Machine Learning in Fetal Cardiology: What to Expect

Artificial intelligence for cervical cancer screening: Scoping review, 2009-2022

BACKGROUND

The intersection of artificial intelligence (AI) with cancer research is increasing, and many of the advances have focused on the analysis of cancer images.

OBJECTIVES

To describe and synthesize the literature on the diagnostic accuracy of AI in early imaging diagnosis of cervical cancer following Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR).

SEARCH STRATEGY

Arksey and O'Malley methodology was used and PubMed, Scopus, and Google Scholar databases were searched using a combination of English and Spanish keywords.

SELECTION CRITERIA

Identified titles and abstracts were screened to select original reports and cross-checked for overlap of cases.

DATA COLLECTION AND ANALYSIS

A descriptive summary was organized by the AI algorithm used, total of images analyzed, data source, clinical comparison criteria, and diagnosis performance.

MAIN RESULTS

We identified 32 studies published between 2009 and 2022. The primary sources of images were digital colposcopy, cervicography, and mobile devices. The machine learning/deep learning (DL) algorithms applied in the articles included support vector machine (SVM), random forest classifier, k-nearest neighbors, multilayer perceptron, C4.5, Naïve Bayes, AdaBoost, XGboots, conditional random fields, Bayes classifier, convolutional neural network (CNN; and variations), ResNet (several versions), YOLO+EfficientNetB0, and visual geometry group (VGG; several versions). SVM and DL methods (CNN, ResNet, VGG) showed the best diagnostic performances, with an accuracy of over 97%.

CONCLUSION

We concluded that the use of AI for cervical cancer screening has increased over the years, and some results (mainly from DL) are very promising. However, further research is necessary to validate these findings.

Applications of artificial intelligence-powered prenatal diagnosis for congenital heart disease

Artificial intelligence (AI) has made significant progress in the medical field in the last decade. The AI-powered analysis methods of medical images and clinical records can now match the abilities of clinical physicians. Due to the challenges posed by the unique group of fetuses and the dynamic organ of the heart, research into the application of AI in the prenatal diagnosis of congenital heart disease (CHD) is particularly active. In this review, we discuss the clinical questions and research methods involved in using AI to address prenatal diagnosis of CHD, including imaging, genetic diagnosis, and risk prediction. Representative examples are provided for each method discussed. Finally, we discuss the current limitations of AI in prenatal diagnosis of CHD, namely Volatility, Insufficiency and Independence (VII), and propose possible solutions.

A deep learning framework for noninvasive fetal ECG signal extraction

Introduction: The availability of proactive techniques for health monitoring is essential to reducing fetal mortality and avoiding complications in fetal wellbeing. In harsh circumstances such as pandemics, earthquakes, and low-resource settings, the incompetence of many healthcare systems worldwide in providing essential services, especially for pregnant women, is critical. Being able to continuously monitor the fetus in hospitals and homes in a direct and fast manner is very important in such conditions. Methods: Monitoring the health of the baby can potentially be accomplished through the computation of vital bio-signal measures using a clear fetal electrocardiogram (ECG) signal. The aim of this study is to develop a framework to detect and identify the R-peaks of the fetal ECG directly from a 12 channel abdominal composite signal. Thus, signals were recorded noninvasively from 70 pregnant (healthy and with health conditions) women with no records of fetal abnormalities. The proposed model employs a recurrent neural network architecture to robustly detect the fetal ECG R-peaks. Results: To test the proposed framework, we performed both subject-dependent (5-fold cross-validation) and independent (leave-one-subject-out) tests. The proposed framework achieved average accuracy values of 94.2% and 88.8%, respectively. More specifically, the leave-one-subject-out test accuracy was 86.7% during the challenging period of vernix caseosa layer formation. Furthermore, we computed the fetal heart rate from the detected R-peaks, and the demonstrated results highlight the robustness of the proposed framework. Discussion: This work has the potential to cater to the critical industry of maternal and fetal healthcare as well as advance related applications.

Pulmonary vascular reactivity in growth restricted fetuses using computational modelling and machine learning analysis of fetal Doppler waveforms

The aim of this study was to investigate the pulmonary vasculature in baseline conditions and after maternal hyperoxygenation in growth restricted fetuses (FGR). A prospective cohort study of singleton pregnancies including 97 FGR and 111 normally grown fetuses was carried out. Ultrasound Doppler of the pulmonary vessels was obtained at 24-37 weeks of gestation and data were acquired before and after oxygen administration. After, Machine Learning (ML) and a computational model were used on the Doppler waveforms to classify individuals and estimate pulmonary vascular resistance (PVR). Our results showed lower mean velocity time integral (VTI) in the main pulmonary and intrapulmonary arteries in baseline conditions in FGR individuals. Delta changes of the main pulmonary artery VTI and intrapulmonary artery pulsatility index before and after hyperoxygenation were significantly greater in FGR when compared with controls. Also, ML identified two clusters: A (including 66% controls and 34% FGR) with similar Doppler traces over time and B (including 33% controls and 67% FGR) with changes after hyperoxygenation. The computational model estimated the ratio of PVR before and after maternal hyperoxygenation which was closer to 1 in cluster A (cluster A 0.98 ± 0.33 vs cluster B 0.78 ± 0.28, p = 0.0156). Doppler ultrasound allows the detection of significant changes in pulmonary vasculature in most FGR at baseline, and distinct responses to hyperoxygenation. Future studies are warranted to assess its potential applicability in the clinical management of FGR.

Anatomical structure characterization of fetal ultrasound images using texture-based segmentation technique via an interactive MATLAB application

OBJECTIVE

To describe the texture characteristics in several anatomical structures within fetal ultrasound images by applying an image segmentation technique through an application developed in MATLAB mathematical processing software.

METHODS

Prospective descriptive observational study with an analytical component. 2D fetal ultrasound images were acquired in patients admitted to the Maternal Fetal Medicine Unit of the Hospital de San José, Bogotá-Colombia. These images were loaded into the developed application to carry out the segmentation and characterization stages by means of 23 numerical texture descriptors. The data were analyzed with central tendency measures and through an embedding process and Euclidean distance.

RESULTS

Forty ultrasound images were included, characterizing 54 structures of the fetal placenta, skull, thorax, and abdomen. By embedding the descriptors, the differentiation of biologically known structures as distinct was achieved, as well as the non-differentiation of similar structures, evidenced using 2D and 3D graphs and numerical data with statistical significance.

