep. Repeated umbilical cord occlusions were performed to achieve severe acidosis. Hemodynamic parameters, ECG, and blood gases were analyzed. The variables used were heart rate baseline, HRV analysis (RMSSD, SDNN, LF, HF, HFnu, Fetal Stress Index (FSI), …), and morphological analysis of decelerations. The gold standard used to classify hypoxia was the fetal arterial pH (pH < 7.10). Different multivariable statistical methods (logistic regression and decision trees) were applied for the detection of acidosis. 21 lambs were instrumented. A total of 130 pairs of FHR/fetal pH analysis were obtained of which 29 in the acidosis group and 101 in the non-acidosis group. After logistic regression model with bootstrap resampling and stepwise backward selection, only one variable was selected, FSI. The AUC of FSI alone in this model was 0.81 with a sensitivity of 0.66, specificity of 0.88, PPV of 0.61, and NPV of 0.90 considering a threshold of 68. Decision trees with CHAID and CART algorithms showed a sensitivity of 0.48 and 0.59, respectively, and a specificity of 0.94 for both. All employed methods identified HRV variables as the most predictive of acidosis. The primary variables selected automatically were those from the HRV. Supporting the use of FHRV measures for the screening of fetal acidosis during labour is interesting.

BOOST ENSEMBLE LEARNING FOR CLASSIFICATION OF CTG SIGNALS

During the process of childbirth, fetal distress caused by hypoxia can lead to various abnormalities. Cardiotocography (CTG), which consists of continuous recording of the fetal heart rate (FHR) and uterine contractions (UC), is routinely used for classifying the fetuses as hypoxic or non-hypoxic. In practice, we face highly imbalanced data, where the hypoxic fetuses are significantly underrepresented. We propose to address this problem by boost ensemble learning, where for learning, we use the distribution of classification error over the dataset. We then iteratively select the most informative majority data samples according to this distribution. In our work, in addition to addressing the imbalanced problem, we also experimented with features that are not commonly used in obstetrics. We extracted a large number of statistical features of fetal heart tracings and uterine activity signals and used only the most informative ones. For classification, we implemented several methods: Random Forest, AdaBoost, k-Nearest Neighbors, Support Vector Machine, and Decision Trees. The paper provides a comparison in the performance of these methods on fetal heart rate tracings available from a public database. Our results show that most applied methods improved their performances considerably when boost ensemble was used.

UNSUPERVISED CLUSTERING AND ANALYSIS OF CONTRACTION-DEPENDENT FETAL HEART RATE SEGMENTS

The computer-aided interpretation of fetal heart rate (FHR) and uterine contraction (UC) has not been developed well enough for wide use in delivery rooms. The main challenges still lie in the lack of unclear and nonstandard labels for cardiotocography (CTG) recordings, and the timely prediction of fetal state during monitoring. Rather than traditional supervised approaches to FHR classification, this paper demonstrates a way to understand the UC-dependent FHR responses in an unsupervised manner. In this work, we provide a complete method for FHR-UC segment clustering and analysis via the Gaussian process latent variable model, and density-based spatial clustering. We map the UC-dependent FHR segments into a space with a visual dimension and propose a trajectory-based FHR interpretation method. Three metrics of FHR trajectory are defined and an open-access CTG database is used for testing the proposed method.

IMPROVING PHASE-RECTIFIED SIGNAL AVERAGING FOR FETAL HEART RATE ANALYSIS

Low umbilical artery pH is a marker for neonatal acidosis and is associated with an increased risk for neonatal complications. The phase-rectified signal averaging (PRSA) features have demonstrated superior discriminatory or diagnostic ability and good interpretability in many biomedical applications including fetal heart rate analysis. However, the performance of PRSA method is sensitive to values of the selected parameters which are usually either chosen based on a grid search or empirically in the literature. In this paper, we examine PRSA method through the lens of dynamical systems theory and reveal the intrinsic connection between state space reconstruction and PRSA. From this perspective, we then introduce a new feature that can better characterize dynamical systems comparing with PRSA. Our experimental results on an open-access intrapartum Cardiotocography database demonstrate that the proposed feature outperforms state-of-the-art PRSA features in pH-based fetal heart rate analysis.

On designing a biosignal-based fetal state assessment system: A systematic mapping study

BACKGROUND AND OBJECTIVE

The patterns present in biosignals, such as fetal heart rate (FHR), are valuable indicators of fetal well-being. In designing biosignal analysis systems, the variety of approaches and technology usage impairs the decision-making for the fundamental units of the systems. There is a need for an updated overview of studies encompassing the biosignal-based fetal state assessment systems. Therefore, we propose a systematic mapping study to identify and synthesize recent research regarding the building blocks that compose these systems.

METHODS

We followed well-established guidelines to perform a systematic mapping of studies regarding the building-blocks that compose the fetal state assessment systems and published between January 2016 and January 2021. A search string was determined based on the mapping questions and the PI (population and intervention) divisions. The search string was applied in digital libraries covering the fields of computer science, engineering, and medical informatics. Then, we applied the forward snowballing technique to complement the resulting set. This process resulted in 75 primary studies selected from a total of 871 papers.

RESULTS

Selected studies were classified according to the publication types, systems design stages, datasets, and predictive capabilities. The results revealed that (i) The majority of the selected studies refer to the method as a type of publication and there is a lack of validation studies; (ii) The CTU-UHB was the most frequent biosignal-based dataset and UCI-CTG was the most frequent feature-based data; (iii) The selected studies made use of the system design stages alone or in a mixed-mode.

CONCLUSION

The results indicated that the well-established classification models achieved competitive results compared with the state-of-the-art methods in data-constrained system designs. Moreover, we identified the need for validation studies in the clinical environment.