Prediction of severe adverse neonatal outcomes at the second stage of labour using machine learning: a retrospective cohort study.

OBJECTIVE: To create a personalised machine learning model for prediction of severe adverse neonatal outcomes (SANO) during the second stage of labour. DESIGN: Retrospective Electronic-Medical-Record (EMR) -based study. POPULATION: A cohort of 73 868 singleton, term deliveries that reached the second stage of labour, including 1346 (1.8%) deliveries with SANO. METHODS: A gradient boosting model was created, analysing 21 million data points from antepartum features (e.g. gravidity and parity) gathered at admission to the delivery unit, and intrapartum data (e.g. cervical dilatation and effacement) gathered during the first stage of labour. Deliveries were allocated to high-risk and low-risk groups based on the Youden index to maximise sensitivity and specificity. MAIN OUTCOME MEASURES: SANO was defined as either umbilical cord pH levels ≤7.1 or 1-minute or 5-minute Apgar score ≤7. RESULTS: The model for prediction of SANO yielded an area under the receiver operating curve (AUC) of 0.761 (95% CI 0.748-0.774). A third of the cohort (33.5%, n = 24 721) were allocated to a high-risk group for SANO, which captured up to 72.1% of these cases (odds ratio 5.3, 95% CI 4.7-6.0; high-risk versus low-risk groups). CONCLUSIONS: Data acquired throughout the first stage of labour can be used to predict SANO during the second stage of labour using a machine learning model. Stratifying parturients at the beginning of the second stage of labour in a 'time out' session, can direct a personalised approach to management of this challenging aspect of labour, as well as improve allocation of staff and resources. TWEETABLE ABSTRACT: Personalised prediction score for severe adverse neonatal outcomes in labour using machine learning model.

Persistence of levator ani sonographic defect detected by three-dimensional transperineal sonography in primiparous women.

OBJECTIVES: Correlation of the sonographic appearance of levator ani muscle (LAM) injury soon after delivery with that at long-term follow-up has not been described fully. We aimed to compare results of three-dimensional (3D) transperineal sonographic (TPS) evaluation of the LAM from the period immediately postpartum with long-term follow-up, to determine whether sonographic findings persist over time. METHODS: Primiparous women (n = 210) who had been examined by 3D-TPS in a previous study to determine LAM trauma 24-72 hours after delivery were invited to participate in a follow-up examination 3-21 months postpartum. We included in this study only women who were not pregnant when approached and who had not given birth in the interim. LAM trauma was diagnosed with 3D-TPS when we observed discontinuity and distortion of the most anteromedial part of the pubovisceral muscle in the coronal C-plane or rendered image. Initial and follow-up 3D-TPS results were compared using Cohen's kappa test for inter-rater agreement. RESULTS: Among the 87 women included in this study we found strong correlation between earlier and later sonographic appearance of LAM: 17/21 women with a sonographic finding of LAM injury in the period immediately postpartum were positive in the follow-up examination, and only 2/66 women negative for LAM damage at the first examination were found to have sonographic evidence of LAM defect at follow-up (Cohen's kappa, 0.805 (95% CI, 0.656-0.954), P < 0.001). CONCLUSIONS: Our findings suggest that 3D-TPS of the LAM is a reliable examination. A sonographic finding of LAM defect identified in the period immediately postpartum persists months or years after delivery; therefore, this test may be performed following delivery, or may be delayed without impacting the result. It is likely that this sonographic defect represents real anatomical disruption and is not an imaging artifact.

Systematic examination of the fetal abdominal precordial veins: a cohort study.

