The October issue of Ultrasound in Obstetrics & Gynecology includes a prospective longitudinal study investigating anthropometric and cardiovascular parameter trajectories across 12 months postpartum in women at high risk for pre-eclampsia (PE), a secondary analysis of a multicenter randomized trial evaluating risk factors for PE in high-risk women receiving low-dose aspirin prophylaxis, an observational study assessing the accuracy of gestational-age estimation using an artificial intelligence system applied to non-targeted ultrasound sweeps, and a retrospective observational study investigating the potential for radiomics features to distinguish between high-grade and low-grade serous ovarian carcinoma and whether their incorporation into machine-learning models can provide additional discriminatory value over models based on conventional ultrasound.
Please see below a selection of articles from the October issue of the Journal chosen specially by the UOG team. To view all UOG content, become an ISUOG member today or login and upgrade.
Postpartum anthropometric and cardiovascular profile following hypertensive disorders of pregnancy: 12-month follow-up
Hypertensive disorders of pregnancy (HDP) are associated with hemodynamic abnormalities that may persist postpartum, but longitudinal data in women at high risk for pre-eclampsia (PE) across the first postpartum year are limited. In this prospective longitudinal study, Chen et al. compared anthropometric and cardiovascular parameters at 6–12 weeks postpartum between women at high risk for PE who did and did not subsequently develop HDP and low risk controls and assessed trajectories of these parameters across the 12-month postpartum period in high-risk women. At 6–12 weeks postpartum, high-risk-with-HDP women showed higher mean arterial pressure, left ventricular mass index, systemic vascular resistance and myocardial performance index, with more impaired global longitudinal strain, with most differences persisting to 12 months. High-risk-without-HDP women showed decreasing weight, waist circumference and waist-to-hip ratio as well as improving cardiac parameters from 6–12 weeks to 12 months postpartum, whilst these improvements were only noted at 12 months in the high-risk-with-HDP group. These findings suggest that women at high risk for PE, particularly those who develop HDP, may have suboptimal anthropometric and cardiovascular trajectories during the first postpartum year, supporting surveillance beyond the standard 6–12-week postpartum visit in this population.
Determinants of preterm pre-eclampsia following low-dose aspirin prophylaxis: evidence from stepped-wedge cluster randomized trial in Asia
Low-dose aspirin is recommended widely as an effective prophylactic intervention for pregnant women at high risk for preterm pre-eclampsia (PE), but discrepancies have been observed regarding its effectiveness across different populations. In this secondary analysis of a multicenter stepped-wedge cluster randomized trial, Lin et al. aimed to identify potential risk factors for preterm PE and any-onset PE in women identified as high-risk by the Fetal Medicine Foundation (FMF) triple test despite receiving low-dose aspirin. Chronic hypertension, higher mean arterial pressure and higher estimated risk were associated with preterm PE and earlier delivery complicated by PE among high-risk women, despite low-dose aspirin prophylaxis. These findings indicate risk factors that could be incorporated into a risk stratification framework to improve detection of residual PE risk in first-trimester screening and improve prevention strategies.
Real-time gestational-age estimation at 11–13 weeks using artificial intelligence on non-targeted ultrasound sweeps
Artificial intelligence (AI) models have demonstrated calculation of gestational-age (GA) estimates comparable to those based on conventional biometry, even in cases in which ultrasound images were acquired by novice operators. Eisenkolb et al. assessed the ability of an AI system applied to non-targeted ultrasound sweeps at 11–13 weeks' gestation to estimate GA, evaluating its performance against crown–rump length (CRL)-based estimates using mean absolute error (MAE) and Bland–Altman analysis. The MAE between AI- and CRL-based GA was 2.1 days for single measurements (however, this was reduced to 1.8 days by accounting for measurement error in CRL) but CRL-based estimates showed narrower limits of agreement and lower variability. This indicates that AI-based GA estimation performs well in the first trimester but is less precise than GA estimation using CRL measured by experienced sonographers.
Differentiating high- and low-grade serous ovarian carcinoma using radiomics: a pilot study
High-grade (HGSC) and low-grade (LGSC) serous ovarian carcinomas have different pathogenesis, response to chemotherapy and prognosis, underscoring the importance of accurately distinguishing between them. In this retrospective observational study, Ciccarone et al. aimed to identify ultrasound-based radiomics features that could be used to distinguish HGSC from invasive LGSC and to develop machine-learning models incorporating radiomics, clinical and conventional ultrasound features for this purpose. Distinct radiomics features between the two conditions were identified but models incorporating these features did not outperform those based on conventional ultrasound. This suggests that radiomics approaches provide limited additional value for reliable preoperative discrimination. Future research may be necessary to develop more accurate and clinically useful artificial intelligence-based predictive models.