Identification and treatment of pregnancies at risk for preterm birth is a central challenge in obstetric research. Many of the known causes of preterm birth originate from mechanical failure in reproductive tissues. To better understand the biomechanical environment of the gravid uterus and its potential contribution to preterm birth, this computational study presents a parametric method for modeling maternal reproductive anatomy during the early second trimester. A finite element modeling approach was built using existing sonographic measurements from early second-trimester maternal anatomy and material properties from published mechanical tests. We applied the same physiologically relevant intrauterine pressure to all models and quantified the resulting tissue stretch. The sensitivity of the stretch in the proximal cervix was explored by varying material properties and sonographic maternal anatomy dimensions. Cervical material properties, particularly the fiber stiffness modulus and ground substance Young's modulus, were found to have the greatest effect on proximal cervix stretch compared to other material properties and sonographic dimensions. Among the sonographic dimension measurements, those defining the region surrounding the proximal cervix had the greatest effect on proximal cervix stretch, including the curvature of the posterior uterine wall and the thickness of the lower uterine segment. The computational modeling approach presented here enables future patient-specific studies of gravid reproductive tissues to elucidate differences between individuals who do and do not deliver preterm. Additionally, this study is foundational for building digital twins to support future virtual clinical studies on diagnostic and therapeutic device design to prevent preterm birth.

The uterine cervix is a soft biological tissue with critical biomechanical functions in pregnancy. It is a mechanical barrier that supports the growing fetus. As pregnancy progresses, the cervix becomes more compliant and eventually opens in late pregnancy to facilitate childbirth. This dual function is facilitated by extensive remodeling of the cervical extracellular matrix (ECM), giving rise to its complex time-dependent material properties. Premature cervical remodeling is known to result in preterm birth, defined as birth before 37 weeks of gestation. While previous work has studied cervical remodeling using various biomechanical methods, it remains unclear how the intrinsic or flow-independent viscoelastic behavior of the cervix is influenced by cervical remodeling. In this study, an anisotropic reactive viscoelastic material model was formulated and investigated under tensile deformation to understand material behavior in cervical remodeling. To calibrate the model, experimental force relaxation data was used from uniaxial tension tests on Rhesus macaque cervical specimens from four gestational time points. The results showed that cervical tissue equilibrium and instantaneous stiffness significantly decreased from the nonpregnant (NP) to the late pregnancy status. In addition, cervical tissue in the late third trimester relaxed faster to equilibrium than the other gestational groups, particularly at prescribed grip-to-grip strains greater than 30%. This fast relaxation to equilibrium helps the cervix dissipate tensile hoop stresses induced by the fetus during labor, preventing its rupture. This work provides insights into time-dependent cervical remodeling features, which are crucial for developing diagnostic methods and treatments for preterm birth.

Objective Researchers and clinicians have called for early pregnancy biomarkers of pre-term delivery, especially in the uterine cervix, a crucial player in delivery timing. The goal of this study was to develop practical biomarkers that can help predict the timing of delivery using ultrasound images. Here we investigated first-order speckle statistics as a lens into the remodeling microstructure of the cervix throughout gestation. Methods Thirty low-risk pregnant subjects were analyzed. Transvaginal ultrasound exams with a prototype linear array transducer were performed five times throughout pregnancy and once post-partum. Parameters of the Nakagami and homodyned K distributions were estimated from the envelopes of raw radiofrequency signals at a range of beam steering angles. The steering angle dependence of these parameters was also analyzed to quantify changes in microstructural anisotropy. Linear mixed effects modeling was performed for each speckle statistics metric, using subject age, parity, location of measurement within the cervix and time remaining to delivery as predictors. Results The metric most influenced by delivery timing was the steering angle dependence of the Nakagami m parameter in the posterior cervix near the internal os, where the linear mixed effects model showed an increase of 1.9±0.5% per wk in the variation of m with respect to the steering angle compared with a variation of m with respect to the steering angle at 40 wk from delivery (p<0.05). All other speckle statistics metrics and their steering angle dependence also had statistically significant relationships with delivery timing. Subject age and parity had a negative correlation with speckle statistics anisotropy metrics, and measurement location (anterior or posterior cervix) was a statistically significant predictor of speckle statistics metrics. Conclusion This investigation was the first to present an analysis of speckle statistics in the in vivo human cervix in a longitudinal pregnancy study. It was shown that speckle statistics parameter estimates are sensitive to changes in the cervix as it progresses toward delivery, suggesting the clinical potential of these tools for detecting microstructural remodeling in the cervix.