CONCLUSION

The texture characterization of the labeled structures in fetal ultrasound images through the numerical descriptors allows the accurate discrimination of these structures.

Simulator, machine learning, and artificial intelligence: Time has come to assist prenatal ultrasound diagnosis

In this Commentary authors investigated and extended the role of simulator in assisting obstetric sonographers in training program. The interconnection of different digitalized technologies such as digital data, artificial neuronal and convolutional networks, machine and deep learning, telemedicine, and output are discussed and contribute to the generation of artificial intelligence.

A Clinical Approach to Semiautomated Three-Dimensional Fetal Brain Biometry-Comparing the Strengths and Weaknesses of Two Diagnostic Tools: 5DCNS+TM and SonoCNSTM

(1) Objective: We aimed to evaluate the accuracy and efficacy of AI-assisted biometric measurements of the fetal central nervous system (CNS) by comparing two semiautomatic postprocessing tools. We further aimed to discuss the additional value of semiautomatically generated sagittal and coronal planes of the CNS. (2) Methods: Three-dimensional (3D) volumes were analyzed with two semiautomatic software tools, 5DCNS+™ and SonoCNS™. The application of 5DCNS+™ results in nine planes (axial, coronal and sagittal) displayed in a single template; SonoCNS™ depicts three axial cutting sections. The tools were compared regarding automatic biometric measurement accuracy. (3) Results: A total of 129 fetuses were included for final analysis. Our data indicate that, in terms of the biometric quantification of head circumference (HC), biparietal diameter (BPD), transcerebellar diameter (TCD) and cisterna magna (CM), the accuracy of SonoCNS™ was higher with respect to the manual measurement of an experienced examiner compared to 5DCNS+™, whereas it was the other way around regarding the diameter of the posterior horn of the lateral ventricle (Vp). The inclusion of four orthogonal coronal views in 5DCNS+™ gives valuable information regarding spatial arrangements, particularly of midline structures. (4) Conclusions: Both tools were able to ease assessment of the intracranial anatomy, highlighting the additional value of automated algorithms in clinical use. SonoCNS™ showed a superior accuracy of plane reconstruction and biometry, but volume reconstruction using 5DCNS+™ provided more detailed information, which is needed for an entire neurosonogram as suggested by international guidelines.

A multicenter study on two-stage transfer learning model for duct-dependent CHDs screening in fetal echocardiography

Duct-dependent congenital heart diseases (CHDs) are a serious form of CHD with a low detection rate, especially in underdeveloped countries and areas. Although existing studies have developed models for fetal heart structure identification, there is a lack of comprehensive evaluation of the long axis of the aorta. In this study, a total of 6698 images and 48 videos are collected to develop and test a two-stage deep transfer learning model named DDCHD-DenseNet for screening critical duct-dependent CHDs. The model achieves a sensitivity of 0.973, 0.843, 0.769, and 0.759, and a specificity of 0.985, 0.967, 0.956, and 0.759, respectively, on the four multicenter test sets. It is expected to be employed as a potential automatic screening tool for hierarchical care and computer-aided diagnosis. Our two-stage strategy effectively improves the robustness of the model and can be extended to screen for other fetal heart development defects.

The Role of Artificial Intelligence in Prediction, Risk Stratification, and Personalized Treatment Planning for Congenital Heart Diseases

This narrative review delves into the potential of artificial intelligence (AI) in predicting, stratifying risk, and personalizing treatment planning for congenital heart disease (CHD). CHD is a complex condition that affects individuals across various age groups. The review highlights the challenges in predicting risks, planning treatments, and prognosticating long-term outcomes due to CHD's multifaceted nature, limited data, ethical concerns, and individual variabilities. AI, with its ability to analyze extensive data sets, presents a promising solution. The review emphasizes the need for larger, diverse datasets, the integration of various data sources, and the analysis of longitudinal data. Prospective validation in real-world clinical settings, interpretability, and the importance of human clinical expertise are also underscored. The ethical considerations surrounding privacy, consent, bias, monitoring, and human oversight are examined. AI's implications include improved patient outcomes, cost-effectiveness, and real-time decision support. The review aims to provide a comprehensive understanding of AI's potential for revolutionizing CHD management and highlights the significance of collaboration and transparency to address challenges and limitations.

Machine learning applied in maternal and fetal health: a narrative review focused on pregnancy diseases and complications

Introduction

Machine learning (ML) corresponds to a wide variety of methods that use mathematics, statistics and computational science to learn from multiple variables simultaneously. By means of pattern recognition, ML methods are able to find hidden correlations and accomplish accurate predictions regarding different conditions. ML has been successfully used to solve varied problems in different areas of science, such as psychology, economics, biology and chemistry. Therefore, we wondered how far it has penetrated into the field of obstetrics and gynecology.

Aim

To describe the state of art regarding the use of ML in the context of pregnancy diseases and complications.

Methodology

Publications were searched in PubMed, Web of Science and Google Scholar. Seven subjects of interest were considered: gestational diabetes mellitus, preeclampsia, perinatal death, spontaneous abortion, preterm birth, cesarean section, and fetal malformations.

Current state

ML has been widely applied in all the included subjects. Its uses are varied, the most common being the prediction of perinatal disorders. Other ML applications include (but are not restricted to) biomarker discovery, risk estimation, correlation assessment, pharmacological treatment prediction, drug screening, data acquisition and data extraction. Most of the reviewed articles were published in the last five years. The most employed ML methods in the field are non-linear. Except for logistic regression, linear methods are rarely used.

Future challenges

To improve data recording, storage and update in medical and research settings from different realities. To develop more accurate and understandable ML models using data from cutting-edge instruments. To carry out validation and impact analysis studies of currently existing high-accuracy ML models.

Conclusion

The use of ML in pregnancy diseases and complications is quite recent, and has increased over the last few years. The applications are varied and point not only to the diagnosis, but also to the management, treatment, and pathophysiological understanding of perinatal alterations. Facing the challenges that come with working with different types of data, the handling of increasingly large amounts of information, the development of emerging technologies, and the need of translational studies, it is expected that the use of ML continue growing in the field of obstetrics and gynecology.