OBJECTIVES: Examination of the fetal venous system is a necessary part of complete fetal organ scanning to confirm landmark anatomy, such as the ductus venosus and course of the umbilical veins, and, whenever cardiovascular anomalies are identified, to exclude associated anomalous development of the fetal veins. We aimed to develop a protocol for systematic examination of the fetal venous system during midtrimester targeted organ scanning. METHODS: We included low-risk women with a singleton fetus presenting between January 2011 and June 2013 to our center for routine midtrimester (20-24 weeks) targeted organ scanning. Imaging of the venous system was added to the booked scan and comprised two-dimensional color Doppler scanning of the fetal abdomen in three discrete planes, two transverse and one longitudinal. The more caudal plane was obtained in a ventral or lateral transverse abdominal plane to image the umbilical vein, left portal vein, portal sinus, anterior right portal vein, posterior right portal vein, main portal vein and splenic vein and artery. Moving cephalad, a ventral or lateral transverse plane was obtained to image the right, middle and left hepatic veins and inferior vena cava (IVC). Finally, a longitudinal anteroposterior plane showed the umbilical vein, ductus venosus, IVC and left hepatic vein. In some cases the pulsed Doppler waveform of a given target vessel was also examined. Three-dimensional/4D ultrasound was applied as necessary, when anomalous cases were encountered. RESULTS: We examined 1810 women. Their body mass index ranged from 19 to 40 (mean, 24.7). In 38 (2.1%) women, the target anatomy was not visualized satisfactorily owing to maternal body habitus. A T-shaped configuration of the portal system vessels was observed in 63% of cases, an X-shaped configuration in 25% and an H-shaped configuration in 12%. During the study period, 24 congenital anomalies of the precordial venous system were diagnosed: nine cases of persistent right umbilical vein, seven of agenesis of the ductus venosus, five of anomalous portal venous drainage and three of interrupted IVC with azygos continuation. CONCLUSIONS: Examination of the fetal venous system is feasible with the application of three abdominal planes. While a venous system scan is not practicable as part of a screening-level examination, mastery of the normal anatomy is an essential part of the professional knowledge base, in order to provide ready and complete scanning of the system in cases of suspected anomalies or disordered cardiac function.

Sonographic finding of postpartum levator ani muscle injury correlates with pelvic floor clinical examination.

OBJECTIVES: Correlation of the sonographic finding of levator ani muscle (LAM) injuries with clinical examination in primiparous women following vaginal delivery has not been fully described. We aimed to examine the correlation of three-dimensional transperineal ultrasound (3D-TPS) finding of LAM defects with results of clinical examination of the pelvic floor, at intermediate follow-up. METHODS: Subjects were primiparae 3-21 months following vaginal delivery, who had not become pregnant or delivered in the interim. On 3D-TPS, LAM trauma was diagnosed when discontinuity and distortion were visible in the most anteromedial part of the pubovisceral muscle in the coronal C-plane or rendered image. Clinical examination was performed by a physiotherapist who was blinded to the ultrasound results, and included palpation of the medial and lateral parts of the LAM mass, evaluation of tissue quality and whether there was any palpable gap. Muscle strength was evaluated using the modified Oxford scale. RESULTS: Eighty-seven women were included, 19 (21.8%) of whom were found to have a sonographic LAM injury. Oxford score palpation parameter of asymmetric muscle mass or texture was significantly correlated with the finding of a LAM defect: of 68 women with normal 3D-TPS, 22 (32.4%) were found to have asymmetry of muscle mass or tissue quality on clinical examination vs 12 (63.2%) of 19 women with sonographic evidence of LAM injury (P = 0.016). Muscle strength and endurance parameters did not significantly correlate with the 3D-TPS findings. CONCLUSION: Our findings suggest that persistent 3D-TPS LAM injury after primary vaginal delivery has clinical expression in changes in mass and texture of the LAM, as assessed by palpation.

Fetal tricuspid annular plane systolic excursion (f-TAPSE): evaluation of fetal right heart systolic function with conventional M-mode ultrasound and spatiotemporal image correlation (STIC) M-mode.