The uterus is central to the establishment, maintenance, and delivery of a healthy pregnancy. Biomechanics is an important contributor to pregnancy success, and alterations to normal uterine biomechanical functions can contribute to an array of obstetric pathologies. Few studies have characterized the passive mechanical properties of the gravid human uterus, and ethical limitations have largely prevented the investigation of mid-gestation periods. To address this key knowledge gap, this study seeks to characterize the structural, compositional, and time-dependent micro-mechanical properties of the nonhuman primate (NHP) uterine layers in nonpregnancy and at three time-points in pregnancy: early 2nd, early 3rd, and late 3rd trimesters. Distinct material and compositional properties were noted across the different tissue layers, with the nonpregnant endometrium and pregnant decidua being the least stiff, most viscous, least diffusible, and most hydrated layers of the NHP uterus. Pregnancy induced notable compositional and structural changes in the decidua and myometrium but had no effect on their micro-mechanical properties. Further comparison to published human data revealed marked similarities across species, with minor differences noted for the perimetrium and nonpregnant endometrium. This work provides insights into the material properties of the NHP uterus and demonstrates the validity of NHPs as a model for studying certain aspects of human uterine biomechanics.

Preterm birth (PTB) is the leading cause of perinatal death, affecting 10% of pregnancies. Currently, transvaginal ultrasound (TVUS) measurement of cervical length (CL) is the sole quantitative imaging metric for PTB risk, but offers limited predictive value. While computational models of cervical biomechanics show promise as PTB risk predictors, they require precise clinician-provided measurements. AI-enabled ultrasound segmentation offers a solution by automatically extracting anatomical features, thus addressing the labeling bottleneck. This study utilizes an ensemble of deep learning-based multi-class segmentation models trained on diverse TVUS data (N = 246) and evaluated on an out-of-distribution dataset (N = 29). High agreement (Dice metric ~ 0.8) between expert and model labels demonstrates the utility of AI tools in accurately measuring cervical geometry. Ultimately, this can enhance biomechanical models and more sophisticated AI-based models to better predict birth timing, specifically targeting PTB risk.

PURPOSE:Speckle statistics are fundamentally related to the resolution of an ultrasound image. Existing models for speckle statistics do not account for changes in resolution with imaging depth, a reality of clinical pulse-echo ultrasound, thereby posing challenges to interpretation. The purpose of this work is to evaluate and address this shortfall in first-order speckle statistics analysis. METHODS:Simulated ultrasonic speckle from a low scatterer density medium was created with known acquisition parameters, and speckle statistics of the Nakagami and homodyned K distributions were estimated with and without compensating for the expected change in the resolution cell size, defined by the ultrasound pulse volume, over the field of view. Compensation methods using resolution estimates from the predicted acoustic field, image autocorrelation, or spectral analysis were compared. The absolute number of scatterers per cubic millimeter was also calculated from corrected speckle statistics estimates. The results were validated with experiments in a low scatterer density phantom using a clinical scanner, and in in vivo rhesus macaque cervix and human Achilles tendon images. RESULTS:It was shown in both simulations and phantoms that, when no compensation was applied, the expansion of the resolution cell size caused a saturation of Nakagami m and homodyned K parameter estimates at depths beyond the focal distance, even when scatterer concentration was constant throughout the depth of the phantoms. This confounding factor was reduced by compensating for the changing resolution cell size, resulting in a decrease in the normalized root mean squared errors between estimates at focus and estimates at all depths (7.8 ± 0.3% to 5.1 ± 0.3% in m, 68 ± 2% to 9.1 ± 0.3% in α, and 40 ± 1% to 6.8 ± 0.3% in k in simulated acquisitions; 21.5 ± 0.4% to 15.4 ± 0.4% in m, 291 ± 15% to 76 ± 12% in α, and 39 ± 1% to 21 ± 1% in k in phantom acquisitions). Images from in vivo analysis before and after compensation resulted in an increase in median contrast-to-noise ratio between the cervix and background and the Achilles tendon and background. CONCLUSION:Compensation for changes in ultrasonic resolution with depth prevents saturation in speckle statistics estimates, thus providing more relevant information about tissue properties. The methods described in this work reduce the system dependence of speckle statistics analysis, thus addressing a barrier to the clinical application and adoption of speckle statistics.