Application and Progress of Artificial Intelligence in Fetal Ultrasound

Prenatal ultrasonography is the most crucial imaging modality during pregnancy. However, problems such as high fetal mobility, excessive maternal abdominal wall thickness, and inter-observer variability limit the development of traditional ultrasound in clinical applications. The combination of artificial intelligence (AI) and obstetric ultrasound may help optimize fetal ultrasound examination by shortening the examination time, reducing the physician's workload, and improving diagnostic accuracy. AI has been successfully applied to automatic fetal ultrasound standard plane detection, biometric parameter measurement, and disease diagnosis to facilitate conventional imaging approaches. In this review, we attempt to thoroughly review the applications and advantages of AI in prenatal fetal ultrasound and discuss the challenges and promises of this new field.

A Machine Learning Method for Automated Description and Workflow Analysis of First Trimester Ultrasound Scans

Obstetric ultrasound assessment of fetal anatomy in the first trimester of pregnancy is one of the less explored fields in obstetric sonography because of the paucity of guidelines on anatomical screening and availability of data. This paper, for the first time, examines imaging proficiency and practices of first trimester ultrasound scanning through analysis of full-length ultrasound video scans. Findings from this study provide insights to inform the development of more effective user-machine interfaces, of targeted assistive technologies, as well as improvements in workflow protocols for first trimester scanning. Specifically, this paper presents an automated framework to model operator clinical workflow from full-length routine first-trimester fetal ultrasound scan videos. The 2D+t convolutional neural network-based architecture proposed for video annotation incorporates transfer learning and spatio-temporal (2D+t) modelling to automatically partition an ultrasound video into semantically meaningful temporal segments based on the fetal anatomy detected in the video. The model results in a cross-validation A1 accuracy of 96.10% , F1=0.95 , precision =0.94 and recall =0.95 . Automated semantic partitioning of unlabelled video scans (n=250) achieves a high correlation with expert annotations ( ρ = 0.95, p=0.06 ). Clinical workflow patterns, operator skill and its variability can be derived from the resulting representation using the detected anatomy labels, order, and distribution. It is shown that nuchal translucency (NT) is the toughest standard plane to acquire and most operators struggle to localize high-quality frames. Furthermore, it is found that newly qualified operators spend 25.56% more time on key biometry tasks than experienced operators.

Performance evaluation of computer-aided automated master frame selection techniques for fetal echocardiography

PURPOSE

Fetal echocardiography is widely used for the assessment of fetal heart development and detection of congenital heart disease (CHD). Preliminary examination of the fetal heart involves the four-chamber view which indicates the presence of all the four chambers and its structural symmetry. Examination of various cardiac parameters is generally done using the clinically selected diastole frame. This largely depends on the expertise of the sonographer and is prone to intra- and interobservational errors. To overcome this, automated frame selection technique is proposed for the recognition of fetal cardiac chamber from fetal echocardiography.

METHODS

Three techniques have been proposed in this research study to automate the process of determining the frame referred as "Master Frame" that can be used for the measurement of the cardiac parameters. The first method uses frame similarity measures (FSM) for the determination of the master frame from the given cine loop ultrasonic sequences. FSM makes use of similarity measures such as correlation, structural similarity index (SSIM), peak signal to noise ratio (PSNR), and mean square error (MSE) to identify the cardiac cycle, and all the frames in one cardiac cycle are superimposed to form the master frame. The final master frame is obtained by considering the average of the master frame obtained using each similarity measure. The second method uses averaging of ± 20% from the midframes (AMF). The third method uses averaging of all the frames (AAF) of the cine loop sequence. Both diastole and master frames have been annotated by the clinical experts, and their ground truths are compared for validation. No segmentation techniques have been used to avoid the variability of the performance of various segmentation techniques. All the proposed schemes were evaluated using six fidelity metrics such as Dice coefficient, Jaccard ratio, Hausdorff distance, structural similarity index, mean absolute error, and Pratt figure of merit.

RESULTS

The three proposed techniques were tested on the frames extracted from 95 ultrasound cine loop sequences between 19 and 32 weeks of gestation. The feasibility of the techniques was determined by the computation of fidelity metrics between the master frame derived and the diastole frame chosen by the clinical experts. The FSM-based identified master frame found to closely match with manually chosen diastole frame and also ensures statistically significant. The method also detects automatically the cardiac cycle. The resultant master frame obtained through AMF though found to be identical to that of the diastole frame, the size of the chambers found to be reduced that can lead to inaccurate chamber measurement. The master frame obtained through AAF was not found to be identical to that of clinical diastole frame.

CONCLUSION

It can be concluded that the frame similarity measure (FSM)-based master frame can be introduced in the clinical routine for segmentation followed by cardiac chamber measurements. Such automated master frame selection also overcomes the manual intervention of earlier reported techniques in the literature. The fidelity metrics assessment further confirms the suitability of proposed master frame for automated fetal chamber recognition.

Multi-centre deep learning for placenta segmentation in obstetric ultrasound with multi-observer and cross-country generalization

The placenta is crucial to fetal well-being and it plays a significant role in the pathogenesis of hypertensive pregnancy disorders. Moreover, a timely diagnosis of placenta previa may save lives. Ultrasound is the primary imaging modality in pregnancy, but high-quality imaging depends on the access to equipment and staff, which is not possible in all settings. Convolutional neural networks may help standardize the acquisition of images for fetal diagnostics. Our aim was to develop a deep learning based model for classification and segmentation of the placenta in ultrasound images. We trained a model based on manual annotations of 7,500 ultrasound images to identify and segment the placenta. The model's performance was compared to annotations made by 25 clinicians (experts, trainees, midwives). The overall image classification accuracy was 81%. The average intersection over union score (IoU) reached 0.78. The model's accuracy was lower than experts' and trainees', but it outperformed all clinicians at delineating the placenta, IoU = 0.75 vs 0.69, 0.66, 0.59. The model was cross validated on 100 2nd trimester images from Barcelona, yielding an accuracy of 76%, IoU 0.68. In conclusion, we developed a model for automatic classification and segmentation of the placenta with consistent performance across different patient populations. It may be used for automated detection of placenta previa and enable future deep learning research in placental dysfunction.

