Search

Objectives: To develop a deep learning (DL)-model using convolutional neural networks (CNN) to automatically identify the fetal head position at transperineal ultrasound in the second stage of labor., Material and Methods: Prospective, multicenter study including singleton, term, cephalic pregnancies in the second stage of labor. We assessed the fetal head position using transabdominal ultrasound and subsequently, obtained an image of the fetal head on the axial plane using transperineal ultrasound and labeled it according to the transabdominal ultrasound findings. The ultrasound images were randomly allocated into the three datasets containing a similar proportion of images of each subtype of fetal head position (occiput anterior, posterior, right and left transverse): the training dataset included 70 %, the validation dataset 15 %, and the testing dataset 15 % of the acquired images. The pre-trained ResNet18 model was employed as a foundational framework for feature extraction and classification. CNN 1 was trained to differentiate between occiput anterior (OA) and non-OA positions, CNN 2 classified fetal head malpositions into occiput posterior (OP) or occiput transverse (OT) position, and CNN 3 classified the remaining images as right or left OT. The DL-model was constructed using three convolutional neural networks (CNN) working simultaneously for the classification of fetal head positions. The performance of the algorithm was evaluated in terms of accuracy, sensitivity, specificity, F1-score and Cohen's kappa., Results: Between February 2018 and May 2023, 2154 transperineal images were included from eligible participants across 16 collaborating centers. The overall performance of the model for the classification of the fetal head position in the axial plane at transperineal ultrasound was excellent, with an of 94.5 % (95 % CI 92.0--97.0), a sensitivity of 95.6 % (95 % CI 96.8-100.0), a specificity of 91.2 % (95 % CI 87.3-95.1), a F1-score of 0.92 and a Cohen's kappa of 0.90. The best performance was achieved by the CNN 1 - OA position vs fetal head malpositions - with an accuracy of 98.3 % (95 % CI 96.9-99.7), followed by CNN 2 - OP vs OT positions - with an accuracy of 93.9 % (95 % CI 89.6-98.2), and finally, CNN 3 - right vs left OT position - with an accuracy of 91.3 % (95 % CI 83.5-99.1)., Conclusions: We have developed a DL-model capable of assessing fetal head position using transperineal ultrasound during the second stage of labor with an excellent overall accuracy. Future studies should validate our DL model using larger datasets and real-time patients before introducing it into routine clinical practice., Competing Interests: Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2024 The Author(s). Published by Elsevier B.V. All rights reserved.)