Speckle statistics estimation is a useful quantitative ultrasound tool for characterizing tissue microstructure. However, because of their elongated geometry, fibrillar tissue components like collagen may not be described well by speckle statistics models. The purpose of this study is to perform a systematic analysis of the effects of microstructural anisotropy on speckle statistics estimation. We created phantoms made of wool fibers to correlate speckle statistics estimates to elongated scatterer geometries. Phantoms were attached to a calibrated spring to induce fiber alignment by applying a known tension. Ultrasonic beams were steered to 0, ±5, and ±10 degrees. Nakagami and homodyned K distribution parameters were calculated from each steered acquisition. Applying tension (0 to 3±0.2 N) induced alignment in the wool fibers such that speckle statistics estimates exhibited increased dependence on beam steering angle. Whereas an isotropically scattering phantom exhibited 4%, 55%, 25% and 17% total changes in Nakagami m, Nakagami Ω, homodyned K α, and homodyned K k metrics (respectively) over all steering angles with reference to the 0 degree estimate, changes of 35%, 177%, 151%, and 23% were observed in wool fiber phantoms. The same experiment was repeated in the Achilles tendon of a human subject (28%, 190%, 140%, and 53%) and in the cervix of a Rhesus macaque (9%, 76%, 58%, and 11%) to demonstrate sensitivity in vivo. This study demonstrates how speckle statistics parameters can be used to measure the degree of alignment of anisotropic acoustic scatterers separately from spatial density.

The coordinated biomechanical performance of maternal tissues facilitates healthy pregnancy. Quantifying uterine and cervical biomechanical function has been challenging due to minimal data on the anatomy’s shape, size, and material properties across gestation. Addressing this challenge, this study quantifies structural features of human pregnancy by assessing maternal reproductive tissues and estimated fetal weight in 47 low-risk pregnancies at four gestation times. Uterocervical size and estimated fetal weight were measured via ultrasound, and cervical stiffness was measured via mechanical aspiration. Patient-specific uterocervical solid models were built for each time point, and uterocervical dimensions and cervical stiffness rate changes were assessed between time points. We found that uterine growth rates are time- and direction-dependent, with cervical softening occurring fastest in early gestation and cervical shortening fastest in late gestation. In conclusion, this work enables computational modeling platforms (i.e., digital twins) to explore the structural performance of the uterus and cervix in pregnancy.

Cervical remodeling is critical for a healthy pregnancy. Premature tissue changes can lead to preterm birth (PTB), and the absence of remodeling can lead to post-term birth, causing significant morbidity. Comprehensive characterization of cervical material properties is necessary to uncover the mechanisms behind abnormal cervical softening. Quantifying cervical material properties during gestation is challenging in humans. Thus, a nonhuman primate (NHP) model is employed for this study. In this study, cervical tissue samples were collected from Rhesus macaques before pregnancy and at three gestational time points. Indentation and tension mechanical tests were conducted, coupled with digital image correlation (DIC), constitutive material modeling, and inverse finite element analysis (IFEA) to characterize the equilibrium material response of the macaque cervix during pregnancy. Results show, as gestation progresses: (1) the cervical fiber network becomes more extensible (nonpregnant versus pregnant locking stretch: 2.03 +/- 1.09 versus 2.99 +/- 1.39) and less stiff (nonpregnant versus pregnant initial stiffness: 272 +/- 252 kPa versus 43 +/- 43 kPa); (2) the ground substance compressibility does not change much (nonpregnant versus pregnant bulk modulus: 1.37 +/- 0.82 kPa versus 2.81 +/- 2.81 kPa); (3) fiber network dispersion increases, moving from aligned to randomly oriented (nonpregnant versus pregnant concentration coefficient: 1.03 +/- 0.46 versus 0.50 +/- 0.20); and (4) the largest change in fiber stiffness and dispersion happen during the second trimester. These results, for the first time, reveal the remodeling process of a nonhuman primate cervix and its distinct regimes throughout the entire pregnancy.