Organ-specific learning curves of sonographers performing first-trimester anatomical screening and impact of score-based evaluation on ultrasound image quality

INTRODUCTION

First-trimester anatomical screening (FTAS) by ultrasound has been introduced in many countries as screening for aneuploidies, but also as early screening for fetal structural abnormalities. While a lot of emphasis has been put on the detection rates of FTAS, little is known about the performance of quality control programs and the sonographers' learning curve for FTAS. The aims of the study were to evaluate the performance of a score-based quality control system for the FTAS and to assess the learning curves of sonographers by evaluating the images of the anatomical planes that were part of the FTAS protocol.

METHODS

Between 2012-2015, pregnant women opting for the combined test in the North-Netherlands were also invited to participate in a prospective cohort study extending the ultrasound investigation to include a first-trimester ultrasound performed according to a protocol. All anatomical planes included in the protocol were documented by pictures stored for each examination in logbooks. The logbooks of six sonographers were independently assessed by two fetal medicine experts. For each sonographer, logbooks of examination 25-50-75 and 100 plus four additional randomly selected logbooks were scored for correct visualization of 12 organ-system planes. A plane specific score of at least 70% was considered sufficient. The intra-class correlation coefficient (ICC), was used to measure inter-assessor agreement for the cut-off scores. Organ-specific learning curves were defined by single-cumulative sum (CUSUM) analysis.

RESULTS

Sixty-four logbooks were assessed. Mean duration of the scan was 22 ± 6 minutes and mean gestational age was 12+6 weeks. In total 57% of the logbooks graded as sufficient. Most sufficient scores were obtained for the fetal skull (88%) and brain (70%), while the lowest scores were for the face (29%) and spine (38%). Five sonographers showed a learning curve for the skull and the stomach, four for the brain and limbs, three for the bladder and kidneys, two for the diaphragm and abdominal wall and one for the heart and spine and none for the face and neck.

CONCLUSION

Learning curves for FTAS differ per organ system and per sonographer. Although score-based evaluation can validly assess image quality, more dynamic approaches may better reflect clinical performance.

Abnormalities of the Width of the Four-Chamber View and the Area, Length, and Width of the Ventricles to Identify Fetuses at High-Risk for D-Transposition of the Great Arteries and Tetralogy of Fallot

OBJECTIVES

The prenatal detection of D-Transposition of the great arteries (D-TGA) and tetralogy of Fallot (TOF) has been reported to be less than 50% to as high as 77% when adding the outflow tracts to the four-chamber screening protocol. Because many examiners still struggle with the outflow tract examination, this study evaluated whether changes in the size and shape of the heart in the 4CV as well as the ventricles occurred in fetuses with D-TGA and TOF could be used to screen for these malformations.

METHODS

Forty-four fetuses with the pre-and post-natal diagnosis of D-TGA and 44 with TOF were evaluated between 19 and 36 weeks of gestation in which the 4CV was imaged. Measurements of the end-diastolic width, length, area, and global sphericity index were measured for the four-chamber view and the right and left ventricles. Using z-score computed values, logistic regression was performed between the 88 study and 200 control fetuses using the hierarchical forward selection protocol.

RESULTS

Logistic regression identified 10 variables that correctly classified 83/88 of fetuses with TOF and TGA, for a sensitivity of 94%. Six of 200 normal controls were incorrectly classified for a false-positive rate of 3%. The area under the receiver operator classification curve was 98.1%. The true positive rate for D-TGA was 93.2%, with a false-negative rate to 6.8%. The true positive rate for TOF was 95.5%, with a false negative rate of 4.5%.

CONCLUSIONS

Measurements of the 4CV and of the RV and LV may help identify fetuses at risk for D-TGA or TOF.

A second look at intrapartum fetal surveillance and future directions

Intrapartum fetal surveillance aims to predict significant fetal hypoxia and institute timely intervention to avoid fetal injury, and do so without unnecessary operative delivery of fetuses at no risk of intrapartum hypoxia. However, the configuration and application of current clinical guidelines inadvertently undermine these aims because of persistent failure to incorporate increased understanding of fetal cardiovascular physiology and adaptations to oxygen deprivation, advances in signal acquisition/processing, and related technologies. Consequently, the field on intrapartum fetal surveillance is stuck in rudimentary counts of the fetal R-R intervals and visual assessment of very common, but nonspecific fetal heart decelerations and fetal heart rate variability. The present authors argue that the time has come to move away from classifications of static morphological appearances of FHR decelerations, which do not assist the thinking clinician in understanding how the fetus defends itself and compensates for intrapartum hypoxic ischaemic insults or the patterns that suggest progressive loss of compensation. We also reappraise some of the controversial aspects of intrapartum fetal surveillance in modern obstetric practice, the current state of flux in training and certification, and contemplate the future of the field particularly in the context of the emerging role of artificial intelligence.

Learning deep architectures for the interpretation of first-trimester fetal echocardiography (LIFE) - a study protocol for developing an automated intelligent decision support system for early fetal echocardiography

BACKGROUND

Congenital Heart Disease represents the most frequent fetal malformation. The lack of prenatal identification of congenital heart defects can have adverse consequences for the neonate, while a correct prenatal diagnosis of specific cardiac anomalies improves neonatal care neurologic and surgery outcomes. Sonographers perform prenatal diagnosis manually during the first or second-trimester scan, but the reported detection rates are low. This project's primary objective is to develop an Intelligent Decision Support System that uses two-dimensional video files of cardiac sweeps obtained during the standard first-trimester fetal echocardiography (FE) to signal the presence/absence of previously learned key features.

METHODS

The cross-sectional study will be divided into a training part of the machine learning approaches and the testing phase on previously unseen frames and eventually on actual video scans. Pregnant women in their 12-13 + 6 weeks of gestation admitted for routine first-trimester anomaly scan will be consecutively included in a two-year study, depending on the availability of the experienced sonographers in early fetal cardiac imaging involved in this research. The Data Science / IT department (DSIT) will process the key planes identified by the sonographers in the two- dimensional heart cine loop sweeps: four-chamber view, left and right ventricular outflow tracts, three vessels, and trachea view. The frames will be grouped into the classes representing the plane views, and then different state-of-the- art deep-learning (DL) pre-trained algorithms will be tested on the data set. The sonographers will validate all the intermediary findings at the frame level and the meaningfulness of the video labeling.

DISCUSSION

FE is feasible and efficient during the first trimester. Still, the continuous training process is impaired by the lack of specialists or their limited availability. Therefore, in our study design, the sonographer benefits from a second opinion provided by the developed software, which may be very helpful, especially if a more experienced colleague is unavailable. In addition, the software may be implemented on the ultrasound device so that the process could take place during the live examination.

