Intra- and interobserver variability in fetal ultrasound measurements

Use of focus point for plane acquisition to improve reproducibility in fetal biometry

OBJECTIVE

To assess the reproducibility of ultrasound measurements of fetal biometry using a 'focus point' to assist the acquisition of the relevant plane.

METHODS

This was a study of 80 women with a singleton non-anomalous pregnancy who attended University College London Hospital, London, UK, between 18 and 37 weeks' gestation. Planes to measure head circumference (HC), abdominal circumference (AC) and femur length (FL) were obtained four times by two different sonographers with different levels of experience, who were blinded to one another; the first set of images was obtained with reference to a standard image, and the second set of images was obtained using the focus point technique. The focus point was defined as a unique fetal anatomical landmark in each plane (cavum septi pellucidi for HC, two-thirds of the umbilical vein for AC and one of the two extremities of the diaphysis for FL). Once identified, the focus point was maintained in view while the sonographer rotated the probe along three axes (x, y, z) to acquire the relevant plane. Sonographers were either in training or had > 3000 scans worth of experience. Intra- and interobserver reproducibility were assessed using Bland-Altman plots, and absolute values and percentages for mean difference and 95% limits of agreement (LoA) were reported.

RESULTS

Overall reproducibility was good, with all 95% LoA < 8%. Reproducibility was improved by use of the focus point compared with the standard technique for both intraobserver comparison (95% LoA, < 4% vs < 6%) and interobserver comparison (95% LoA, < 7% vs < 8%). These findings were independent of sonographer seniority and plane acquired.

CONCLUSIONS

Reproducibility of fetal biometry assessment is improved with use of the focus point for plane acquisition, regardless of sonographer experience. We propose that this method should be implemented in clinical practice and training programs in fetal biometry. © 2023 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Artificial Intelligence in Obstetric Anomaly Scan: Heart and Brain

BACKGROUND

The ultrasound scan represents the first tool that obstetricians use in fetal evaluation, but sometimes, it can be limited by mobility or fetal position, excessive thickness of the maternal abdominal wall, or the presence of post-surgical scars on the maternal abdominal wall. Artificial intelligence (AI) has already been effectively used to measure biometric parameters, automatically recognize standard planes of fetal ultrasound evaluation, and for disease diagnosis, which helps conventional imaging methods. The usage of information, ultrasound scan images, and a machine learning program create an algorithm capable of assisting healthcare providers by reducing the workload, reducing the duration of the examination, and increasing the correct diagnosis capability. The recent remarkable expansion in the use of electronic medical records and diagnostic imaging coincides with the enormous success of machine learning algorithms in image identification tasks.

OBJECTIVES

We aim to review the most relevant studies based on deep learning in ultrasound anomaly scan evaluation of the most complex fetal systems (heart and brain), which enclose the most frequent anomalies.

Memory-based unsupervised video clinical quality assessment with multi-modality data in fetal ultrasound

In obstetric sonography, the quality of acquisition of ultrasound scan video is crucial for accurate (manual or automated) biometric measurement and fetal health assessment. However, the nature of fetal ultrasound involves free-hand probe manipulation and this can make it challenging to capture high-quality videos for fetal biometry, especially for the less-experienced sonographer. Manually checking the quality of acquired videos would be time-consuming, subjective and requires a comprehensive understanding of fetal anatomy. Thus, it would be advantageous to develop an automatic quality assessment method to support video standardization and improve diagnostic accuracy of video-based analysis. In this paper, we propose a general and purely data-driven video-based quality assessment framework which directly learns a distinguishable feature representation from high-quality ultrasound videos alone, without anatomical annotations. Our solution effectively utilizes both spatial and temporal information of ultrasound videos. The spatio-temporal representation is learned by a bi-directional reconstruction between the video space and the feature space, enhanced by a key-query memory module proposed in the feature space. To further improve performance, two additional modalities are introduced in training which are the sonographer gaze and optical flow derived from the video. Two different clinical quality assessment tasks in fetal ultrasound are considered in our experiments, i.e., measurement of the fetal head circumference and cerebellar diameter; in both of these, low-quality videos are detected by the large reconstruction error in the feature space. Extensive experimental evaluation demonstrates the merits of our approach.

BabyNet＋＋: Fetal birth weight prediction using biometry multimodal data acquired less than 24 hours before delivery

Accurate prediction of fetal weight at birth is essential for effective perinatal care, particularly in the context of antenatal management, which involves determining the timing and mode of delivery. The current standard of care involves performing a prenatal ultrasound 24 hours prior to delivery. However, this task presents challenges as it requires acquiring high-quality images, which becomes difficult during advanced pregnancy due to the lack of amniotic fluid. In this paper, we present a novel method that automatically predicts fetal birth weight by using fetal ultrasound video scans and clinical data. Our proposed method is based on a Transformer-based approach that combines a Residual Transformer Module with a Dynamic Affine Feature Map Transform. This method leverages tabular clinical data to evaluate 2D+t spatio-temporal features in fetal ultrasound video scans. Development and evaluation were carried out on a clinical set comprising 582 2D fetal ultrasound videos and clinical records of pregnancies from 194 patients performed less than 24 hours before delivery. Our results show that our method outperforms several state-of-the-art automatic methods and estimates fetal birth weight with an accuracy comparable to human experts. Hence, automatic measurements obtained by our method can reduce the risk of errors inherent in manual measurements. Observer studies suggest that our approach may be used as an aid for less experienced clinicians to predict fetal birth weight before delivery, optimizing perinatal care regardless of the available expertise.

Deep learning for estimation of fetal weight throughout the pregnancy from fetal abdominal ultrasound

BACKGROUND

Fetal weight is currently estimated from fetal biometry parameters using heuristic mathematical formulas. Fetal biometry requires measurements of the fetal head, abdomen, and femur. However, this examination is prone to inter- and intraobserver variability because of factors, such as the experience of the operator, image quality, maternal characteristics, or fetal movements. Our study tested the hypothesis that a deep learning method can estimate fetal weight based on a video scan of the fetal abdomen and gestational age with similar performance to the full biometry-based estimations provided by clinical experts.

OBJECTIVE

This study aimed to develop and test a deep learning method to automatically estimate fetal weight from fetal abdominal ultrasound video scans.

STUDY DESIGN

A dataset of 900 routine fetal ultrasound examinations was used. Among those examinations, 800 retrospective ultrasound video scans of the fetal abdomen from 700 pregnant women between 15 6/7 and 41 0/7 weeks of gestation were used to train the deep learning model. After the training phase, the model was evaluated on an external prospectively acquired test set of 100 scans from 100 pregnant women between 16 2/7 and 38 0/7 weeks of gestation. The deep learning model was trained to directly estimate fetal weight from ultrasound video scans of the fetal abdomen. The deep learning estimations were compared with manual measurements on the test set made by 6 human readers with varying levels of expertise. Human readers used standard 3 measurements made on the standard planes of the head, abdomen, and femur and heuristic formula to estimate fetal weight. The Bland-Altman analysis, mean absolute percentage error, and intraclass correlation coefficient were used to evaluate the performance and robustness of the deep learning method and were compared with human readers.

RESULTS

Bland-Altman analysis did not show systematic deviations between readers and deep learning. The mean and standard deviation of the mean absolute percentage error between 6 human readers and the deep learning approach was 3.75%±2.00%. Excluding junior readers (residents), the mean absolute percentage error between 4 experts and the deep learning approach was 2.59%±1.11%. The intraclass correlation coefficients reflected excellent reliability and varied between 0.9761 and 0.9865.

CONCLUSION

This study reports the use of deep learning to estimate fetal weight using only ultrasound video of the fetal abdomen from fetal biometry scans. Our experiments demonstrated similar performance of human measurements and deep learning on prospectively acquired test data. Deep learning is a promising approach to directly estimate fetal weight using ultrasound video scans of the fetal abdomen.

Fetal biometry and amniotic fluid volume assessment end-to-end automation using Deep Learning

Fetal biometry and amniotic fluid volume assessments are two essential yet repetitive tasks in fetal ultrasound screening scans, aiding in the detection of potentially life-threatening conditions. However, these assessment methods can occasionally yield unreliable results. Advances in deep learning have opened up new avenues for automated measurements in fetal ultrasound, demonstrating human-level performance in various fetal ultrasound tasks. Nevertheless, the majority of these studies are retrospective in silico studies, with a limited number including African patients in their datasets. In this study we developed and prospectively assessed the performance of deep learning models for end-to-end automation of fetal biometry and amniotic fluid volume measurements. These models were trained using a newly constructed database of 172,293 de-identified Moroccan fetal ultrasound images, supplemented with publicly available datasets. the models were then tested on prospectively acquired video clips from 172 pregnant people forming a consecutive series gathered at four healthcare centers in Morocco. Our results demonstrate that the 95% limits of agreement between the models and practitioners for the studied measurements were narrower than the reported intra- and inter-observer variability among expert human sonographers for all the parameters under study. This means that these models could be deployed in clinical conditions, to alleviate time-consuming, repetitive tasks, and make fetal ultrasound more accessible in limited-resource environments.