Objective: To investigate the relationship of umbilical vein flow (UVF) measured close to term with abnormal fetal growth and adverse perinatal outcome in a cohort of pregnancies at low risk of placental insufficiency., Methods: This was a prospective multicenter observational study conducted across two tertiary maternity units. Patients with a singleton appropriate-for-gestational-age fetus between 35 and 38 weeks' gestation were included. Pregnancies at higher risk of placental insufficiency or with fetal anomalies were excluded. At ultrasound examination, the abdominal circumference (AC), umbilical vein diameter and peak velocity of the umbilical vein were measured, and, using these variables, a new variable, UVF/AC, was calculated. The primary outcome was the occurrence of severely stunted fetal growth, defined as a greater than 40-percentile drop between estimated fetal weight at the third-trimester ultrasound and birth weight. The occurrence of adverse perinatal outcome (defined as one of the following: neonatal acidosis (umbilical artery pH < 7.15 and/or base excess > 12 mmol/L) at birth, 5-min Apgar score < 7, neonatal resuscitation or neonatal intensive care unit admission) was analyzed as a secondary outcome., Results: Between April 2021 and March 2023, 365 women were included in the study. The mean UVF/AC at enrolment was 6.4 ± 2.6 mL/min/cm, and 35 (9.6%) cases were affected by severely stunted fetal growth. Severely stunted fetal growth was associated with a lower mean UVF/AC (5.4 ± 2.6 vs 6.5 ± 2.6 mL/min/cm; P = 0.02) and a higher frequency of UVF/AC < 10 th percentile (8/35 (22.9%) vs 28/330 (8.5%); P = 0.01). Moreover, UVF/AC showed an area under the receiver-operating-characteristics curve (AUC) of 0.65 (95% CI, 0.55-0.75; P = 0.004) in predicting the occurrence of severely stunted fetal growth, and the optimal cut-off value of UVF/AC for discriminating between normal and severely stunted fetal growth was 7.2 mL/min/cm. This value was associated with a sensitivity and specificity of 0.77 (95% CI, 0.60-0.90) and 0.33 (95% CI, 0.28-0.39), and positive and negative predictive values of 0.11 (95% CI, 0.07-0.15) and 0.93 (95% CI, 0.87-0.97), respectively. Regarding the occurrence of adverse perinatal outcome, this was associated independently with maternal age (adjusted odds ratio (aOR), 0.93 (95% CI, 0.87-0.99); P = 0.04), UVF/AC Z-score (aOR, 0.53 (95% CI, 0.30-0.87); P = 0.01) and augmentation of labor (aOR, 2.69 (95% CI, 1.28-5.69); P = 0.009). UVF/AC showed an AUC of 0.65 (95% CI, 0.56-0.73; P = 0.005) in predicting the occurrence of adverse perinatal outcome, and the optimal cut-off value of UVF/AC for discriminating between normal and adverse perinatal outcome was 6.7 mL/min/cm. This value was associated with a sensitivity and specificity of 0.70 (95% CI, 0.54-0.83) and 0.40 (95% CI, 0.34-0.45), and positive and negative predictive values of 0.14 (95% CI, 0.09-0.19) and 0.91 (95% CI, 0.85-0.95), respectively., Conclusions: Our data demonstrate an association between reduced UVF close to term, severely stunted fetal growth and adverse perinatal outcome in a cohort of low-risk pregnant women, with a moderate ability to rule out and a poor ability to rule in either outcome. Further studies are needed to establish whether the assessment of UVF can improve the identification of fetuses at risk of subclinical placental insufficiency and adverse perinatal outcome. © 2023 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology., (© 2023 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.)

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., (© 2022 International Society of Ultrasound in Obstetrics and Gynecology.)

Objectives: To evaluate the role of the cervical sliding sign (CSS) in the prediction of the outcome of induction of labor (IOL)., Study Design: Two-center prospective observational cohort study involving a non-consecutive series of uncomplicated singleton term pregnancies, planned for IOL, with a fetus in cephalic presentation and unfavorable cervix as defined by a Bishop score ≤ 6. The Bishop score was evaluated by transvaginal digital examination and the cervical length and CSS by transvaginal ultrasound. The presence of CSS was defined as the sliding of the anterior cervical lip on the posterior one under gentle pressure of the transvaginal probe. The primary outcome of the study was successful vaginal delivery within 24 h. The secondary outcome was the induction-to-active-labor time. The interobserver agreement for the CSS was also evaluated., Results: Over a period of 12 months, 179 women were included. The CSS was found in 86 (48.0 %) patients and was associated with an increased likelihood of vaginal delivery within 24 h (60/86 or 69.8 % vs 27/93 or 29.0 %, P < 0.001) and a shorter induction-to-active-labor time (954 ± 618 min vs 1416 ± 660 min, P < 0.001). Multivariable regression analysis showed that the CSS was an independent predictor of vaginal delivery within 24 h (aOR 5.37, 95 % CI 2.26-12.75) and shorter induction-to-active-labor time interval (HR 1.81, 95 % CI 1.19-2.74). The interobserver variability based on intraclass correlation coefficient for the CSS was excellent (ICC = 0.90)., Conclusion: In women undergoing IOL with an unfavorable cervix, the CSS is associated with a higher frequency of vaginal delivery within 24 h and a shorter induction-to-active-labor time., Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (Copyright © 2022 Elsevier B.V. All rights reserved.)

