OBJECTIVE: In this study we use transperineal ultrasound to identify how many women will achieve a normal vaginal delivery without substantial damage to the levator ani or anal sphincter muscles, and to create a model to predict patient characteristics associated with successful atraumatic normal vaginal delivery.
STUDY DESIGN: This is a retrospective, secondary analysis of data sets gathered in the context of an interventional perinatal imaging study. A total of 660 primiparas, carrying an uncomplicated singleton pregnancy, underwent an antepartum and postpartum interview, vaginal exam (Pelvic Organ Prolapse Quantification), and 4-dimensional translabial ultrasound. Ultrasound data were analyzed for levator trauma and/or overdistention and residual sphincter defects. Postprocessing analysis of ultrasound volumes was performed blinded against clinical data and analyzed against obstetric data retrieved from the local maternity database. Levator avulsion was diagnosed if the muscle insertion at the inferior pubic ramus at the plane of minimal hiatal dimensions and within 5 mm above this plane on tomographic ultrasound imaging was abnormal, ie the muscle was disconnected from the inferior pubic ramus. Hiatal overdistensibility (microtrauma) was diagnosed if there was a peripartum increase in hiatal area on Valsalva by >20% with the resultant area ≥25 cm2. A sphincter defect was diagnosed if a gap of >30 degrees was seen in ≥4 of 6 tomographic ultrasound imaging slices bracketing the external anal sphincter. Two models were tested: a first model that defines severe pelvic floor trauma as either obstetric anal sphincter injury or levator avulsion, and a second, more conservative model, that also included microtrauma.
RESULTS: A total of 504/660 women (76%) returned for postpartum follow-up as described previously. In all, 21 patients were excluded due to inadequate data or intercurrent pregnancy, leaving 483 women for analysis. Model 1 defined nontraumatic vaginal delivery as excluding operative delivery, obstetric anal sphincter injuries, and sonographic evidence of levator avulsion or residual sphincter defect. Model 2 also excluded microtrauma. Of 483 women, 112 (23%) had a cesarean delivery, 103 (21%) had an operative vaginal delivery, and 17 (4%) had a third-/fourth-degree tear, leaving 251 women who could be said to have had a normal vaginal delivery. On ultrasound, in model 1, 27 women (6%) had an avulsion and 31 (6%) had a residual defect, leaving 193/483 (40%) who met the criteria for atraumatic normal vaginal delivery. In model 2, an additional 33 women (7%) had microtrauma, leaving only 160/483 (33%) women who met the criteria for atraumatic normal vaginal delivery. On multivariate analysis, younger age and earlier gestation at time of delivery remained highly significant as predictors of atraumatic normal vaginal delivery in both models, with increased hiatal area on Valsalva also significant in model 2 (all P ≤ .035).
CONCLUSION: The prevalence of significant pelvic floor trauma after vaginal child birth is much higher than generally assumed. Rates of obstetric anal sphincter injury are often underestimated and levator avulsion is not included as a consequence of vaginal birth in most obstetric text books. In this study less than half (33-40%) of primiparous women achieved an atraumatic normal vaginal delivery.
MATERIAL AND METHODS: A retrospective observational study utilizing data from two studies with identical inclusion criteria and assessment protocols between 2005 and 2014. A total of 1148 primiparae with an uncomplicated singleton pregnancy were recruited and assessed with translabial ultrasound at 36 weeks antepartum and 871 (76%) returned for reassessment 3-6 months postpartum. The ultrasound data of vaginally parous women were analyzed for levator avulsion and microtrauma. The former was diagnosed if the muscle insertion at the inferior pubic ramus in the plane of minimal hiatal dimensions and within 5 mm above were abnormal on tomographic ultrasound imaging. Microtrauma was diagnosed in women with an intact levator and if there was a postpartum increase in hiatal area on Valsalva by >20% with the resultant area ≥25 cm2 .
RESULTS: The complete datasets of 844 women were analyzed. Among them, 609 delivered vaginally: by normal vaginal delivery in 452 (54%), a vacuum birth in 102 (12%) and a forceps delivery in 55 (6%). Levator avulsion was diagnosed in 98 and microtrauma in 97. On multivariate analysis, increasing maternal age, lower body mass index and lower bladder neck descent were associated with avulsion. Increased bladder neck descent and a family history of cesarean section (CS) were associcated with microtrauma.
CONCLUSIONS: Maternal age, body mass index, bladder neck descent and family history of CS are antenatal predictors for levator trauma.
METHODS: This was a retrospective ancillary analysis involving 660 nulliparous women carrying an uncomplicated singleton term pregnancy in a prospective perinatal intervention trial at two Australian tertiary obstetric units. They had been seen antenatally and at 3-6 months postpartum for a standardized clinical assessment between 2007 and 2014. Primary outcome measures were sonographically diagnosed LAM and external anal sphincter (EAS) trauma.
RESULTS: The incidence of LAM avulsion (11.5% vs. 21.3%, P = 0.01) and composite trauma, i.e., LAM avulsion ± EAS injury (29.2% vs. 39.7%, P = 0.03) were higher in one of the two hospitals, where the forceps delivery rate was also higher (10.9% vs. 2.6%, P
RESULTS: The oocyte AI model demonstrated area under the curve (AUC) up to 0.65 on two blind test datasets. High sensitivity for predicting competent oocytes (83-88%) was offset by lower specificity (26-36%). Exclusion of confounding biological variables (male factor infertility and maternal age ≥35 years) improved AUC up to 14%, primarily due to increased specificity. AI score correlated with size of the zona pellucida and perivitelline space, and ooplasm appearance. AI score also correlated with blastocyst expansion grade and morphological quality. The sum of AI scores from oocytes in group culture images predicted the formation of two or more usable blastocysts (AUC 0.77).
CONCLUSION: An AI model to evaluate oocyte competence was developed using federated learning, representing an essential step in protecting patient data. The AI model was significantly predictive of oocyte competence, as defined by usable blastocyst formation, which is a critical factor for IVF success. Potential clinical utility ranges from selective oocyte fertilization to guiding treatment decisions regarding additional rounds of oocyte retrieval.
DESIGN: In total, 10,677 oocyte images with associated metadata were collected prospectively by eight IVF clinics across six countries. AI training used federated learning, where data were retained on regional servers to comply with data privacy laws. The final AI model required a single image as input to evaluate oocyte competence, which was defined by the formation of a usable blastocyst (≥expansion grade 3 by day 5 or 6 post ICSI).