COMFormer: Classification of Maternal-Fetal and Brain Anatomy Using a Residual Cross-Covariance Attention Guided Transformer in Ultrasound

Monitoring the healthy development of a fetus requires accurate and timely identification of different maternal-fetal structures as they grow. To facilitate this objective in an automated fashion, we propose a deep-learning-based image classification architecture called the COMFormer to classify maternal-fetal and brain anatomical structures present in 2-D fetal ultrasound (US) images. The proposed architecture classifies the two subcategories separately: maternal-fetal (abdomen, brain, femur, thorax, mother's cervix (MC), and others) and brain anatomical structures [trans-thalamic (TT), trans-cerebellum (TC), trans-ventricular (TV), and non-brain (NB)]. Our proposed architecture relies on a transformer-based approach that leverages spatial and global features using a newly designed residual cross-variance attention block. This block introduces an advanced cross-covariance attention (XCA) mechanism to capture a long-range representation from the input using spatial (e.g., shape, texture, intensity) and global features. To build COMFormer, we used a large publicly available dataset (BCNatal) consisting of 12 400 images from 1792 subjects. Experimental results prove that COMFormer outperforms the recent CNN and transformer-based models by achieving 95.64% and 96.33% classification accuracy on maternal-fetal and brain anatomy, respectively.

[A pragmatic comparison of fetal biometry curves]

INTRODUCTION

The fetal biometrics charts recommended in France for ultrasound screening include measurements of head circumference (HC), biparietal diameter (BIP), abdominal circumference (AC) and femur length (FL). New international growth standards have been recommended since 2022. The aim of this work is to quantitatively describe the differences between these biometric curves.

METHODS

The biometry curves from the French College for Fetal Ultrasound, OMS and INTERGROWTH-21 are pragmatically compared based on their original quantile regression equations (superposition and quantification of differences in millimeters and in proportion) for different percentiles of clinical interest.

RESULTS

Compared with the new charts, CFEF underestimates HC<-3DS and AC<10eP. The proportions of differences between the CFEF and INTERGROWTH-21 or WHO curves always remained <5%. The proportions of difference of the 3rd percentile of HC and FL, 10th and 90th percentile of AC were always lower than 2%, 2%, 5% and 4% respectively, between OMS and INTERGROWTH-21.

CONCLUSION

The switch to prescriptive standards suggests an improvement in the detection of fetuses with AC<10th percentile, an improvement in the detection of prenatal onset microcephaly, with no argument for a decrease in the detection rate of severe constitutional bone disease or modification of obstetrical guidelines.

Evolving the Era of 5D Ultrasound? A Systematic Literature Review on the Applications for Artificial Intelligence Ultrasound Imaging in Obstetrics and Gynecology

Artificial intelligence (AI) has gained prominence in medical imaging, particularly in obstetrics and gynecology (OB/GYN), where ultrasound (US) is the preferred method. It is considered cost effective and easily accessible but is time consuming and hindered by the need for specialized training. To overcome these limitations, AI models have been proposed for automated plane acquisition, anatomical measurements, and pathology detection. This study aims to overview recent literature on AI applications in OB/GYN US imaging, highlighting their benefits and limitations. For the methodology, a systematic literature search was performed in the PubMed and Cochrane Library databases. Matching abstracts were screened based on the PICOS (Participants, Intervention or Exposure, Comparison, Outcome, Study type) scheme. Articles with full text copies were distributed to the sections of OB/GYN and their research topics. As a result, this review includes 189 articles published from 1994 to 2023. Among these, 148 focus on obstetrics and 41 on gynecology. AI-assisted US applications span fetal biometry, echocardiography, or neurosonography, as well as the identification of adnexal and breast masses, and assessment of the endometrium and pelvic floor. To conclude, the applications for AI-assisted US in OB/GYN are abundant, especially in the subspecialty of obstetrics. However, while most studies focus on common application fields such as fetal biometry, this review outlines emerging and still experimental fields to promote further research.

Maternal PlGF and umbilical Dopplers predict pregnancy outcomes at diagnosis of early-onset fetal growth restriction

BACKGROUNDSevere, early-onset fetal growth restriction (FGR) causes significant fetal and neonatal mortality and morbidity. Predicting the outcome of affected pregnancies at the time of diagnosis is difficult, thus preventing accurate patient counseling. We investigated the use of maternal serum protein and ultrasound measurements at diagnosis to predict fetal or neonatal death and 3 secondary outcomes: fetal death or delivery at or before 28+0 weeks, development of abnormal umbilical artery (UmA) Doppler velocimetry, and slow fetal growth.METHODSWomen with singleton pregnancies (n = 142, estimated fetal weights [EFWs] below the third centile, less than 600 g, 20+0 to 26+6 weeks of gestation, no known chromosomal, genetic, or major structural abnormalities) were recruited from 4 European centers. Maternal serum from the discovery set (n = 63) was analyzed for 7 proteins linked to angiogenesis, 90 additional proteins associated with cardiovascular disease, and 5 proteins identified through pooled liquid chromatography and tandem mass spectrometry. Patient and clinician stakeholder priorities were used to select models tested in the validation set (n = 60), with final models calculated from combined data.RESULTSThe most discriminative model for fetal or neonatal death included the EFW z score (Hadlock 3 formula/Marsal chart), gestational age, and UmA Doppler category (AUC, 0.91; 95% CI, 0.86-0.97) but was less well calibrated than the model containing only the EFW z score (Hadlock 3/Marsal). The most discriminative model for fetal death or delivery at or before 28+0 weeks included maternal serum placental growth factor (PlGF) concentration and UmA Doppler category (AUC, 0.89; 95% CI, 0.83-0.94).CONCLUSIONUltrasound measurements and maternal serum PlGF concentration at diagnosis of severe, early-onset FGR predicted pregnancy outcomes of importance to patients and clinicians.TRIAL REGISTRATIONClinicalTrials.gov NCT02097667.FUNDINGThe European Union, Rosetrees Trust, Mitchell Charitable Trust.

Clinical use of the canine foetal kidney formula in dogs of different maternal sizes during the last ten days before parturition

Data concerning the use of the canine foetal kidney length (L) formula in the prediction of parturition timing are still lacking. In our study, we aimed to evaluate the accuracy of the L formula in predicting the parturition date during the last 10 days of pregnancy. Twenty-five clinically healthy pregnant bitches, aged 2-9 years and weighing 3.5-52.2 kg, were ultrasonographically monitored from -11 to 0 days before parturition (dbp). Kidney L was measured for the three most caudal foetuses, and the parturition day was estimated using the kidney formula, whose accuracy was calculated as the percentage of cases estimated (ranges of ± 1 or ± 2 days) on the actual parturition date. A K-proportions test was performed to identify differences in the accuracy among maternal sizes and the sex ratio of pups, and a two-proportions z-test was performed to identify differences between litter size classes (≤ 7 vs. > 7 pups) and time ranges (-11/-5 and -4/0 dbp). An accuracy of 35% within ± 2 days was found in the range of -11/-5 dbp and an accuracy of 30% within ± 2 days was found in the range of -4/0 dbp. The accuracy differed between small (53% ±1 day and 60% ±2 days) and large (10% within ± 1 and ± 2 days) bitches (P = 0.019 within ± 1 day, and P = 0.007 within ± 2 days). For small litter sizes, the accuracy was 38% within ± 1 day and 44% within ± 2 days, and for large litter sizes, it was 14% within ± 1 and ± 2 days. A threshold value was found between litter size classes within ± 2 days. The use of the L formula during the last ten days of pregnancy did not seem to warrant good accuracy in the prediction of parturition date. Further studies on different maternal sizes should be performed.

Use of artificial intelligence and deep learning in fetal ultrasound imaging

Deep learning is considered the leading artificial intelligence tool in image analysis in general. Deep-learning algorithms excel at image recognition, which makes them valuable in medical imaging. Obstetric ultrasound has become the gold standard imaging modality for detection and diagnosis of fetal malformations. However, ultrasound relies heavily on the operator's experience, making it unreliable in inexperienced hands. Several studies have proposed the use of deep-learning models as a tool to support sonographers, in an attempt to overcome these problems inherent to ultrasound. Deep learning has many clinical applications in the field of fetal imaging, including identification of normal and abnormal fetal anatomy and measurement of fetal biometry. In this Review, we provide a comprehensive explanation of the fundamentals of deep learning in fetal imaging, with particular focus on its clinical applicability. © 2022 International Society of Ultrasound in Obstetrics and Gynecology.

Quantifying Simulated Contamination Deposition on Healthcare Providers Using Image Analysis

INTRODUCTION

Simulation-based research has played an important role in improving care for communicable diseases. Unfortunately, few studies have attempted to quantify the level of contamination in these simulation activities. We aim to assess the feasibility and provide validity evidence for using integrated density values and area of contamination (AOC) to differentiate various levels of simulated contamination.