OBJECTIVES: Fetal tricuspid annular plane systolic excursion (f-TAPSE) is a modified method to measure the vertical movement of the tricuspid valve annulus by M-mode ultrasound, in order to assess the fetal right heart. Evaluation of right heart function is well-recognized in pediatric and adult cardiology, but has not been studied widely in the fetus. We aimed to study f-TAPSE in the second half of gestation in normal fetuses, to establish reference ranges for this measure, to evaluate the usefulness of spatiotemporal image correlation (STIC) M-mode in obtaining it, and to compare conventional M-mode and STIC M-mode-based measures of f-TAPSE. METHODS: We recruited gravidae presenting to our centers from 20 to 38 weeks for targeted organ scans, fetal echocardiography or third-trimester fetal surveillance, with structurally normal singleton fetuses and verified gestational age (GA). Because of the small number of subjects at the lower limit, fetuses at 20 and those at 21 weeks were combined into a single group ('21 weeks'). During the booked scan, in addition to standard biometry, M-mode was applied to the tricuspid annulus, parallel to the ventricular septum, and the amplitude of the resulting wave was measured. To allow comparison with STIC M-mode, a STIC volume was acquired and saved. In post-processing, the volume was rotated to show an apical four-chamber view, and f-TAPSE was investigated in a similar fashion to that used for conventional M-mode. Two to three measures of TAPSE were taken and the results averaged. In thirty women, measurements were performed by two observers and inter- and intraobserver variation were calculated. RESULTS: We examined 341 fetuses at GA 20-39 weeks. Conventional M-mode f-TAPSE values ranged from a mean of 3.6 (± 1.1) mm at 21 weeks to a mean of 8.6 (± 1.5) mm at 39 weeks. In 45 cases we were unable to perform conventional M-mode ultrasound because of fetal lie; in eight cases STIC volumes were found in post-processing to be unsuitable for analysis. STIC f-TAPSE values ranged from a mean of 4.2 (± 1.4) mm at 21 weeks to a mean of 8.3 (± 1.5) mm at 39 weeks. Scatterplots of f-TAPSE measures obtained with conventional M-mode and with STIC M-mode were created vs GA and estimated fetal weight (EFW). For both modalities, f-TAPSE increased linearly with GA and with EFW. Good correlation was found between the two methods (Pearson's R(2) = 0.904). No significant difference was found in mean or variance of the distributions or slopes of the regression equations. Inter- and intraobserver variation (intraclass correlation coefficient) in conventional M-mode and STIC M-mode f-TAPSE measures were 0.94 and 0.97, respectively. CONCLUSION: F-TAPSE in normal fetuses increases over the course of gestation and correlates to EFW. F-TAPSE measurement is easy to perform and available on all ultrasound machines; STIC f-TAPSE is possible on machines with STIC capability and produces similar measures with a greater success rate. We suggest the addition of f-TAPSE measurement to fetal right cardiac function evaluation.

Three-dimensional transperineal ultrasound findings associated with anal incontinence after intrapartum sphincter tears in primiparous women.

OBJECTIVE: Three-dimensional transperineal sonography (3D-TPS) performed in women following third- or fourth-degree intrapartum tears repaired with the overlapping technique demonstrates characteristic signs. The aim of this study was to investigate the correlation of these signs with clinical complaints of incontinence. METHODS: This was a prospective observational study. Sixty primiparous women underwent 3D-TPS 3-42 (mean, 10.6) months after surgical repair of third- or fourth-degree postpartum sphincter tears with the overlapping technique and were evaluated for clinical degree of incontinence using the St Mark's Incontinence Score (SMIS) questionnaire. The following signs were assessed on 3D-TPS: interruption of the internal anal sphincter or external anal sphincter, 'half moon' sign, changes in the mucosal folds and thickening of the external anal sphincter in the area of sphincter repair. As a comparison group, 27 primiparous women after normal vaginal delivery, without clinically recognized anal sphincter tears, were evaluated similarly, 3-37 (mean, 9.9) months postpartum. RESULTS: Abnormal sonographic signs were apparent in 35 (58.3%) of 60 women in the study group, and 39 (65%) of 60 had some clinical complaints of incontinence 3-42 months after delivery, most of a mild degree. Higher SMIS results were found in women of the study group than in those of the comparison group (mean (SD) 2.80 (0.481) vs. 1.15 (0.365); P = 0.018). The rates of incontinence were similar between the women in the study group with normal ultrasound findings and the women in the comparison group (9/25 vs. 10/27; relative risk (RR) = 0.97, 95% CI, 0.47-1.97). CONCLUSIONS: Sonographic signs of anal sphincter tear and repair had disappeared at follow-up examination in almost half of the patients, and therefore this examination should be deferred from the early postpartum period. A substantial proportion of women report some complaint of incontinence after sphincter repair, most of a slight degree. Such complaints are associated with abnormal 3D-TPS findings at follow up, while in women with a normal 3D-TPS scan the rate of incontinence complaints is similar to that in women after normal delivery.