Objective: To describe the incidence, clinical features and perinatal outcome of late-onset fetal growth restriction (FGR) associated with genetic syndrome or aneuploidy, structural malformation or congenital infection., Methods: This was a retrospective multicenter cohort study of patients who attended one of four tertiary maternity hospitals in Italy. We included consecutive singleton pregnancies between 32 + 0 and 36 + 6 weeks' gestation with either fetal abdominal circumference (AC) or estimated fetal weight < 10 th percentile for gestational age or a reduction in AC of > 50 percentiles from the measurement at an ultrasound scan performed between 18 and 32 weeks. The study group consisted of pregnancies with late-onset FGR and a genetic syndrome or aneuploidy, structural malformation or congenital infection (anomalous late-onset FGR). The presence of congenital anomalies was ascertained postnatally in neonates with abnormal findings on antenatal investigation or detected after birth. The control group consisted of pregnancies with structurally and genetically normal fetuses with late-onset FGR. Composite adverse perinatal outcome was defined as the presence of at least one of stillbirth, 5-min Apgar score < 7, admission to the neonatal intensive care unit (NICU), need for respiratory support at birth, neonatal jaundice and neonatal hypoglycemia. The primary aims of the study were to assess the incidence and clinical features of anomalous late-onset FGR, and to compare the perinatal outcome of such cases with that of fetuses with non-anomalous late-onset FGR., Results: Overall, 1246 pregnancies complicated by late-onset FGR were included in the study, of which 120 (9.6%) were allocated to the anomalous late-onset FGR group. Of these, 11 (9.2%) had a genetic syndrome or aneuploidy, 105 (87.5%) had an isolated structural malformation, and four (3.3%) had a congenital infection. The most frequent structural defects associated with late-onset anomalous FGR were genitourinary malformations (28/105 (26.7%)) and limb malformation (21/105 (20.0%)). Compared with the non-anomalous late-onset FGR group, fetuses with anomalous late-onset FGR had an increased incidence of composite adverse perinatal outcome (35.9% vs 58.3%; P < 0.01). Newborns with anomalous, compared to those with non-anomalous, late-onset FGR showed a higher frequency of need for respiratory support at birth (25.8% vs 9.0%; P < 0.01), intubation (10.0% vs 1.1%; P < 0.01), NICU admission (43.3% vs 22.6%; P < 0.01) and longer hospital stay (median, 24 days (range, 4-250 days) vs 11 days (range, 2-59 days); P < 0.01)., Conclusions: Most pregnancies complicated by anomalous late-onset FGR have structural malformations rather than genetic abnormality or infection. Fetuses with anomalous late-onset FGR have an increased incidence of complications at birth and NICU admission and a longer hospital stay compared with fetuses with isolated late-onset FGR. © 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology., (© 2022 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.)

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 F 1 -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 F 1 -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., (© 2021 International Society of Ultrasound in Obstetrics and Gynecology.)

Objective: The objective of this study was to assess the position of the conus medullaris (CM) at the first trimester 3D ultrasound in a cohort of structurally normal fetuses., Methods: This was a multicenter prospective study involving a consecutive series of structurally normal fetuses between 11 and 13 weeks of gestation (CRL between 45 and 84 mm). All fetuses were submitted to 3D transvaginal ultrasound using a sagittal view of the spine as the starting plane of acquisition. At offline analysis, the position of the CM was evaluated by 2 independent operators with a quantitative and a qualitative method: (1) the distance between the most caudal part of the CM and the distal end of the coccyx (CMCd) was measured; (2) a line perpendicular to the fetal spine joining the tip of the CM to the anterior abdominal wall was traced to determine the level of this line in relation to the umbilical cord insertion (conus to abdomen line, CAL). Interobserver agreement for the CCMd was evaluated. Linear regression analysis was used to determine the association between the CMCd and CRL, and a normal range was computed based on the best-fit model. The absence of congenital anomalies was confirmed in all cases after birth., Results: In the study period between December 2019 and March 2020, 143 fetuses were recruited. In 130 fetuses (90.9%), the visualization of the CM was feasible. The mean value of the CMCd was 1.09 ± 0.16 cm. The 95% limits of agreement for the interobserver variability in measurement of the CMCd were 0.24 and 0.26 cm. The interobserver variability based on the intra-class correlation coefficient (ICC) for the CCMd was good (ICC = 0.81). We found a positive linear relationship between the CCMd and CRL. In all these fetuses, the CAL encountered the abdominal wall at or above the level of the cord insertion., Conclusion: In normal fetuses, the assessment of the CM position is feasible at the first trimester 3D ultrasound with a good interobserver agreement. The CM level was never found below the fetal umbilical cord insertion, while the CMCd was noted to increase according to the gestational age, confirming the "ascension" of the CM during fetal life., (© 2021 S. Karger AG, Basel.)