METHODS

An increasing number of simulated contamination spots using fluorescent marker were applied on a manikin chest to simulate a contaminated healthcare provider. An ultraviolet light was used to illuminate the manikin to highlight the simulated contamination. Images of increasing contamination levels were captured using a camera with different exposure settings. Image processing software was used to measure 2 outcomes: (1) natural logarithm of integrated density; and (2) AOC. Mixed-effects linear regression models were used to assess the effect of contamination levels and exposure settings on both outcome measures. A standardized "proof-of-concept" exercise was set up to calibrate and formalize the process for human subjects.

RESULTS

A total of 140 images were included in the analyses. Dose-response relationships were observed between contamination levels and both outcome measures. For each increment in the number of contaminated simulation spots (ie, simulated contaminated area increased by 38.5 mm 2 ), on average, log-integrated density increased by 0.009 (95% confidence interval, 0.006-0.012; P < 0.001) and measured AOC increased by 37.8 mm 2 (95% confidence interval, 36.7-38.8 mm 2 ; P < 0.001), which is very close to actual value (38.5 mm 2 ). The "proof-of-concept" demonstration further verified results.

CONCLUSIONS

Integrated density and AOC measured by image processing can differentiate various levels of simulated, fluorescent contamination. The AOC measured highly agrees with the actual value. This method should be optimized and used in the future research to detect simulated contamination deposited on healthcare providers.

Is accuracy of estimated fetal weight improved by better image quality scores?

OBJECTIVE

To assess in a group of ultrasound operators of various levels of experience the predictive value of systematic quality scoring to assess estimated fetal weight (EFW) validity.

METHODS

Screenshots, sonographer experience, and neonate birth weight were collected for 131 ultrasound examinations in the 7 days before birth. The difference (EFW error) between projected birth weight (EFW + [30 g × interval in days to birth]) and actual birth weight was then assessed (absolute value). Three senior sonographers rated all the screenshots (International Society of Ultrasound in Obstetrics and Gynecology 16-point score for image quality) and interobserver reproducibility was assessed concomitantly. The impact of the score on EFW accuracy was then assessed (univariate analysis). Receiver operating characteristic curves allowed us to assess the score's positive predictive value (PPV) for accurate EFW.

RESULTS

Mean birth weight was 2998 ± 954 g and mean EFW error was 8.6% ± 7.1%. Both the sonographer's experience and score significantly impacted the EFW error (P < 0.05). The PPVs of systematic image scores for identifying an EFW error greater than 10% and greater than 15% were appropriate for clinical use (areas under the curve 0.61 and 0.70, respectively). Score reproducibility was modest.

CONCLUSION

Low image scores and limited ultrasound expertise are associated with an increased risk of inaccurate EFW.

Automatic Detection and Measurement of Renal Cysts in Ultrasound Images: A Deep Learning Approach

Ultrasonography is widely used for diagnosis of diseases in internal organs because it is nonradioactive, noninvasive, real-time, and inexpensive. In ultrasonography, a set of measurement markers is placed at two points to measure organs and tumors, then the position and size of the target finding are measured on this basis. Among the measurement targets of abdominal ultrasonography, renal cysts occur in 20-50% of the population regardless of age. Therefore, the frequency of measurement of renal cysts in ultrasound images is high, and the effect of automating measurement would be high as well. The aim of this study was to develop a deep learning model that can automatically detect renal cysts in ultrasound images and predict the appropriate position of a pair of salient anatomical landmarks to measure their size. The deep learning model adopted fine-tuned YOLOv5 for detection of renal cysts and fine-tuned UNet++ for prediction of saliency maps, representing the position of salient landmarks. Ultrasound images were input to YOLOv5, and images cropped inside the bounding box and detected from the input image by YOLOv5 were input to UNet++. For comparison with human performance, three sonographers manually placed salient landmarks on 100 unseen items of the test data. These salient landmark positions annotated by a board-certified radiologist were used as the ground truth. We then evaluated and compared the accuracy of the sonographers and the deep learning model. Their performances were evaluated using precision-recall metrics and the measurement error. The evaluation results show that the precision and recall of our deep learning model for detection of renal cysts are comparable to standard radiologists; the positions of the salient landmarks were predicted with an accuracy close to that of the radiologists, and in a shorter time.

Impact of maternal body mass index on the accuracy of third trimester sonographic estimation of fetal weight

OBJECTIVES

To Determine whether maternal body mass index (BMI) can affect the accuracy of sonographic estimation of fetal weight (EFW) in the third trimester when compared to neonatal birthweight (BW).

METHODS

Secondary analysis from our original prospective cohort of pregnant women beyond 34 weeks, distributed in 4 groups according to their BMI: normal, overweight, obese and morbid obese. Fetal biometry and fluid measurements were obtained by two experienced sonographers, blinded for patient's clinical information and to each other's measurements. Average EFW and neonatal BW were converted into gestational-specific Z-scores. Interobserver correlation coefficient (ICC) and Cronbach's reliability coefficient (CRC) were calculated. Bland-Altman (BA) plots were constructed to assess the level of accuracy.

RESULTS

100 women were enrolled (800 measurements obtained by 17 sonographers): 17 had normal BMI (17%), 27 were overweight (27%), 29 were obese (29%) and 27 were morbidly obese (27%). There was no statistical difference for GA at delivery (p = 0.74), EFW (p = 0.05) or BW (p = 0.09) between groups (Table 1). Mean Z-score for EFW was - 0.17 (SD 0.81) and for neonatal BW was - 0.25 (SD 0.74). ICC was 0.69 (95% CI 0.57, 0.78) and CRC was 0.82. Mean Z-score difference was small (Table 2). When stratifying according to BMI categories, the ICC ranged from 0.49 to 0.76. Reliability indices ranged from 0.66 to 0.86. The Z-scores' differences were overall small with no statistical difference (Table 3). BA showed evenly distributed interobserver differences (Fig. 1).

CONCLUSIONS

When performed by trained sonographers, fetal weight estimation in the third trimester is accurate when compared to neonatal birthweight at increasing BMI categories.

[The French National Committee on Obstetrical and Fetal Ultrasound guidelines 2022 (CNEOF)]

The 2022 CNEOF guidelines (Conférence nationale d'échographie obstétricale et fœtale) report has been recently issued. It presents the necessary evolutions for the years to come, in a philosophy of optimal security for patients and caregivers, through a homogenization of ultrasound screening practices. As a source of changes in practices, this new version raises concerns, and even reticence, which must be heard and addressed, by reminding that this report is not fixed and can be adapted to the realities of practice over time and their feedback. This short text presents the CNEOF, the novelties of the 2022 report and details some important parts of the report that have been a source of questioning in the month following its publication. The aim of this text is to present a summary (in addition to the full report) to reassure, through education, all the parties involved in this medical practice which is so exciting and of major importance for perinatal health. Thus, the types of ultrasound examinations (screening, diagnostic, expertise…), the conditions of their realization, dating, biometries and the items part of the ultrasound reports are presented with elements of precision useful for their implementation.

Is Maternal Cardiovascular Performance Impaired in Altitude-Associated Fetal Growth Restriction?

Mundo, William, Lilian Toledo-Jaldin, Alexandrea Heath-Freudenthal, Jaime Huayacho, Litzi Lazo-Vega, Alison Larrea-Alvarado, Valquiria Miranda-Garrido, Rodrigo Mizutani, Lorna G. Moore, Any Moreno-Aramayo, Richard Gomez, Patricio Gutierrez, and Colleen G. Julian. Is maternal cardiovascular performance impaired in altitude-associated fetal growth restriction? High Alt Med Biol. 23:352-360, 2022. Introduction: The incidence of fetal growth restriction (FGR) is elevated in high-altitude resident populations. This study aims to determine whether maternal central hemodynamics during the last trimester of pregnancy are altered in high-altitude FGR. Methods: In this cross-sectional study of maternal-infant pairs (FGR, n = 27; controls, n = 26) residing in La Paz, Bolivia, maternal heart rate, cardiac output (CO), stroke volume, and systemic vascular resistance (SVR) were assessed using continuous-wave Doppler ultrasound. Transabdominal Doppler ultrasound was used for uterine artery (UtA) resistance indices and fetal measures. Maternal venous soluble fms-like tyrosine kinase-1 (sFlt1) levels were measured. Results: FGR pregnancies had reduced CO, elevated SVR and UtA resistance, fetal brain sparing, and increased maternal sFlt1 versus controls. Maternal SVR was positively associated with UtA resistance and inversely associated with middle cerebral artery resistance and birth weight. Maternal sFlt1 was greater in FGR than controls and positively associated with UtA pulsatility index. Women with elevated sFlt1 levels also tended to have lower CO and higher SVR. Conclusion: Noninvasive assessment of maternal cardiovascular function may be an additional method for detecting high-risk pregnancies at high altitudes, thereby informing the need for increased surveillance and appropriate allocation of resources to minimize adverse outcomes.