Objective The objective of this study was to survey national utilization of cervical length (CL) ultrasound on labor and delivery (L&D) for the evaluation of preterm labor (PTL) and identify provider attitudes and barriers to utilization. Study Design Survey was emailed to Obstetrics and Gynecology Residency and Maternal-Fetal Medicine Fellowship program and advertised via links on obstetric-related Facebook interest groups. The survey was open from August 4, 2020 to January 4, 2021. Characteristics between respondents who did and did not report the use of CL ultrasound for PTL evaluation were compared with chi-square analysis. Results There were 214 respondents across 42 states. One hundred and thirty-four respondents (63%) reported any use of CL in the evaluation of PTL and eighty (37%) denied it. There was a significant difference in practice location, practice type, delivery volume, and region between those who did and did not utilize CL ultrasound on L&D. Those who did use CL ultrasound were more likely to report no barriers to use (40 vs. 4%, p < 0.001). The most common barriers involved the availability of transvaginal ultrasound (31%), sterilization of transvaginal ultrasound probe (32%), limited availability of persons able to perform/interpret CL imaging (38%). Nineteen percent believed CL ultrasound had little/no utility in clinical practice. Those who did not use CL ultrasound in the evaluation of PTL were significantly more likely to report the feeling that there was little/no utility of CL ultrasound in clinical practice (37 vs. 7%, p < 0.001) and to report transvaginal ultrasound availability as barriers to use (63 vs. 12%, p < 0.001). Conclusion CL ultrasound is used nationally in PTL evaluation. However, significant barriers limit widespread adoption. These barriers can be addressed through the dissemination of information and practice guidelines, addition of CL ultrasound education in residency training and through CME opportunities after training, and providing support/resources/access for those looking to add this tool to their practice environment. Key Points

The coordinated biomechanical performance, such as uterine stretch and cervical barrier function, within maternal reproductive tissues facilitates healthy human pregnancy and birth. Quantifying normal biomechanical function and detecting potentially detrimental biomechanical dysfunction (e.g., cervical insufficiency, uterine overdistention, premature rupture of membranes) is difficult, largely due to minimal data on the shape and size of maternal anatomy and material properties of tissue across gestation. This study quantitates key structural features of human pregnancy to fill this knowledge gap and facilitate three-dimensional modeling for biomechanical pregnancy simulations to deeply explore pregnancy and childbirth. These measurements include the longitudinal assessment of uterine and cervical dimensions, fetal weight, and cervical stiffness in 47 low-risk pregnancies at four time points during gestation (late first, middle second, late second, and middle third trimesters). The uterine and cervical size were measured via 2-dimensional ultrasound, and cervical stiffness was measured via cervical aspiration. Trends in uterine and cervical measurements were assessed as time-course slopes across pregnancy and between gestational time points, accounting for specific participants. Patient-specific computational solid models of the uterus and cervix, generated from the ultrasonic measurements, were used to estimate deformed uterocervical volume. Results show that for this low-risk cohort, the uterus grows fastest in the inferior-superior direction from the late first to middle second trimester and fastest in the anterior-posterior and left-right direction between the middle and late second trimester. Contemporaneously, the cervix softens and shortens. It softens fastest from the late first to the middle second trimester and shortens fastest between the late second and middle third trimester. Alongside the fetal weight estimated from ultrasonic measurements, this work presents holistic maternal and fetal patient-specific biomechanical measurements across gestation.