TRIAL REGISTRATION

The study is registered under the name "Learning deep architectures for the Interpretation of Fetal Echocardiography (LIFE)", project number 408PED/2020, project code PN-III-P2-2.1-PED-2019.

TRIAL REGISTRATION

ClinicalTrials.gov , unique identifying number NCT05090306, date of registration 30.10.2020.

Multiomics, artificial intelligence, and precision medicine in perinatology

Technological advances in omics evaluation, bioinformatics, and artificial intelligence have made us rethink ways to improve patient outcomes. Collective quantification and characterization of biological data including genomics, epigenomics, metabolomics, and proteomics is now feasible at low cost with rapid turnover. Significant advances in the integration methods of these multiomics data sets by machine learning promise us a holistic view of disease pathogenesis and yield biomarkers for disease diagnosis and prognosis. Using machine learning tools and algorithms, it is possible to integrate multiomics data with clinical information to develop predictive models that identify risk before the condition is clinically apparent, thus facilitating early interventions to improve the health trajectories of the patients. In this review, we intend to update the readers on the recent developments related to the use of artificial intelligence in integrating multiomic and clinical data sets in the field of perinatology, focusing on neonatal intensive care and the opportunities for precision medicine. We intend to briefly discuss the potential negative societal and ethical consequences of using artificial intelligence in healthcare. We are poised for a new era in medicine where computational analysis of biological and clinical data sets will make precision medicine a reality. IMPACT: Biotechnological advances have made multiomic evaluations feasible and integration of multiomics data may provide a holistic view of disease pathophysiology. Artificial Intelligence and machine learning tools are being increasingly used in healthcare for diagnosis, prognostication, and outcome predictions. Leveraging artificial intelligence and machine learning tools for integration of multiomics and clinical data will pave the way for precision medicine in perinatology.

A review on deep-learning algorithms for fetal ultrasound-image analysis

Deep-learning (DL) algorithms are becoming the standard for processing ultrasound (US) fetal images. A number of survey papers in the field is today available, but most of them are focusing on a broader area of medical-image analysis or not covering all fetal US DL applications. This paper surveys the most recent work in the field, with a total of 153 research papers published after 2017. Papers are analyzed and commented from both the methodology and the application perspective. We categorized the papers into (i) fetal standard-plane detection, (ii) anatomical structure analysis and (iii) biometry parameter estimation. For each category, main limitations and open issues are presented. Summary tables are included to facilitate the comparison among the different approaches. In addition, emerging applications are also outlined. Publicly-available datasets and performance metrics commonly used to assess algorithm performance are summarized, too. This paper ends with a critical summary of the current state of the art on DL algorithms for fetal US image analysis and a discussion on current challenges that have to be tackled by researchers working in the field to translate the research methodology into actual clinical practice.

Applications of artificial intelligence in obstetrics

Artificial intelligence, which has been applied as an innovative technology in multiple fields of healthcare, analyzes large amounts of data to assist in disease prediction, prevention, and diagnosis, as well as in patient monitoring. In obstetrics, artificial intelligence has been actively applied and integrated into our daily medical practice. This review provides an overview of artificial intelligence systems currently used for obstetric diagnostic purposes, such as fetal cardiotocography, ultrasonography, and magnetic resonance imaging, and demonstrates how these methods have been developed and clinically applied.

Artificial Intelligence in Pediatric Cardiology: A Scoping Review

The evolution of AI and data science has aided in mechanizing several aspects of medical care requiring critical thinking: diagnosis, risk stratification, and management, thus mitigating the burden of physicians and reducing the likelihood of human error. AI modalities have expanded feet to the specialty of pediatric cardiology as well. We conducted a scoping review searching the Scopus, Embase, and PubMed databases covering the recent literature between 2002-2022. We found that the use of neural networks and machine learning has significantly improved the diagnostic value of cardiac magnetic resonance imaging, echocardiograms, computer tomography scans, and electrocardiographs, thus augmenting the clinicians' diagnostic accuracy of pediatric heart diseases. The use of AI-based prediction algorithms in pediatric cardiac surgeries improves postoperative outcomes and prognosis to a great extent. Risk stratification and the prediction of treatment outcomes are feasible using the key clinical findings of each CHD with appropriate computational algorithms. Notably, AI can revolutionize prenatal prediction as well as the diagnosis of CHD using the EMR (electronic medical records) data on maternal risk factors. The use of AI in the diagnostics, risk stratification, and management of CHD in the near future is a promising possibility with current advancements in machine learning and neural networks. However, the challenges posed by the dearth of appropriate algorithms and their nascent nature, limited physician training, fear of over-mechanization, and apprehension of missing the 'human touch' limit the acceptability. Still, AI proposes to aid the clinician tomorrow with precision cardiology, paving a way for extremely efficient human-error-free health care.

Inside the "black box": Embedding clinical knowledge in data-driven machine learning for heart disease diagnosis

Background

Ischemic heart disease (IHD) caused by the narrowing of coronary arteries is a major cause of morbidity and mortality worldwide. Clinical diagnosis involves complex, costly, and potentially invasive procedures.

Objective

To address this problem, we introduce a novel clinical knowledge-enhanced machine learning (ML) pipeline to assist in timely and cost-effective IHD prediction.

Methods

Unlike conventional data-driven "black box" ML approaches, we propose an effective mechanism to engage clinical expertise and gain insight into the "black box" at each stage of model development, including data analysis, preprocessing, selecting the most clinically discriminative features, and model evaluation. One-hot feature encoding is introduced to expose hidden bias and highlight the important elements and features.

Results

Experimental results on the benchmark Cleveland IHD dataset showed that the proposed clinical knowledge-enhanced ML pipeline overperformed state-of-the-art data-driven ML models, using even fewer features. Our model based on one-hot feature encoding and support vector machine achieved the best accuracy of 94.4% and sensitivity 95% by using only 7 discriminative attributes.

Conclusion

We share insights and discuss the effectiveness of incorporating clinical input in machine learning to improve model performance, as well as addressing some practical issues such as data bias and interpretability. We hope this preliminary study on engaging clinical expertise to explore the "black box" would improve the trustworthiness of AI and its potential wider uptake in the medical field.