Semiautomated Software to Improve Stability and Reduce Operator-Induced Variation in Vascular Ultrasound Speckle Tracking

OBJECTIVES

Ultrasound is useful in predicting arteriovenous fistula (AVF) maturation, which is essential for hemodialysis in end-stage renal disease patients. We developed ultrasound software that measures circumferential vessel wall strain (distensibility) using conventional ultrasound Digital Imaging and Communications in Medicine (DICOM) data. We evaluated user-induced variability in measurement of arterial wall distensibility and upon finding considerable variation we developed and tested 2 methods for semiautomated measurement.

METHODS

Ultrasound scanning of arteries of 10 subjects scheduled for AVF surgery were performed. The top and bottom of the vessel wall were tracked using the Kanade-Lucas-Tomasi (KLT) feature-tracking algorithm over the stack of images in the DICOM cine loops. The wall distensibility was calculated from the change of vessel diameter over time. Two semiautomated methods were used for comparison.

RESULTS

The location of points selected by users for the cine loops varied significantly, with a maximum spread of up to 120 pixels (7.8 mm) for the top and up to 140 pixels (9.1 mm) for the bottom of the vessel wall. This variation in users' point selection contributed to the variation in distensibility measurements (ranging from 5.63 to 41.04%). Both semiautomated methods substantially reduced variation and were highly correlated with the median distensibility values obtained by the 10 users.

CONCLUSIONS

Minimizing user-induced variation by standardizing point selection will increase reproducibility and reliability of distensibility measurements. Our recent semiautomated software may help expand use in clinical studies to better understand the role of vascular wall compliance in predicting the maturation of fistulas.

Student Competency for Midtrimester Obstetrics Scan upon Completion of the Master’s Degree in Medical Sonography

Objectives

To evaluate the competency of medical sonographer students who have completed training to estimate the gestational age (GA) and perform fetal biometric measurements compared to obstetricians.

Methods

We conducted a cross-sectional observational study at the end of the medical sonographer students' practice sessions. In total, 80 midtrimester (18–28 weeks) pregnant women were recruited, and an ultrasound was performed according to the International Society of Sonography in Obstetrics and Gynecology (ISUOG) guideline. Estimated GA calculated from fetal biometric measurements was compared between medical sonographer students and qualified obstetricians. Subsequently, images were randomly evaluated by maternal-fetal medicine specialists to assess the measurement performance.

Results

There was no significant difference in the estimated GA between the medical sonographer students and obstetricians (mean difference, 0.01 ± 2.92 day, p = 0.89). However, there was a significant difference in the measurement of the head circumference (HC) and abdominal circumference (AC) (p < 0.001). The overall image quality of the fetal head, abdomen, and femur was considered a good to excellent score (77.5%–80%). There was a perfect and nearly perfect agreement regarding the presence of the placenta previa, adequacy of amniotic fluid, and position of the placenta (k = 0.9–1.0).

Conclusions

The medical sonographer students demonstrated competency in GA estimation by fetal biometry measurement similar to obstetricians. However, the quality of the acquired images, according to the ISUOG recommendation, needs improvement, and this should be emphasized in the sonography course curriculum. The results suggest that medical sonographers can relieve obstetricians' workload for ultrasound screening in midtrimester pregnancies.

Placenta segmentation in ultrasound imaging: Addressing sources of uncertainty and limited field-of-view

Automatic segmentation of the placenta in fetal ultrasound (US) is challenging due to the (i) high diversity of placenta appearance, (ii) the restricted quality in US resulting in highly variable reference annotations, and (iii) the limited field-of-view of US prohibiting whole placenta assessment at late gestation. In this work, we address these three challenges with a multi-task learning approach that combines the classification of placental location (e.g., anterior, posterior) and semantic placenta segmentation in a single convolutional neural network. Through the classification task the model can learn from larger and more diverse datasets while improving the accuracy of the segmentation task in particular in limited training set conditions. With this approach we investigate the variability in annotations from multiple raters and show that our automatic segmentations (Dice of 0.86 for anterior and 0.83 for posterior placentas) achieve human-level performance as compared to intra- and inter-observer variability. Lastly, our approach can deliver whole placenta segmentation using a multi-view US acquisition pipeline consisting of three stages: multi-probe image acquisition, image fusion and image segmentation. This results in high quality segmentation of larger structures such as the placenta in US with reduced image artifacts which are beyond the field-of-view of single probes.

Developing Clinical Artificial Intelligence for Obstetric Ultrasound to Improve Access in Underserved Regions: Protocol for a Computer-Assisted Low-Cost Point-of-Care UltraSound (CALOPUS) Study

BACKGROUND

The World Health Organization recommends a package of pregnancy care that includes obstetric ultrasound scans. There are significant barriers to universal access to antenatal ultrasound, particularly because of the cost and need for maintenance of ultrasound equipment and a lack of trained personnel. As low-cost, handheld ultrasound devices have become widely available, the current roadblock is the global shortage of health care providers trained in obstetric scanning.

OBJECTIVE

The aim of this study is to improve pregnancy and risk assessment for women in underserved regions. Therefore, we are undertaking the Computer-Assisted Low-Cost Point-of-Care UltraSound (CALOPUS) project, bringing together experts in machine learning and clinical obstetric ultrasound.

METHODS

In this prospective study conducted in two clinical centers (United Kingdom and India), participating pregnant women were scanned and full-length ultrasounds were performed. Each woman underwent 2 consecutive ultrasound scans. The first was a series of simple, standardized ultrasound sweeps (the CALOPUS protocol), immediately followed by a routine, full clinical ultrasound examination that served as the comparator. We describe the development of a simple-to-use clinical protocol designed for nonexpert users to assess fetal viability, detect the presence of multiple pregnancies, evaluate placental location, assess amniotic fluid volume, determine fetal presentation, and perform basic fetal biometry. The CALOPUS protocol was designed using the smallest number of steps to minimize redundant information, while maximizing diagnostic information. Here, we describe how ultrasound videos and annotations are captured for machine learning.

RESULTS

Over 5571 scans have been acquired, from which 1,541,751 label annotations have been performed. An adapted protocol, including a low pelvic brim sweep and a well-filled maternal bladder, improved visualization of the cervix from 28% to 91% and classification of placental location from 82% to 94%. Excellent levels of intra- and interannotator agreement are achievable following training and standardization.

CONCLUSIONS

The CALOPUS study is a unique study that uses obstetric ultrasound videos and annotations from pregnancies dated from 11 weeks and followed up until birth using novel ultrasound and annotation protocols. The data from this study are being used to develop and test several different machine learning algorithms to address key clinical diagnostic questions pertaining to obstetric risk management. We also highlight some of the challenges and potential solutions to interdisciplinary multinational imaging collaboration.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR1-10.2196/37374.

A new approach to automatic measure fetal head circumference in ultrasound images using convolutional neural networks

Fetal head circumference (HC) is an important biological index in prenatal ultrasound screening. In the clinic, fetal HC is usually measured manually by sonographers in two dimensional (2D) ultrasound images. The manual method is significantly affected by the inter/intra-observer difference and the process of manual measurement is inconvenient and time-consuming for sonographers. Although several artificial intelligence (AI) approaches had been applied to fetal HC measurement, they had weak generalization ability, especially for the incomplete or blurred skull edge. In this study, a fast and accurate method for fetal HC auto-measurement was proposed. Different from the common region segmentation method, an end-to-end convolutional neural network (CNN) for fetal skull boundary segmentation in 2D ultrasound images is proposed, which is an efficient method to directly segment the boundary of fetal skull by using the proposed double-branch structure. The segmentation results can be directly used to calculate fetal HC without complex post-processing. The proposed approach achieved excellent results: Mean Dice Sore (MDS)±std: 97.98 ± 1.30, Mean Hausdorff Distance (MHD)±std: 1.20 ± 0.68 mm, Mean Absolute Difference (MAD)±std: 1.75 ± 1.60 mm, Mean Difference (MD)±std: 0.08 ± 2.37 mm. Additionally, we drew a Bland-Altman plot to demonstrate that HC measured by the proposed approach has high agreement with the real value. Comprehensive results show that the proposed approach is comparable to the state-of-the-art methods for fetal HC measurement. Meanwhile, our approach belongs to a lightweight network with less parameters, which is convenient for deployment. We hope it could provide help for precision medicine in prenatal ultrasound screening.

Household income, fetal size and birth weight: an analysis of eight populations

BACKGROUND

The age at onset of the association between poverty and poor health is not understood. Our hypothesis was that individuals from highest household income (HI), compared to those with lowest HI, will have increased fetal size in the second and third trimester and birth.

METHODS

Second and third trimester fetal ultrasound measurements and birth measurements were obtained from eight cohorts. Results were analysed in cross-sectional two-stage individual patient data (IPD) analyses and also a longitudinal one-stage IPD analysis.

RESULTS

The eight cohorts included 21 714 individuals. In the two-stage (cross-sectional) IPD analysis, individuals from the highest HI category compared with those from the lowest HI category had larger head size at birth (mean difference 0.22 z score (0.07, 0.36)), in the third trimester (0.25 (0.16, 0.33)) and second trimester (0.11 (0.02, 0.19)). Weight was higher at birth in the highest HI category. In the one-stage (longitudinal) IPD analysis which included data from six cohorts (n=11 062), head size was larger (mean difference 0.13 (0.03, 0.23)) for individuals in the highest HI compared with lowest category, and this difference became greater between the second trimester and birth. Similarly, in the one-stage IPD, weight was heavier in second highest HI category compared with the lowest (mean difference 0.10 (0 .00, 0.20)) and the difference widened as pregnancy progressed. Length was not linked to HI category in the longitudinal model.