Fetal membranes have important mechanical and antimicrobial roles in maintaining pregnancy. However, the small thickness (<800 µm) of fetal membranes places them outside the resolution limits of most ultrasound and magnetic resonance systems. Optical imaging methods like optical coherence tomography (OCT) have the potential to fill this resolution gap. Here, OCT and machine learning methods were developed to characterize the ex vivo properties of human fetal membranes under dynamic loading. A saline inflation test was incorporated into an OCT system, and tests were performed on n = 33 and n = 32 human samples obtained from labored and C-section donors, respectively. Fetal membranes were collected in near-cervical and near-placental locations. Histology, endogenous two photon fluorescence microscopy, and second harmonic generation microscopy were used to identify sources of contrast in OCT images of fetal membranes. A convolutional neural network was trained to automatically segment fetal membrane sub-layers with high accuracy (Dice coefficients >0.8). Intact amniochorion bilayer and separated amnion and chorion were individually loaded, and the amnion layer was identified as the load-bearing layer within intact fetal membranes for both labored and C-section samples, consistent with prior work. Additionally, the rupture pressure and thickness of the amniochorion bilayer from the near-placental region were greater than those of the near-cervical region for labored samples. This location-dependent change in fetal membrane thickness was not attributable to the load-bearing amnion layer. Finally, the initial phase of the loading curve indicates that amniochorion bilayer from the near-cervical region is strain-hardened compared to the near-placental region in labored samples. Overall, these studies fill a gap in our understanding of the structural and mechanical properties of human fetal membranes at high resolution under dynamic loading events.

To characterize vasodynamic response in placentae associated with healthy pregnancies and those complicated by placental-mediated disease with elastography. Sonographic elastography is an emerging technology being applied to clinical placental assessment. Administration of vasoactive agents to perfused placentae is a novel strategy to define the role of vascular dynamics in generation of sonographic parameters. Prospective, descriptive cohort study of patients enrolled for post-delivery placental collection with uncomplicated pregnancies and those with ischemic placental disease (IPD; preeclampsia, fetal growth restriction), gestational hypertension (gHTN) or chronic hypertension (cHTN). Following single lobule placental perfusion, shear wave speeds (SWS) obtained via ultrasound elasticity under physiologic conditions and with sequential introduction of vasodilator and vasoconstrictor, then analyzed descriptively and normalized to individual subject level reference conditions. 13 term patients enrolled with 12 placental regions analyzed per perfusion state in triplicate totaling 689 SWS measurements. Mean SWS computed in each perfusion condition across tissue surfaces (Figure 1). Subject level data normalized to physiologic perfusion conditions and the SWS difference between administration of vasoconstrictor and vasodilator assessed (Figure 2). In healthy placentae, there was a small but consistent increase in SWS from vasodilation to vasoconstriction as the tissue became stiffer. In contrast, only the maternal surface in IPD demonstrated greater stiffness and cHTN, gHTN placentae demonstrated a paradoxical response throughout with increased tissue compliance. Direct vasoactive manipulation of the perfused placenta demonstrates subtle stiffening with vasoconstriction in healthy placentae and a variable response across disease states. This may represent a downstream consequence of disease-related vascular injury or a potentially pathogenic upstream atypical vascular response; the later would represent an opportunity for studies predicting placental-mediated disease.View Large Image Figure ViewerDownload Hi-res image Download (PPT)

Prenatal ultrasound is an indispensable tool used by obstetrical care providers to assist in the everyday care of their pregnant patients. Alongside advancements in imaging, the electronic systems that support this technology have become more advanced. However, it is currently difficult for these individual systems to communicate with each other "out of the box." There is also minimal standardization of the type and format of data transmitted within these systems. Clinicians and system vendors must work collaboratively to create clinical and technical standards to serve as the foundation for increased interoperability among the various systems within each institutional network. Therefore, the Society for Maternal-Fetal Medicine Clinical Informatics Committee established an Ultrasound Electronic Health Record Subcommittee to facilitate collaboration between clinicians, including maternal-fetal medicine subspecialists, and ultrasound network component vendors. Based on the work of this subcommittee, the purpose of this document is to provide: (1) a basic understanding of ultrasound network architecture and capabilities, and (2) best-practice recommendations for electronic health record order design, obstetrical clinical data standards, and billing and coding practices.