A Machine Learning-Based Intrauterine Growth Restriction (IUGR) Prediction Model for Newborns

Intrauterine growth restriction (IUGR) is a condition in which the fetal weight is below the 10th percentile for its gestational age. Prenatal exposure to metals can cause a decrease in fetal growth during gestation thereby reducing birth weight. Therefore, the aim of the present study was to develop a machine learning model for early prediction of IUGR. A total of 126 IUGR and 88 appropriate-for-gestational-age (AGA) samples were collected from the Gynecology Department, Safdarjung Hospital, New Delhi. The predictive models were developed using the Weka software. The models developed using all the features gave the highest accuracy of 95.5% with support vector machine (SMO) algorithm and 88.5% with multilayer perceptron (MLP) algorithm. Further, models developed after feature selection using 14 important and statistically significant variables also gave the highest accuracy of 98.5% with SMO algorithm and 99% with Naïve Bayes (NB) algorithm. The study concluded SMO_31, SMO_14, MLP_31, and NB_14 to be the better classifiers for IUGR prediction.

Deep Learning for Improving the Effectiveness of Routine Prenatal Screening for Major Congenital Heart Diseases

Early prenatal screening with an ultrasound (US) can significantly lower newborn mortality caused by congenital heart diseases (CHDs). However, the need for expertise in fetal cardiologists and the high volume of screening cases limit the practically achievable detection rates. Hence, automated prenatal screening to support clinicians is desirable. This paper presents and analyses potential deep learning (DL) techniques to diagnose CHDs in fetal USs. Four convolutional neural network architectures were compared to select the best classifier with satisfactory results. Hence, dense convolutional network (DenseNet) 201 architecture was selected for the classification of seven CHDs, such as ventricular septal defect, atrial septal defect, atrioventricular septal defect, Ebstein's anomaly, tetralogy of Fallot, transposition of great arteries, hypoplastic left heart syndrome, and a normal control. The sensitivity, specificity, and accuracy of the DenseNet201 model were 100%, 100%, and 100%, respectively, for the intra-patient scenario and 99%, 97%, and 98%, respectively, for the inter-patient scenario. We used the intra-patient DL prediction model to validate our proposed model against the prediction results of three expert fetal cardiologists. The proposed model produces a satisfactory result, which means that our model can support expert fetal cardiologists to interpret the decision to improve CHD diagnostics. This work represents a step toward the goal of assisting front-line sonographers with CHD diagnoses at the population level.

A Study on the Association between Korotkoff Sound Signaling and Chronic Heart Failure (CHF) Based on Computer-Assisted Diagnoses

Background

Korotkoff sound (KS) is an important indicator of hypertension when monitoring blood pressure. However, its utility in noninvasive diagnosis of Chronic heart failure (CHF) has rarely been studied.

Purpose

In this study, we proposed a method for signal denoising, segmentation, and feature extraction for KS, and a Bayesian optimization-based support vector machine algorithm for KS classification.

Methods

The acquired KS signal was resampled and denoised to extract 19 energy features, 12 statistical features, 2 entropy features, and 13 Mel Frequency Cepstrum Coefficient (MFCCs) features. A controlled trial based on the VALSAVA maneuver was carried out to investigate the relationship between cardiac function and KS. To classify these feature sets, the K-Nearest Neighbors (KNN), decision tree (DT), Naive Bayes (NB), ensemble (EM) classifiers, and the proposed BO-SVM were employed and evaluated using the accuracy (Acc), sensitivity (Se), specificity (Sp), Precision (Ps), and F1 score (F1).

Results

The ALSAVA maneuver indicated that the KS signal could play an important role in the diagnosis of CHF. Through comparative experiments, it was shown that the best performance of the classifier was obtained by BO-SVM, with Acc (85.0%), Se (85.3%), and Sp (84.6%).

Conclusions

In this study, a method for noise reduction, segmentation, and classification of KS was established. In the measured data set, our method performed well in terms of classification accuracy, sensitivity, and specificity. In light of this, we believed that the methods described in this paper can be applied to the early, noninvasive detection of heart disease as well as a supplementary monitoring technique for the prognosis of patients with CHF.

Towards deep observation: A systematic survey on artificial intelligence techniques to monitor fetus via Ultrasound Images

Several reviews have been conducted regarding artificial intelligence (AI) techniques to improve pregnancy outcomes. But they are not focusing on ultrasound images. This survey aims to explore how AI can assist with fetal growth monitoring via ultrasound image. We reported our findings using the guidelines for PRISMA. We conducted a comprehensive search of eight bibliographic databases. Out of 1269 studies 107 are included. We found that 2D ultrasound images were more popular (88) than 3D and 4D ultrasound images (19). Classification is the most used method (42), followed by segmentation (31), classification integrated with segmentation (16) and other miscellaneous methods such as object-detection, regression, and reinforcement learning (18). The most common areas that gained traction within the pregnancy domain were the fetus head (43), fetus body (31), fetus heart (13), fetus abdomen (10), and the fetus face (10). This survey will promote the development of improved AI models for fetal clinical applications.

•

Artificial intelligence studies to monitor fetal development via ultrasound images

•

Fetal issues categorized based on four categories — general, head, heart, face, abdomen

•

The most used AI techniques are classification, segmentation, object detection, and RL

•

The research and practical implications are included.

Role of Four-Chamber Heart Ultrasound Images in Automatic Assessment of Fetal Heart: A Systematic Understanding

The fetal echocardiogram is useful for monitoring and diagnosing cardiovascular diseases in the fetus in utero. Importantly, it can be used for assessing prenatal congenital heart disease, for which timely intervention can improve the unborn child’s outcomes. In this regard, artificial intelligence (AI) can be used for the automatic analysis of fetal heart ultrasound images. This study reviews nondeep and deep learning approaches for assessing the fetal heart using standard four-chamber ultrasound images. The state-of-the-art techniques in the field are described and discussed. The compendium demonstrates the capability of automatic assessment of the fetal heart using AI technology. This work can serve as a resource for research in the field.