CONCLUSIONS

The association between HI, an index of poverty, and fetal size is already present in the second trimester.

Task model-specific operator skill assessment in routine fetal ultrasound scanning

Purpose

For highly operator-dependent ultrasound scanning, skill assessment approaches evaluate operator competence given available data, such as acquired images and tracked probe movement. Operator skill level can be quantified by the completeness, speed, and precision of performing a clinical task, such as biometry. Such clinical tasks are increasingly becoming assisted or even replaced by automated machine learning models. In addition to measurement, operators need to be competent at the upstream task of acquiring images of sufficient quality. To provide computer assistance for this task requires a new definition of skill.

Methods

This paper focuses on the task of selecting ultrasound frames for biometry, for which operator skill is assessed by quantifying how well the tasks are performed with neural network-based frame classifiers. We first develop a frame classification model for each biometry task, using a novel label-efficient training strategy. Once these task models are trained, we propose a second task model-specific network to predict two skill assessment scores, based on the probability of identifying positive frames and accuracy of model classification.

Results

We present comprehensive results to demonstrate the efficacy of both the frame-classification and skill-assessment networks, using clinically acquired data from two biometry tasks for a total of 139 subjects, and compare the proposed skill assessment with metrics of operator experience.

Conclusion

Task model-specific skill assessment is feasible and can be predicted by the proposed neural networks, which provide objective assessment that is a stronger indicator of task model performance, compared to existing skill assessment methods.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11548-022-02642-y.

RDHCformer: Fusing ResDCN and Transformers for Fetal Head Circumference Automatic Measurement in 2D Ultrasound Images

Fetal head circumference (HC) is an important biological parameter to monitor the healthy development of the fetus. Since there are some HC measurement errors that affected by the skill and experience of the sonographers, a rapid, accurate and automatic measurement for fetal HC in prenatal ultrasound is of great significance. We proposed a new one-stage network for rotating elliptic object detection based on anchor-free method, which is also an end-to-end network for fetal HC auto-measurement that no need for any post-processing. The network structure used simple transformer structure combined with convolutional neural network (CNN) for a lightweight design, meanwhile, made full use of powerful global feature extraction ability of transformer and local feature extraction ability of CNN to extract continuous and complete skull edge information. The two complement each other for promoting detection precision of fetal HC without significantly increasing the amount of computation. In order to reduce the large variation of intersection over union (IOU) in rotating elliptic object detection caused by slight angle deviation, we used soft stage-wise regression (SSR) strategy for angle regression and added KLD that is approximate to IOU loss into total loss function. The proposed method achieved good results on the HC18 dataset to prove its effectiveness. This study is expected to help less experienced sonographers, provide help for precision medicine, and relieve the shortage of sonographers for prenatal ultrasound in worldwide.

Incorporating personal-device-based point-of-care ultrasound into obstetric care: a validation study

BACKGROUND

Personal-device-based point-of-care-ultrasound (P-POCUS) probes plug directly into a cell phone or tablet to function as its display, creating the potential to increase access to obstetric ultrasonography in complex healthcare settings (COVID units, low resource settings); however, new technology must be proven to be reliable in the obstetric setting before integrating into practice.

OBJECTIVE

To evaluate the intraclass correlation (reliability) of personal-device-based-point-of-care-ultrasound devices as compared with standard ultrasound machines in obstetrics.

STUDY DESIGN

This was a prospective, observational study of patients between 19-39 weeks gestation in an urban, prenatal ultrasound diagnosis center. Each patient underwent assessment by an expert sonographer using standard ultrasound machines and personal-device-based-point-of-care-ultrasound devices to determine estimated fetal weight. The statistical reliability and agreement between the estimated fetal weights was assessed through intraclass correlation coefficients, Bland-Altman plots, and Pearson correlation coefficients.

RESULTS

100 paired sets of scans were performed from October 2020 to December 2020. For the estimated fetal weights, there was near-perfect agreement, with an intraclass correlation coefficient of 0.99 (P<.0001). Bland-Altman analysis showed an average difference of 53 grams, with 95% limit of agreement between -178 grams and 283 grams. Pearson correlation showed near-perfect correlation between the measurements (r=0.99, P<.0001).

CONCLUSION

personal-device-based point-of-care-ultrasound devices are reliable tools for performing basic obstetrical ultrasound and have the potential to increase access to obstetrical ultrasound worldwide.

Deep learning fetal ultrasound video model match human observers in biometric measurements

Objective.This work investigates the use of deep convolutional neural networks (CNN) to automatically perform measurements of fetal body parts, including head circumference, biparietal diameter, abdominal circumference and femur length, and to estimate gestational age and fetal weight using fetal ultrasound videos.Approach.We developed a novel multi-task CNN-based spatio-temporal fetal US feature extraction and standard plane detection algorithm (called FUVAI) and evaluated the method on 50 freehand fetal US video scans. We compared FUVAI fetal biometric measurements with measurements made by five experienced sonographers at two time points separated by at least two weeks. Intra- and inter-observer variabilities were estimated.Main results.We found that automated fetal biometric measurements obtained by FUVAI were comparable to the measurements performed by experienced sonographers The observed differences in measurement values were within the range of inter- and intra-observer variability. Moreover, analysis has shown that these differences were not statistically significant when comparing any individual medical expert to our model.Significance.We argue that FUVAI has the potential to assist sonographers who perform fetal biometric measurements in clinical settings by providing them with suggestions regarding the best measuring frames, along with automated measurements. Moreover, FUVAI is able perform these tasks in just a few seconds, which is a huge difference compared to the average of six minutes taken by sonographers. This is significant, given the shortage of medical experts capable of interpreting fetal ultrasound images in numerous countries.

Segmentation-Based vs. Regression-Based Biomarker Estimation: A Case Study of Fetus Head Circumference Assessment from Ultrasound Images

The fetus head circumference (HC) is a key biometric to monitor fetus growth during pregnancy, which is estimated from ultrasound (US) images. The standard approach to automatically measure the HC is to use a segmentation network to segment the skull, and then estimate the head contour length from the segmentation map via ellipse fitting, usually after post-processing. In this application, segmentation is just an intermediate step to the estimation of a parameter of interest. Another possibility is to estimate directly the HC with a regression network. Even if this type of segmentation-free approaches have been boosted with deep learning, it is not yet clear how well direct approach can compare to segmentation approaches, which are expected to be still more accurate. This observation motivates the present study, where we propose a fair, quantitative comparison of segmentation-based and segmentation-free (i.e., regression) approaches to estimate how far regression-based approaches stand from segmentation approaches. We experiment various convolutional neural networks (CNN) architectures and backbones for both segmentation and regression models and provide estimation results on the HC18 dataset, as well agreement analysis, to support our findings. We also investigate memory usage and computational efficiency to compare both types of approaches. The experimental results demonstrate that even if segmentation-based approaches deliver the most accurate results, regression CNN approaches are actually learning to find prominent features, leading to promising yet improvable HC estimation results.

Assessing fetal growth in Africa: Application of the international WHO and INTERGROWTH-21st standards in a Beninese pregnancy cohort

BACKGROUND

Fetal growth restriction is a major complication of pregnancy and is associated with stillbirth, infant death and child morbidity. Ultrasound monitoring of pregnancy is becoming more common in Africa for fetal growth monitoring in clinical care and research, but many countries have no national growth charts. We evaluated the new international fetal growth standards from INTERGROWTH-21st and WHO in a cohort from southern Benin.

METHODS

Repeated ultrasound and clinical data were collected in women from the preconceptional RECIPAL cohort (241 women with singleton pregnancies, 964 ultrasounds). We modelled fetal biometric parameters including abdominal circumference (AC) and estimated fetal weight (EFW) and compared centiles to INTERGROWTH-21st and WHO standards, using the Bland and Altman method to assess agreement. For EFW, we used INTERGROWTH-21st standards based on their EFW formula (IG21st) as well as a recent update using Hadlock's EFW formula (IG21hl). Proportions of fetuses with measurements under the 10th percentile were compared.

RESULTS

Maternal malaria and anaemia prevalence was 43% and 69% respectively and 11% of women were primigravid. Overall, the centiles in the RECIPAL cohort were higher than that of INTERGROWTH-21st and closer to that of WHO. Consequently, the proportion of fetuses under 10th percentile thresholds was systematically lower when applying IG21st compared to WHO standards. At 27-31 weeks and 33-38 weeks, respectively, 7.4% and 5.6% of fetuses had EFW <10th percentile using IG21hl standards versus 10.7% and 11.6% using WHO standards.

CONCLUSION

Despite high anemia and malaria prevalence in the cohort, IG21st and WHO standards did not identify higher than expected proportions of fetuses under the 10th percentiles of ultrasound parameters or EFW. The proportions of fetuses under the 10th percentile threshold for IG21st charts were particularly low, raising questions about its use to identify growth-restricted fetuses in Africa.