Fetal heart ventricular mass obtained by STIC acquisition combined with inversion mode and VOCAL.

OBJECTIVE: Estimation of fetal heart ventricular mass is important for fetal cardiac evaluation in cases of structural or functional cardiac disorders or extracardiac factors. It may be used with other cardiac parameters to ascertain the severity and prognosis of such disorders, or the nature and timing of intervention. We applied a novel technique combining spatiotemporal image correlation (STIC) with three-dimensional inversion mode and Virtual Organ Computer-aided AnaLysis (VOCAL™) for fetal cardiac mass assessment in healthy fetuses in the second and third trimesters. METHODS: STIC acquisition was performed during fetal quiescence with the abdomen uppermost, at an angle of 30-50°, without color Doppler mapping. Myocardial volume measurements were performed in postprocessing using VOCAL mode, set to 15°. Beginning with the heart in four-chamber view at end diastole, a trace was drawn manually including the myocardium and interventricular septum. Inversion mode colors the intraventricular (anechoic, fluid-filled) voxels; this intraventricular volume was subtracted automatically from the total. Mass was determined by multiplying the result by the estimated fetal myocardial density (1.050 g/cm(3) ). The process was repeated for right and left ventricles. RESULTS: Data from 106 fetuses at 21-38 weeks' gestation were obtained and scatterplots of fetal cardiac ventricular mass distribution were created. Several cases of fetuses with disordered cardiac ventricle (supraventricular tachycardia, hypoplastic left heart syndrome, dilated cardiomyopathy, twin-to-twin transfusion syndrome, Ebstein anomaly, non-immune hydrops fetalis, septate right atrium and diaphragmatic hernia) were examined. Ventricular mass parameters were markedly affected as compared with normal cases of similar gestational age. CONCLUSIONS: STIC acquisition combined with inversion mode and VOCAL is a feasible method of cardiac ventricular mass quantification. This methodology may have added value in fetal cardiac evaluation in cases of anatomic malformation or cardiac dysfunction, or in cases of maternal diabetes.

Added value of three-/four-dimensional ultrasound in offline analysis and diagnosis of congenital heart disease.

OBJECTIVE: Many published studies have shown that application of three-dimensional (3D) and real-time 3D (4D) ultrasound modalities can improve certain aspects of fetal echocardiography, but have left open the question of whether these modalities improved the accuracy of prenatal detection of anatomical fetal cardiovascular malformations. We aimed to determine whether 3D/4D ultrasound improved diagnostic ability in cases of congenital heart disease (CHD). METHODS: Women who attended for early- or midtrimester targeted organ scans had complete fetal echocardiography according to our five-planes protocol, as well as examination of the ductus venosus and longitudinal aortic arch planes, performed with 2D ultrasound combined with 2D color Doppler, spatiotemporal image correlation (STIC), STIC with color Doppler, and STIC with B-flow. Ultrasound data of cases of CHD were stored in a dedicated archive. Stored cases were anonymized and the list order was randomized. Stored 2D ultrasound cineloops and 4D ultrasound volumes were reviewed separately according to a standardized table of 23 specified structures on five required planes of visualization: the upper abdomen, four-chamber view, five-chamber view, pulmonary artery bifurcation view, and three vessels and trachea plane. Separate diagnoses were recorded and finally compared. Diagnoses were confirmed by pathological examination or neonatal echocardiography. RESULTS: During the study period, 13 101 examinations were performed; 181 diagnoses of CHD were made. In 12 of these, 3D/4D ultrasound added to the accuracy of our diagnosis: one right aortic arch with anomalous branching; one transposition of the great arteries with pulmonary atresia diagnosed with tomographic ultrasound imaging (TUI); one segmental interrupted aortic arch diagnosed with TUI; one right ventricle aneurysm diagnosed with B-flow; two agenesis of ductus venosus to the coronary sinus diagnosed by multiplanar reconstruction (MPR) and B-flow; two total anomalous pulmonary venous connection diagnosed with MPR; and four ventricular septal defect (VSD) diagnosed with the aid of virtual planes. There were 12 missed diagnoses and no false-positive results. CONCLUSIONS: 3D/4D ultrasound modalities may have advantages in some aspects of fetal cardiovascular evaluation, however, overall 3D/4D ultrasound modalities had added value in only about 6% of cases of fetal anatomical cardiovascular anomalies.