Medical professional enhancement using explainable artificial intelligence in fetal cardiac ultrasound screening (supplementary)

Diagnostic support tools based on artificial intelligence (AI) have exhibited high performance in various medical fields. However, their clinical application remains challenging because of the lack of explanatory power in AI decisions (black box problem), making it difficult to build trust with medical professionals. Nevertheless, visualizing the internal representation of deep neural networks will increase explanatory power and improve the confidence of medical professionals in AI decisions. We propose a novel deep learning-based explainable representation “graph chart diagram” to support fetal cardiac ultrasound screening, which has low detection rates of congenital heart diseases due to the difficulty in mastering the technique. Screening performance improves using this representation from 0.966 to 0.975 for experts, 0.829 to 0.890 for fellows, and 0.616 to 0.748 for residents in the arithmetic mean of area under the curve of a receiver operating characteristic curve. This is the first demonstration wherein examiners used deep learning-based explainable representation to improve the performance of fetal cardiac ultrasound screening, highlighting the potential of explainable AI to augment examiner capabilities.

Intelligent systems in obstetrics and midwifery: Applications of machine learning

INTRODUCTION

Machine learning is increasingly utilized over recent years in order to develop models that represent and solve problems in a variety of domains, including those of obstetrics and midwifery. The aim of this systematic review was to analyze research studies on machine learning and intelligent systems applications in midwifery and obstetrics.

METHODS

A thorough literature review was performed in four electronic databases (PubMed, APA PsycINFO, SCOPUS, ScienceDirect). Only articles that discussed machine learning and intelligent systems applications in midwifery and obstetrics, were considered in this review. Selected articles were critically evaluated as for their relevance and a contextual synthesis was conducted.

RESULTS

Thirty-two articles were included in this systematic review as they met the inclusion and methodological criteria specified in this study. The results suggest that machine learning and intelligent systems have produced successful models and systems in a broad list of midwifery and obstetrics topics, such as diagnosis, pregnancy risk assessment, fetal monitoring, bladder tumor, etc.

CONCLUSIONS

This systematic review suggests that machine learning represents a very promising area of artificial intelligence for the development of practical and highly effective applications that can support human experts, as well the investigation of a wide range of exciting opportunities for further research.

Novel artificial intelligence approach for automatic differentiation of fetal occiput anterior and non-occiput anterior positions during labor

OBJECTIVES

To describe a newly developed machine-learning (ML) algorithm for the automatic recognition of fetal head position using transperineal ultrasound (TPU) during the second stage of labor and to describe its performance in differentiating between occiput anterior (OA) and non-OA positions.

METHODS

This was a prospective cohort study including singleton term (> 37 weeks of gestation) pregnancies in the second stage of labor, with a non-anomalous fetus in cephalic presentation. Transabdominal ultrasound was performed to determine whether the fetal head position was OA or non-OA. For each case, one sonographic image of the fetal head was then acquired in an axial plane using TPU and saved for later offline analysis. Using the transabdominal sonographic diagnosis as the gold standard, a ML algorithm based on a pattern-recognition feed-forward neural network was trained on the TPU images to discriminate between OA and non-OA positions. In the training phase, the model tuned its parameters to approximate the training data (i.e. the training dataset) such that it would identify correctly the fetal head position, by exploiting geometric, morphological and intensity-based features of the images. In the testing phase, the algorithm was blinded to the occiput position as determined by transabdominal ultrasound. Using the test dataset, the ability of the ML algorithm to differentiate OA from non-OA fetal positions was assessed in terms of diagnostic accuracy. The F₁ -score and precision-recall area under the curve (PR-AUC) were calculated to assess the algorithm's performance. Cohen's kappa (κ) was calculated to evaluate the agreement between the algorithm and the gold standard.

RESULTS

Over a period of 24 months (February 2018 to January 2020), at 15 maternity hospitals affiliated to the International Study group on Labor ANd Delivery Sonography (ISLANDS), we enrolled into the study 1219 women in the second stage of labor. On the basis of transabdominal ultrasound, they were classified as OA (n = 801 (65.7%)) or non-OA (n = 418 (34.3%)). From the entire cohort (OA and non-OA), approximately 70% (n = 824) of the patients were assigned randomly to the training dataset and the rest (n = 395) were used as the test dataset. The ML-based algorithm correctly classified the fetal occiput position in 90.4% (357/395) of the test dataset, including 224/246 with OA (91.1%) and 133/149 with non-OA (89.3%) fetal head position. Evaluation of the algorithm's performance gave an F₁ -score of 88.7% and a PR-AUC of 85.4%. The algorithm showed a balanced performance in the recognition of both OA and non-OA positions. The robustness of the algorithm was confirmed by high agreement with the gold standard (κ = 0.81; P < 0.0001).

CONCLUSIONS

This newly developed ML-based algorithm for the automatic assessment of fetal head position using TPU can differentiate accurately, in most cases, between OA and non-OA positions in the second stage of labor. This algorithm has the potential to support not only obstetricians but also midwives and accoucheurs in the clinical use of TPU to determine fetal occiput position in the labor ward. © 2021 International Society of Ultrasound in Obstetrics and Gynecology.

Artificial intelligence in obstetrics

This study reviews recent advances on the application of artificial intelligence for the early diagnosis of various maternal-fetal conditions such as preterm birth and abnormal fetal growth. It is found in this study that various machine learning methods have been successfully employed for different kinds of data capture with regard to early diagnosis of maternal-fetal conditions. With the more popular use of artificial intelligence, ethical issues should also be considered accordingly.

The Use of Artificial Intelligence in Automation in the Fields of Gynaecology and Obstetrics - an Assessment of the State of Play

The long-awaited progress in digitalisation is generating huge amounts of medical data every day, and manual analysis and targeted, patient-oriented evaluation of this data is becoming increasingly difficult or even infeasible. This state of affairs and the associated, increasingly complex requirements for individualised precision medicine underline the need for modern software solutions and algorithms across the entire healthcare system. The utilisation of state-of-the-art equipment and techniques in almost all areas of medicine over the past few years has now indeed enabled automation processes to enter - at least in part - into routine clinical practice. Such systems utilise a wide variety of artificial intelligence (AI) techniques, the majority of which have been developed to optimise medical image reconstruction, noise reduction, quality assurance, triage, segmentation, computer-aided detection and classification and, as an emerging field of research, radiogenomics. Tasks handled by AI are completed significantly faster and more precisely, clearly demonstrated by now in the annual findings of the ImageNet Large-Scale Visual Recognition Challenge (ILSVCR), first conducted in 2015, with error rates well below those of humans. This review article will discuss the potential capabilities and currently available applications of AI in gynaecological-obstetric diagnostics. The article will focus, in particular, on automated techniques in prenatal sonographic diagnostics.