Influence of maternal body mass index on interobserver variability of fetal ultrasound biometry and amniotic-fluid assessment in late pregnancy

OBJECTIVE

To determine the interobserver reproducibility of fetal ultrasound biometric and amniotic-fluid measurements in the third trimester of pregnancy, according to maternal body mass index (BMI) category.

METHODS

This was a prospective cohort study of women with a singleton gestation beyond 34 weeks, recruited into four groups according to BMI category: normal (18.0-24.9 kg/m² ), overweight (25.0-29.9 kg/m²⁾ , obese (30.0-39.9 kg/m² ) and morbidly obese (≥ 40 kg/m² ). Multiple pregnancies, women with diabetes and pregnancies with a fetal growth, structural or genetic abnormality were excluded. In each woman, fetal biometric (biparietal diameter (BPD), head circumference, abdominal circumference (AC), femur length (FL) and estimated fetal weight) and amniotic-fluid (amniotic-fluid index (AFI) and maximum vertical pocket (MVP)) measurements were obtained by two experienced sonographers or physicians, blinded to gestational age and each other's measurements. Differences in measurements between observers were expressed as gestational age-specific Z-scores. The interobserver intraclass correlation coefficient (ICC) and Cronbach's reliability coefficient (CRC) were calculated. Bland-Altman analysis was used to assess the degree of reproducibility.

RESULTS

In total, 110 women were enrolled prospectively (including 1320 measurements obtained by 17 sonographers or physicians). Twenty (18.2%) women had normal BMI, 30 (27.3%) women were overweight, 30 (27.3%) women were obese and 30 (27.3%) women were morbidly obese. Except for AFI (ICC, 0.65; CRC, 0.78) and MVP (ICC, 0.49; CRC, 0.66), all parameters had a very high level of interobserver reproducibility (ICC, 0.72-0.87; CRC, 0.84-0.93). When assessing reproducibility according to BMI category, BPD measurements had a very high level of reproducibility (ICC ≥ 0.85; CRC > 0.90) in all groups. The reproducibility of AC and FL measurements increased with increasing BMI, while the reproducibility of MVP measurements decreased. Among the biometric parameters, the difference between the BMI categories in measurement-difference Z-score was significant only for FL. Interobserver differences for biometric measurements fell within the 95% limits of agreement.

CONCLUSION

Obesity does not seem to impact negatively on the reproducibility of ultrasound measurements of fetal biometric parameters when undertaken by experienced sonographers or physicians who commonly assess overweight, obese and morbidly obese women. © 2021 International Society of Ultrasound in Obstetrics and Gynecology.

Fusion imaging in brain structure measurements on a fetus phantom, combining real-time ultrasound with magnetic resonance imaging

Objectives

To assess synchronisation of MRI and US in measuring foetus phantom head structures; inter‐method, intra‐ and inter‐observer differences on biparietal diameter (BPD), head diameter, anterio‐posterior head diameter (HAP) and lateral ventricle structures (VS).

Methods

Fusion Imaging (FI) has been performed by combining MRI and US simultaneously. Axial scans of 1.5 Tesla MRI on a foetus phantom were acquired and uploaded on a US machine (EPIQ 7G, Philips). A PercuNav US tracker allowed the system to recognise and display the position of the transducer. A fetal phantom tracker was used as a phantom reference. Real‐time US of the phantom head was performed by synchronising the uploaded MRI images using different landmarks. Synchronisation has been assessed by taking measurements after rotating the US probe by 90. Measurements were taken by three different observers twice. Differences in measurements between MRI and US, inter‐, intra‐observer differences in all measurements were assessed.

Results

BPD, HAP and VS measurements before rotation were 0.13 ± 0.06 cm, 0.46 ± 0.09 cm and 0.4 ± 0.23 cm (width) and mean 0.6 ± 0.25 cm (length) larger at MRI than at US using any number of landmarks. After US probe rotation VS were 0.3 ± 0.24 cm in width and 0.3 ± 0.27 cm in length. Intra‐ and inter‐observer differences in all measurements were small.

Conclusions

FI showed good synchronisation in measurements. BPD, HAP and VS were larger at MRI than US, likely a result of the way images are generated. Intra‐, inter‐observer differences between measurements were small. This can be important when reporting geometric measures from FI.

Induction of labor for a suspected large-for-gestational-age/macrosomic fetus

Fetal macrosomia is defined as a birth weight of >4000 g, while the term large for gestational age (LGA) is defined as an estimated fetal weight >90th centile for gestational age. Current data indicate that a significant proportion of the babies are LGA. Pregnancies involving LGA babies are associated with increased maternal and perinatal morbidity including caesarean section, postpartum hemorrhage, shoulder dystocia, and birth trauma. To reduce these complications, labor induction has been suggested as a possible solution. However, despite some high-quality evidence in favor of labor induction for suspected macrosomia/LGA, existing guidelines do not support routine induction of labor in this population. The aim of this paper is to critically appraise the available evidence and clinical practice recommendations and highlight the importance of shared decision making and individualized care based on clear counselling regarding the lack of a sensitive diagnostic tool for estimating fetal weight in the third trimester.

Ultrasound in the prediction of birthweight discordance in dichorionic twins

INTRODUCTION

Large birthweight discrepancy has been identified as a risk factor for perinatal morbidity and mortality in twin pregnancies. However, it remains unclear whether such discordance can be predicted by various biological indices with specific cut-off values, and how these depend on the gestational age. We aimed to determine the most effective way to predict large birthweight discordance at various gestational ages.

MATERIAL AND METHODS

A retrospective cohort study of dichorionic twins, live-born between 2008 and 2018, was conducted. Discordances in biparietal diameter, head circumference, humerus and femur length, abdominal circumference, and estimated fetal weight were calculated-([larger twin - smaller twin] / larger twin) × 100%-and compared between those with and without a large birthweight discordance (≥20%). Receiver operating characteristic curves were constructed to analyze the predictive characteristics of each parameter.

RESULTS

Of 598 dichorionic twin pregnancies included, 83 (13.9%) had a birthweight discordance ≥20%. Group differences in biparietal diameter and head circumference discordance were the earliest to emerge (before 20 weeks of gestation), but became insignificant after 36 weeks, followed by humerus and femur length, estimated fetal weight discordance (after 20 weeks), and abdominal circumference discordance (after 28 weeks). The best predictors (with cut-off values) were discordance in biparietal diameter ≥7.8% at <20 weeks, head circumference ≥4.5% at 20-23⁺⁶  weeks, humerus length ≥4.5% at 24-27⁺⁶  weeks, and estimated fetal weight discordance (≥11.6% at 28-31⁺⁶  weeks, ≥10.5% at 32-35⁺⁶  weeks, and ≥15.0% ≥36 weeks), with sensitivity and specificity of 52%-77% and 69%-82%, respectively.

CONCLUSIONS

Different predictors and cut-off values may be useful for predicting large inter-twin birthweight discordance in dichorionic twins at different gestational ages. It is more accurate to use biparietal diameter and head circumference discordance in the early second trimester, humerus length discordance in the late second trimester, and estimated fetal weight discordance in the third trimester.

Quality control of fetal biometric evaluation and Doppler ultrasound

In recent years quality control in obstetric ultrasound has become recommended and an essential component of obstetric scanning. This is to minimize the inaccuracy and variability related to fetal measurements, to provide an effective quality assurance system to sonographers to certify their practice and decrease the impact of medical litigations. For a quality control system in obstetric ultrasound to be useful clinically, multiple strategies need to be employed: certified training, practical standardization exercise, image storing, qualitative and quantitative quality control. Qualitative quality control consists of the evaluation of images obtained for fetal biometry and Doppler scans using an objective score against predefined criteria. Quantitative quality control consists of analyzing quantitatively the performance of a sonographer and the impact on measurements values. Quantitative analysis could be performed either using estimates of intraobserver or interobserver reproducibility of plane acquisition and caliper placements.

Fetal cerebellar growth and Sylvian fissure maturation: international standards from Fetal Growth Longitudinal Study of INTERGROWTH-21st Project

OBJECTIVE

To construct international ultrasound-based standards for fetal cerebellar growth and Sylvian fissure maturation.

METHODS

Healthy, well nourished pregnant women, enrolled at < 14 weeks' gestation in the Fetal Growth Longitudinal Study (FGLS) of INTERGROWTH-21^st , an international multicenter, population-based project, underwent serial three-dimensional (3D) fetal ultrasound scans every 5 ± 1 weeks until delivery in study sites located in Brazil, India, Italy, Kenya and the UK. In the present analysis, only those fetuses that underwent developmental assessment at 2 years of age were included. We measured the transcerebellar diameter and assessed Sylvian fissure maturation using two-dimensional ultrasound images extracted from available 3D fetal head volumes. The appropriateness of pooling data from the five sites was assessed using variance component analysis and standardized site differences. For each Sylvian fissure maturation score (left or right side), mean gestational age and 95% CI were calculated. Transcerebellar diameter was modeled using fractional polynomial regression, and goodness of fit was assessed.

