Search

This is a video of our lightning talk created for the 2020 Summer Biomechanics, Bioengineering, and Biotransport Conference (SB3C), held online June 17-20, 2020: https://archive.sb3c.org/sb3c2020/ Pregnancy, though ubiquitous, is not well understood, in part due to the difficulties surrounding studying the gravid environment. In order to better understand normal gestation processes, an automated patient-specific parametric modeling framework of the pregnant uterus and cervix based on 2D ultrasound was developed. 29 healthy pregnant patients were recruited to have 2D ultrasound images collected at 4 gestational ages in the standing and supine positions. From these images, three trained sonographers took a set of dimension measurements of the uterus and cervix to characterize their overall size and orientation. A solid model was constructed in Solidworks, and defined in a design table such that the 2D ultrasound dimension measurements could be imported and patient-specific models automatically generated. Across all cases, 86% of models automatically generated correctly. Uterine and cervical volume were determined across gestational age using the novel volume estimating abilities of the model. The uterine volume was shown to increase with gestational age, and the cervical volume showed no clear trend with gestational age. The development of these models allows for rapid creation of patient-specific solid models of pregnant anatomy, providing insight to the evolution of the uterus and cervix in normal gestation.

This work is a summation of Divya Rajasekharan and Arielle Feder’s 2020 summer project of parametrically modeling the pelvic bone and vaginal canal in the Myers Soft Tissue Lab at Columbia University. In order to simulate the mechanics of birth, accurate models of reproductive anatomy must be acquired, including the pelvic bone and vaginal canal. The vaginal canal modeling method is based off of measurements from existing literature, capturing the vaginal axis and transverse width. Though currently employing patient averages, we have built the model such that patient-specific measurements of the vaginal canal can be used in the future. The pelvic bone modeling method introduces simplifications that capture the structures relevant to fetal head decent, while also capturing patient-specific pelvic inlet shape. Thus, the pelvic bone is reduced to a ring that can capture most pelvic inlet shapes and solid body to represent the sacrum. In the future, these structures will be implemented into a finite element analysis simulation of birth and their effect on the mechanics of delivery explored.

This work is a summation of Divya Rajasekharan and Arielle Feder’s 2020 summer project of capturing the coronal shape of the uterus in late pregnancy for patient-specific parametric models in the Myers Soft Tissue Lab at Columbia University. In order to simulate the mechanics of birth, accurate models of reproductive anatomy must be acquired. With MRI images costly to obtain and not routinely taken during pregnancy, wide implementation of a patient-specific modeling framework of pregnancy must be based on 2D ultrasound images. Previously, our lab has used a parametric modeling framework based on 2D ultrasound images that assumes an ellipsoidal coronal uterine shape. However, this shape does not accurately capture anatomy in late gestation. Therefore, we investigate utilization of two additional modeling methods based on observations from solid models of the uterus and cervix from segmented MRI images of late gestation pregnancy. The first modeling method is based off the same measurements taken from 2D ultrasound, with additional assumptions about coronal shape derived from MRI images. The second modeling method uses numerous uterine diameter and wall thickness measurements from MRI. All models are compared using finite element analysis with simplistic loading conditions for 5 patients. It was found that uterine wall thickness throughout the uterine body has a significant effect on the magnitude and pattern of stress and stretch. It is also noted that direction of principle strains is consistent across all models. For future simulations of birth, we will develop a solid modeling method that incorporates aspects of both tested models.