Challenges of Developing Robust AI for Intrapartum Fetal Heart Rate Monitoring

Background: CTG remains the only non-invasive tool available to the maternity team for continuous monitoring of fetal well-being during labour. Despite widespread use and investment in staff training, difficulty with CTG interpretation continues to be identified as a problem in cases of fetal hypoxia, which often results in permanent brain injury. Given the recent advances in AI, it is hoped that its application to CTG will offer a better, less subjective and more reliable method of CTG interpretation. Objectives: This mini-review examines the literature and discusses the impediments to the success of AI application to CTG thus far. Prior randomised control trials (RCTs) of CTG decision support systems are reviewed from technical and clinical perspectives. A selection of novel engineering approaches, not yet validated in RCTs, are also reviewed. The review presents the key challenges that need to be addressed in order to develop a robust AI tool to identify fetal distress in a timely manner so that appropriate intervention can be made. Results: The decision support systems used in three RCTs were reviewed, summarising the algorithms, the outcomes of the trials and the limitations. Preliminary work suggests that the inclusion of clinical data can improve the performance of AI-assisted CTG. Combined with newer approaches to the classification of traces, this offers promise for rewarding future development.

The Diagnostic Yield of Fetal Echocardiography Indications in the Current Era

BACKGROUND

The aim of this study was to examine the diagnostic yield of current fetal echocardiography (FE) indications representing a recent era.

METHODS

FE reports of all pregnancies referred to two provincial FE programs from 2009 to 2018 were examined, identifying the indication for FE (14 categories), gestational age at referral, and whether there was no fetal heart disease (FHD), mild or possible FHD (e.g., simple ventricular septal defect, possible coarctation), or moderate or severe FHD.

RESULTS

Over the study period, there were 19,310 unique FE referrals in Alberta (23.3 ± 5.4 weeks' gestation), including 1,907 (9.9%) with moderate or severe and 654 (3.4%) with mild or possible FHD. The most common referral indications included extracardiac pathology or markers (29.7%), maternal diabetes (18.3%), suspected FHD (17.7%), and family history of heart defects (17.7%). The highest yield for moderate or severe FHD was suspected FHD (41.1%; 95% CI, 39.4%-42.7%), followed by suspected or confirmed genetic disorder (15.4%; 95% CI, 12.6%-18.2%), twins or multiples (10.6%; 95% CI, 8.7%-12.5%), oligohydramnios (8.0%; 95% CI, 4.1%-11.9%), extracardiac pathology or markers (6.4%; 95% CI, 5.8%-7.1%), and heart not well seen (5.8%; 95% CI, 4.0%-7.6%). Lowest yields were observed for maternal diabetes (2.2%; 95% CI, 1.7%-2.7%) and family history of heart defects (1.7%; 95% CI, 1.3%-2.2%). Excluding suspected FHD, with two or more FE indications, all other indications demonstrated significant increases in yield of mild or possible (3.5% vs 1.9%, P < .001) and moderate or severe (7.2% vs 2.9%, P < .001) FHD.

CONCLUSIONS

Suspected FHD provides the highest diagnostic yield of moderate or severe FHD. In contrast, maternal diabetes and family history of heart defects, among the most common referral indications, had diagnostic yields approaching general population risks. Even in the absence of suspected FHD, having two or more referral indications importantly increases the diagnostic yield of all other FE indications.

Prospective Artificial Intelligence to Dissect the Dengue Immune Response and Discover Therapeutics

Dengue virus (DENV) poses a serious threat to global health as the causative agent of dengue fever. The virus is endemic in more than 128 countries resulting in approximately 390 million infection cases each year. Currently, there is no approved therapeutic for treatment nor a fully efficacious vaccine. The development of therapeutics is confounded and hampered by the complexity of the immune response to DENV, in particular to sequential infection with different DENV serotypes (DENV1-5). Researchers have shown that the DENV envelope (E) antigen is primarily responsible for the interaction and subsequent invasion of host cells for all serotypes and can elicit neutralizing antibodies in humans. The advent of high-throughput sequencing and the rapid advancements in computational analysis of complex data, has provided tools for the deconvolution of the DENV immune response. Several types of complex statistical analyses, machine learning models and complex visualizations can be applied to begin answering questions about the B- and T-cell immune responses to multiple infections, antibody-dependent enhancement, identification of novel therapeutics and advance vaccine research.

How Do Machines Learn? Artificial Intelligence as a New Era in Medicine

With an increased number of medical data generated every day, there is a strong need for reliable, automated evaluation tools. With high hopes and expectations, machine learning has the potential to revolutionize many fields of medicine, helping to make faster and more correct decisions and improving current standards of treatment. Today, machines can analyze, learn, communicate, and understand processed data and are used in health care increasingly. This review explains different models and the general process of machine learning and training the algorithms. Furthermore, it summarizes the most useful machine learning applications and tools in different branches of medicine and health care (radiology, pathology, pharmacology, infectious diseases, personalized decision making, and many others). The review also addresses the futuristic prospects and threats of applying artificial intelligence as an advanced, automated medicine tool.

Model-Agnostic Method for Thoracic Wall Segmentation in Fetal Ultrasound Videos

The application of segmentation methods to medical imaging has the potential to create novel diagnostic support models. With respect to fetal ultrasound, the thoracic wall is a key structure on the assessment of the chest region for examiners to recognize the relative orientation and size of structures inside the thorax, which are critical components in neonatal prognosis. In this study, to improve the segmentation performance of the thoracic wall in fetal ultrasound videos, we proposed a novel model-agnostic method using deep learning techniques: the Multi-Frame + Cylinder method (MFCY). The Multi-frame method (MF) uses time-series information of ultrasound videos, and the Cylinder method (CY) utilizes the shape of the thoracic wall. To evaluate the achieved improvement, we performed segmentation using five-fold cross-validation on 538 ultrasound frames in the four-chamber view (4CV) of 256 normal cases using U-net and DeepLabv3+. MFCY increased the mean values of the intersection over union (IoU) of thoracic wall segmentation from 0.448 to 0.493 for U-net and from 0.417 to 0.470 for DeepLabv3+. These results demonstrated that MFCY improved the segmentation performance of the thoracic wall in fetal ultrasound videos without altering the network structure. MFCY is expected to facilitate the development of diagnostic support models in fetal ultrasound by providing further accurate segmentation of the thoracic wall.