RESULTS

Of those children in the original FGLS cohort who had developmental assessment at 2 years of age, 1130 also had an available 3D ultrasound fetal head volume. The sociodemographic characteristics and pregnancy/perinatal outcomes of the study sample confirmed the health and low-risk status of the population studied. In addition, the fetuses had low morbidity and adequate growth and development at 2 years of age. In total, 3016 and 2359 individual volumes were available for transcerebellar-diameter and Sylvian-fissure analysis, respectively. Variance component analysis and standardized site differences showed that the five study populations were sufficiently similar on the basis of predefined criteria for the data to be pooled to produce international standards. A second-degree fractional polynomial provided the best fit for modeling transcerebellar diameter; we then estimated gestational-age-specific 3^rd , 50^th and 97^th smoothed centiles. Goodness-of-fit analysis comparing empirical centiles with smoothed centile curves showed good agreement. The Sylvian fissure increased in maturation with advancing gestation, with complete overlap of the mean gestational age and 95% CIs between the sexes for each development score. No differences in Sylvian fissure maturation between the right and left hemispheres were observed.

CONCLUSION

We present, for the first time, international standards for fetal cerebellar growth and Sylvian fissure maturation throughout pregnancy based on a healthy fetal population that exhibited adequate growth and development at 2 years of age. © 2020 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

Fetal biometry in ultrasound: A new approach to assess the long-term impact of simulation on learning patterns

CONTEXT

Simulation-based education (SBE) has demonstrated its acceptability and effectiveness in improving ultrasound training. Because of the high cost of its implementation (investment in equipment and supervision), a pragmatic assessment of the transfer of skills learned in SBE to clinical practice and the identification of its optimal scheduling conditions have been requested to optimize its input.

OBJECTIVES

To quantify the long-term impact of simulation-based education (SBE) on the adequate performance of ultrasound fetal biometry measurements (I). The secondary objective was to identify the temporal patterns that enhanced SBE input in learning (II).

METHODS

Trainees were arbitrarily assigned to a 6-month course in obstetric ultrasound with or without an SBE workshop. In the SBE group, the workshop was implemented 'before' or at an 'early' or a 'late-stage' of the course. Those who did not receive SBE were the control group. The ultrasound skills of all trainees were prospectively collected, evaluated by calculating the delta between OSAUS (Objective Structured Assessment of Ultrasound Skills) scores before and after the course (I). Concomitantly, the accuracy of trainees' measurements was assessed throughout the course by verifying their correlation with the corresponding measurements by their supervisors. The percentage of trainees able to perform five consecutive sets of correct measurements in the control group and in each SBE subgroup were compared (II).

RESULTS

The study included 61 trainees (39 SBE and 22 controls). Comparisons between groups showed no significant difference in the quantitative assessment of skill enhancement (difference in the pre- and post-internship OSAUS score: 1.09 ± 0.87 in the SBE group and 0.72 ± 0.98 in the control group) (I). Conversely, the predefined acceptable skill level was reached by a significantly higher proportion of trainees in the 'early' SBE subgroup (74%, compared with 30% in the control group, P<0.01)(II).

CONCLUSIONS

The quantitative assessment does not support the existence of long-term benefits from SBE training, although the qualitative assessment confirmed SBE helped to raise the minimal level within a group when embedded in an 'early' stage of a practical course.

Gestational Age-Specific Distribution of the Hammersmith Neonatal Neurological Examination Scores Among Low-Risk Neonates in Ghana

OBJECTIVE

To describe gestational age-specific distribution of scores for the Hammersmith Neonatal Neurological Examination (HNNE) up to 48 h after birth in a low-risk, term-born, single-center sample in Ghana.

STUDY DESIGN

This is a nested substudy of a larger prospective study (IMPRINT: Impact of Malaria in Pregnancy on Infant Neurodevelopment) comprising 140 low-risk, term-born neonates at Korle Bu Teaching Hospital in Accra, Ghana, between November 2018 and February 2019. The sample was stratified into three gestational age groups: early-term (37 + 0-38 + 6, weeks + days; n = 61), full-term (39 + 0-40 + 6, weeks + days; n = 52), and late/post-term (41 + 0-42 + 6, weeks + days; n = 27). Neonates were administered the 34-item HNNE by trained physicians. As per the original British scoring system, raw scores for the Ghanaian sample were plotted and scores > 10th centile were assigned a score of 1, 5th-10th centile 0.5, and < 5th centile 0.

RESULTS

The range of raw scores for 16/34 HNNE items varied with gestational age. Specifically, 100% (7/7), 50% (5/10), 33% (1/3), 33% (1/3), 20% (1/5), and 14% (1/7) of items within the orientation and behavior, tone, abnormal signs/patterns, movements, tone patterns, and reflexes subdomain, respectively showed a different distribution of scores above the 10th centile across the three gestational age groups.

CONCLUSION

Differences in gestational age-specific results within our sample in comparison to the original British sample could be, albeit unlikely, due to misclassification of gestational age, unmeasured maternal or fetal morbidity, or perhaps more likely, variation in testing or test conditions, or some combination of these. Genetic variation in neurological development is also a possibility. Further research is warranted to determine the reasons for differences. Our findings highlight the need to determine the accuracy and reliability of standardized neurologic assessments in predicting neurodevelopmental risk for infants in low- and middle-income countries.

Scale-aware Auto-context-guided Fetal US Segmentation with Structured Random Forests

Abstract Accurate measurement of fetal biometrics in ultrasound at different trimesters is essential in assisting clinicians to conduct pregnancy diagnosis. However, the accuracy of manual segmentation for measurement is highly user-dependent. Here, we design a general framework for automatically segmenting fetal anatomical structures in two-dimensional (2D) ultrasound (US) images and thus make objective biometric measurements available. We first introduce structured random forests (SRFs) as the core discriminative predictor to recognize the region of fetal anatomical structures with a primary classification map. The patch-wise joint labeling presented by SRFs has inherent advantages in identifying an ambiguous/fuzzy boundary and reconstructing incomplete anatomical boundary in US. Then, to get a more accurate and smooth classification map, a scale-aware auto-context model is injected to enhance the contour details of the classification map from various visual levels. Final segmentation can be obtained from the converged classification map with thresholding. Our framework is validated on two important biometric measurements, which are fetal head circumference (HC) and abdominal circumference (AC). The final results illustrate that our proposed method outperforms state-of-the-art methods in terms of segmentation accuracy.

How often do we incidentally find a fetal abnormality at the routine third-trimester growth scan? A population-based study

BACKGROUND

Third-trimester scans are increasingly used to try to prevent adverse outcomes associated with abnormalities of fetal growth. Unexpected fetal malformations detected at third-trimester growth scans are rarely reported.

OBJECTIVE

To determine the incidence and type of fetal malformations detected in women attending a routine third-trimester growth scan.

STUDY DESIGN

This was a population-based study of all women with singleton pregnancy attending antenatal care over a 2-year period in Oxfordshire, UK. Women who had a viable singleton pregnancy at dating scan were included. Women had standard obstetrical care including the offer of a routine dating scan and combined screening for trisomies; a routine anomaly scan at 18 to 22 weeks; and a routine third-trimester growth scan at 36 weeks. The third-trimester scan comprises assessment of fetal presentation, amniotic fluid, biometry, umbilical and middle cerebral artery Dopplers, but no formal anatomic assessment is undertaken. Scans are performed by certified sonographers or clinical fellows (n=54), and any suspected abnormalities are evaluated by a team of fetal medicine specialists. We assessed the frequency and type of incidental congenital malformations identified for the first time at this third-trimester scan. All babies were followed-up after birth for a minimum of 6 months.

RESULTS

There were 15,244 women attending routine antenatal care. Anomalies were detected in 474 (3.1%) fetuses as follows: 103 (21.7%) were detected before the anomaly scan, 174 (36.7%) at the anomaly scan, 11 (2.3%) after the anomaly scan and before the third-trimester scan, 43 (9.1%) at the third-trimester scan and 143 (30.2%) after birth. The 43 abnormalities were found in a total of 13,023 women who had a 36 weeks scan, suggesting that in 1 out of 303 (95% confidence interval, 233-432) women attending such a scan, a new malformation was detected. Anomalies detected at the routine third-trimester scan were of the urinary tract (n=30), central nervous system (5), simple ovarian cysts (4), chromosomal (1), splenic cyst (1), skeletal dysplasia (1), and cutaneous lymphangioma (1). Most urinary tract anomalies were renal pelvic dilatation, which showed spontaneous resolution in 57% of the cases.

CONCLUSION

When undertaking a program of routine third-trimester growth scans in women who have had previous screening scans, an unexpected congenital malformation is detected in approximately 1 in 300 women.