In ultrasound shear wave elasticity (SWE) imaging, a number of algorithms exist for estimating the shear wave speed (SWS) from spatiotemporal displacement data. However, no method provides a well-calibrated and practical uncertainty metric, hindering SWE's clinical adoption and utility in downstream decision-making. Here, we designed a deep learning SWS estimator that simultaneously outputs a quantitative and well-calibrated uncertainty value for each estimate. Our deep neural network (DNN) takes as input a single 2D spatiotemporal plane of tracked displacement data and outputs the two parameters m and σ of a log-normal probability distribution. For training and testing, we used in vivo 2D-SWE data of the cervix collected from 30 pregnant subjects, totaling 551 acquisitions and >2 million space-time plots. Points were grouped by uncertainty into bins to assess uncertainty calibration: the predicted uncertainty closely matched the root-mean-square estimation error, with an average absolute percent deviation of 3.84 created a leave-one-out ensemble model that estimated uncertainty with better calibration (1.45 data. Lastly, we applied the DNN to an external dataset to evaluate its generalizability. We have made the trained model, SweiNet, openly available to provide the research community with a fast SWS estimator that also outputs a well-calibrated estimate of the predictive uncertainty.

severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes COVID-19 with a spectrum of disease outcomes in pregnancy and rare congenital infections. Viral tropism for cells in placental tissues is suggested to be low, but SARS-CoV-2 prevalence in placental tissues from women with COVID-19 is unknown. The placenta can produce a robust antiviral innate immune response, though it is unknown whether SARS-CoV-2 induces placental innate immunity and pathologic changes. Though SARS-CoV-2 detection in placental tissues is thought to be infrequent, immune factors associated with viral persistence are unknown. Placental tissues were collected from pregnant women, who were either healthy (N=20; 10 labored, 10 unlabored) or had active (N=53) or resolved COVID-19 disease (N=90) at delivery. Resolved COVID-19 disease was defined as delivery >10 days from symptom onset or diagnosis. We extracted RNA from chorionic villi (CV) and chorioamniotic membranes (CAM) for quantitative PCR for SARS-CoV-2 and a panel of interferon (IFN; ifnb), IFN stimulated genes (ISG; mxa, ifit1) and cytokines (il6). Placental pathology was evaluated in samples where SARS-CoV-2 RNA was detected. Statistical analysis included Student’s t-test and Fisher’s exact test. SARS-CoV-2 viral RNA (vRNA) was detected in placental tissues from 3.5% (5/143) of women with a history of COVID-19 (2 active, 3 resolved; 3 CAM, 3 CV) and 0% (0/20) in healthy controls. Median viral load was 1.0 x 107 copies/mg (range: 4 x 105 – 1 x 1011 copies/mg). An extremely high viral load (1.2 x 1011 copies/mg) was detected in CV 6 days after COVID-19 diagnosis in a patient with a placental basal infarct. The longest interval between diagnosis and SARS-CoV-2 vRNA detection in the placenta (CAM) was 17 weeks after COVID-19 diagnosis. In the CAM, innate immune gene expression was significantly higher in patients with either active or resolved COVID-19 versus controls (active: ifit1, mxa, il6; resolved: ifnb, ifit1, mxa, il6; all p<0.05). In CV, a more limited panel of innate immune genes was significantly elevated compared to controls (active: ifnb; resolved: ifnb and il6; all p<0.05). Notably, the case with detectable SARS-CoV-2 vRNA 17 weeks after COVID-19 diagnosis had very low il6 gene expression in the CV compared to controls. Our data suggest SARS-CoV-2 vRNA can rarely persist in the placenta up to 4 months after diagnosis. High viral loads may correlate with placental pathology. The persistence of the innate immune response, most prominent in CAM, suggests the placental antiviral response remains active for weeks to months after COVID-19. Further study is necessary to determine the impact of the placental innate immune response on viral clearance and whether a sustained elevation in innate immunity may impact fetal health.