Shunt diameter in agenesis of the ductus venosus with extrahepatic portosystemic shunt impacts on prognosis.

OBJECTIVES: Agenesis of the ductus venosus (ADV) is a rare condition in which there are two variants of umbilical vein drainage: intrahepatic shunt or extrahepatic (portosystemic) shunt. It has been posited that the extrahepatic variant carries a poorer prognosis. However, in the absence of associated anomalies there is still a wide variation in outcome. We evaluated the portal system in cases of ADV and aimed to identify parameters that might predict outcome. METHODS: We conducted a retrospective study of cases of ADV with extrahepatic shunt that had been examined in two centers, and collected new cases prospectively. The route of the shunt was depicted using two-dimensional (2D) and three-dimensional (3D) ultrasound imaging. In an attempt to characterize portal system and shunt developmental variations and their possible impact on outcome, the diameter of the shunt was compared with the diameter of the umbilical vein and the entire portal vasculature was assessed. Poor outcome was defined as persistent morbidity or fetal or neonatal death. RESULTS: Twenty-two cases of ADV were identified: nine retrospectively and 13 prospectively. Aberrant shunts from the umbilical vein were identified to the right atrium, coronary sinus, inferior vena cava (IVC) and iliac vein. In seven of 22 cases (32%) a wide connection was observed. In six of these seven cases (86%) the outcome was poor, including four with severe associated anomalies and two with hepatic dysfunction. In five of these cases, cardiomegaly with tricuspid regurgitation was observed, as well as underdevelopment of the portal system. In only five of 15 cases with a narrow shunt (33%) was the outcome poor, and in all five cases the poor outcome was related to associated anomalies. CONCLUSION: In cases of ADV with extrahepatic shunt, portal system development is impacted by the diameter of the shunt. If the shunt is narrow, the portal system will have developed normally. A wide shunt is associated with underdevelopment or absence of the portal system. In cases of ADV with extrahepatic shunt, prognosis is determined by the severity of associated anomalies, the diameter of the shunt and development of the portal system.

Added value of the gray-scale whirlpool sign in the diagnosis of adnexal torsion.

OBJECTIVE: Adnexal torsion is a common gynecologic emergency affecting females of all ages. Expedient diagnosis and treatment are important, particularly in young fertile patients to preserve ovarian viability. Classical parameters for the clinical and sonographic diagnosis of adnexal torsion have very high false-positive rates, approaching 50%. The sonographic 'whirlpool' sign has been shown to be effective for visualizing the torsed part in the prenatal diagnosis of malrotation of the midgut with volvulus, as well as scrotal and ovarian torsion. We aimed to evaluate the efficacy of the whirlpool sign in the diagnosis of torsion, as compared to a protocol based on 'classic' sonographic signs of torsion alone. METHODS: This was a retrospective chart review. Files of all patients who underwent laparoscopy for suspected torsion at our center between January 2006 and May 2009 were extracted and reviewed. Ultrasound reports were retrieved from our computerized database. Patients were assigned to the study group if the whirlpool sign had been investigated during pre-procedural evaluation or to the control group if only the standard protocol had been applied. RESULTS: Eighty women were referred for laparoscopy for suspected adnexal torsion during the study period. In 22 women the ultrasound investigation had included the whirlpool sign (study group) while 58 had been examined by the standard protocol (control group). Twenty women in the study group had a positive whirlpool sign on ultrasound, 18 of whom (90.0%) had confirmed torsion on laparoscopy. In the control group 32 of 58 (55.2%) women had confirmed torsion on laparoscopy. CONCLUSION: The addition of the sonographic whirlpool sign to the preoperative sonographic evaluation of patients with suspected torsion appears to improve the rate of true-positive diagnoses as confirmed by laparoscopy. Copyright © 2010 ISUOG. Published by John Wiley & Sons, Ltd.