Training the trainees: a pilot study of inter-observer discrepancy and learning curve in the maternal foetal unit of a tertiary centre

Our study aims to present the preliminary findings of an ongoing prospective cohort study that assesses the trainees' ability to perform foetal biometry during the third trimester of pregnancy. Sixty-three women with third-trimester singleton pregnancies were included. A biometry scan was performed byboth residents and a foetal medicine specialist in the Third department of Obstetrics and Gynaecology of Attikon University Hospital. For each case, the ultrasonographic measurements of the two operator groups were compared. The mean difference of the resident group compared to the specialist group was: for the biparietal diameter +1.3 mm (CI 95%, range -10.6 to +13,3, ±1.96 SD), for the occipitofrontal diameter -2.6 mm (CI 95%, range -31.5 to +26.2), for the anterior-posterior abdominal diameter -2.6 mm (CI 95%, range -17.9 to +12.8), for the transverse abdominal diameter -0.7 mm (CI 95%, range -17.1 to +15.7) and for the femur length -1.1 mm (CI 95%, range -11.7 to +9.6). We observed that, among all biometric parameters, the most accurate -based on the specialist group were the head circumference measurements. The highest discrepancy was noted for the abdominal assessment. Given that foetal biometry is of utmost importance in obstetrical clinical evaluation and management, a study that highlights the weaknesses of residents in this field could open new horizons in optimising the learning procedure.Impact statementWhat is already known on this subject? After review of the literature, we found only a few studies on inter- and intra-observer discrepancy in foetal biometry measurements among specialists.What the results of this study add? To our knowledge, our study is the first to evaluate residents' capacity of performing a biometry scan, by comparing their measurements to those of MFM specialists.What the implications are of these findings to clinical practice and/or further research? The need for constant evaluation of residents is indisputable. Our study could help to improve their ultrasound skills by giving emphasis on residents' weaknesses. With further research on this subject, a standard system of evaluation could be formed and determine the duration and type of training required for each resident.

International standards for fetal brain structures based on serial ultrasound measurements from Fetal Growth Longitudinal Study of INTERGROWTH-21st Project

OBJECTIVE

To create prescriptive growth standards for five fetal brain structures, measured using ultrasound, in healthy, well-nourished women at low risk of impaired fetal growth and poor perinatal outcome, taking part in the Fetal Growth Longitudinal Study (FGLS) of the INTERGROWTH-21^st Project.

METHODS

This was a complementary analysis of a large, population-based, multicenter, longitudinal study. The sample analyzed was selected randomly from the overall FGLS population, ensuring an equal distribution among the eight diverse participating sites and of three-dimensional (3D) ultrasound volumes across pregnancy (range: 15-36 weeks' gestation). We measured, in planes reconstructed from 3D ultrasound volumes of the fetal head at different timepoints in pregnancy, the size of the parieto-occipital fissure (POF), Sylvian fissure (SF), anterior horn of the lateral ventricle, atrium of the posterior horn of the lateral ventricle (PV) and cisterna magna (CM). Fractional polynomials were used to construct the standards. Growth and development of the infants were assessed at 1 and 2 years of age to confirm their adequacy for constructing international standards.

RESULTS

From the entire FGLS cohort of 4321 women, 451 (10.4%) were selected at random. After exclusions, 3D ultrasound volumes from 442 fetuses born without a congenital malformation were used to create the charts. The fetal brain structures of interest were identified in 90% of cases. All structures, except the PV, showed increasing size with gestational age, and the size of the POF, SF, PV and CM showed increasing variability. The 3^rd , 5^th , 50^th , 95^th and 97^th smoothed centiles are presented. The 5^th centiles for the POF and SF were 3.1 mm and 4.7 mm at 22 weeks' gestation and 4.6 mm and 9.9 mm at 32 weeks, respectively. The 95^th centiles for the PV and CM were 8.5 mm and 7.5 mm at 22 weeks and 8.6 mm and 9.5 mm at 32 weeks, respectively.

CONCLUSIONS

We have produced prescriptive size standards for fetal brain structures based on prospectively enrolled pregnancies at low risk of abnormal outcome. We recommend these as international standards for the assessment of measurements obtained using ultrasound from fetal brain structures. © 2020 Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.

The management of error in ultrasound fetal growth monitoring

It is important to understand the uncertainty in fetal measurements when using them in the management of pregnancy. The aim of this essay is to provide background on errors and uncertainty, describing error sources and their potential impact, with guidance on improving accuracy. Errors can be systematic or random, arising from equipment, image plane selection, measurement method and caliper placement and influenced by image quality, training and experience. The uncertainty in measurements is larger than clinically significant differences in fetal size and growth. Errors can be reduced by implementing equipment acceptance testing, written procedures, training and audit.

Spatio-temporal visual attention modelling of standard biometry plane-finding navigation

We present a novel multi-task neural network called Temporal SonoEyeNet (TSEN) with a primary task to describe the visual navigation process of sonographers by learning to generate visual attention maps of ultrasound images around standard biometry planes of the fetal abdomen, head (trans-ventricular plane) and femur. TSEN has three components: a feature extractor, a temporal attention module (TAM), and an auxiliary video classification module (VCM). A soft dynamic time warping (sDTW) loss function is used to improve visual attention modelling. Variants of the model are trained on a dataset of 280 video clips, each containing one of the three biometry planes and lasting 3-7 seconds, with corresponding real-time recorded gaze tracking data of an experienced sonographer. We report the performances of the different variants of TSEN for visual attention prediction at standard biometry plane detection. The best model performance is achieved using bi-directional convolutional long-short term memory (biCLSTM) in both TAM and VCM, and it outperforms a previous spatial model on all static and dynamic saliency metrics. As an auxiliary task to validate the clinical relevance of the visual attention modelling, the predicted visual attention maps were used to guide standard biometry plane detection in consecutive US video frames. All spatio-temporal TSEN models achieve higher scores compared to a spatial-only baseline; the best performing TSEN model achieves F1 scores on these standard biometry planes of 83.7%, 89.9% and 81.1%, respectively.

[Training in fetal ultrasound biometry: Prospective assesment of Objective Structured Assessment of Ultrasound Skills (OSAUS) efficiency]

BACKGROUND

Fetal biometry quality directly influences obstetrical care relevance. However, obstetrician proficiencies are heterogeneous in particular during initial training.

OBJECTIVES

To assess the predictive value of OSAUS scale to identify operators with enough command to perform a valid estimation of fetal weight (EFW) (I). This study also assesses OSAUS intra-operator inter-exams variability (II) and pass/fail score relevancy (III).

METHODS

Lecturers in Nancy University Hospital assessed trainees' proficiency for EWF systematically and prospectively through OSAUS scale. The trainee assessment was performed right after the one of the senior operator (reference EFW) on three consecutive patients during standard care ultrasounds. To ensure variability in proficiency within the sample, previous practice was taken into account during enrollment ("novices" and "intermediates" for<20 and 20 past exams, respectively). Correlation between mean OSAUS and validity of EFW (a valid EFW was defined by a difference with the reference EWF<0.8 Z-score) and variability between consecutive assessments were assessed.

RESULTS

The study population was constituted of 8 "novice" and 8 "intermediate" trainees. Association between OSAUS and EFW validity was significant (P<0.03) (I). Intra-operator inter-exams variability was majored in the "novice" group (coefficients of variation were 25% vs. 10% in "novice" and "intermediate" group respectively) (II). Within the sample, specificity and positive predictive value of a pass/fail score OSAUS>3.5 to predict EFW validity were 77% and 71%, respectively (III).

CONCLUSION

A 3.5 OSAUS pass/fail score could provide a relevant threshold to estimate operator proficiency in assessing fetal biometry in an autonomous and secure way.

Are there specific factors associated with prenatally undiagnosed foetal macrosomia?

OBJECTIVE

Fetal macrosomia is known to increase maternal and neonatal complications, but 20%-50% of the macrosomic fetuses are prenatally undiagnosed. Our objective was to identify specific factors associated with undiagnosed fetal macrosomia in women without diabetes.

METHODS

Retrospective case-control study in a tertiary maternity unit between January 1st and December 31st, 2016. Inclusion of all women delivering after 37 weeks of a single live-born macrosomic infant, i.e., with a birth weight ≥ 90th percentile for gestational age (GA). Women with pre-existing or gestational diabetes were excluded. To identify specific factors associated with undiagnosed foetal macrosomia, we compared risk factors for macrosomia, maternal characteristics, father's body mass index (BMI) and prenatal follow up between two groups depending on whether macrosomia was prenatally diagnosed or not.

RESULTS

Among 428 macrosomic newborns, 224 (52.3 %) were prenatally undiagnosed. Known risk factors for macrosomia, maternal characteristics (such as low socio-economic level, low education level) and father's BMI were similar between the two groups. The prenatal follow up was comparable between the two groups. Ultrasound estimated foetal weight during the 3rd trimester was lower in the undiagnosed macrosomic foetuses compared to diagnosed macrosomic foetuses (2130±279 vs 2445±333, p<0.001).

CONCLUSIONS

No specific factor of undiagnosed macrosomia was identified, and women with prenatally undiagnosed fetal macrosomia had the same risk factors than women with diagnosed macrosomia. Our study suggests that our groups have different growth curves. This hypothesis has yet to be studied.