Childbirth is a high-risk event that obstetrics has been slow to model with modern digital tools. BirthView, a computer simulation that visualises fetal and maternal anatomy during labour, now forms the basis for a proposed “digital twin” called Birth4Cast. The problem is that clinicians cannot reliably predict whether a natural birth will succeed or turn into an emergency Caesarean section, which carries greater risks for both mother and baby. Birth4Cast would use fetal MRI data, machine learning, and finite element analysis to simulate a specific woman’s labour before it happens, showing whether the fetal head can rotate and descend safely. If the simulation predicts failure, a planned Caesarean can be scheduled instead. The research must first solve challenges in image segmentation and estimating soft-tissue properties and uterine forces. If successful, the system could reduce emergency interventions and their long-term consequences. This is primarily an engineering and computing project that adapts structural simulation methods—used for decades in car and aircraft design—to a medical context, with potential to quietly improve maternity care by making a hidden process visible and predictable.
View original technical description
Childbirth is an event that all of us went through yet we may not realise that it carried a higher risk of trauma than any other event we may go through during our lifetime! Obstetrics and midwifery are the clinical specialties that are primarily concerned with childbirth. They have experienced substantial clinical and pharmacological improvements in the last decades but have been slow to adopt state of the art digital technologies such as computer graphics (CG), extended reality (XR), the finite element method (FEM), machine learning (ML) and artificial intelligence (AI). These technologies may shed more light on the non-transparent childbirth process and therefore be well-positioned to address the challenge of problematic childbirths that may have long-term consequences to fetus and mother. BirthView is a computer-based technology that visualises the fetal and maternal anatomies. It uses a variant of the FEM to calculate the interaction of the fetus with the maternal anatomy during the second stage of labour. The FEM has seen widespread use since the 1960s mainly to calculate structural deformations from applied forces and moments. Popular applications are in vehicle and aircraft design, structural engineering and more recently in biomechanics and soft tissue modelling. BirthView’s integration of accurate virtual feto-maternal anatomical models and algorithms to calculate their interaction has resulted in an XR/CG based simulation that mimics the childbirth process realistically. This “realism” is validated by the observation of the “cardinal movements” of childbirth that include the fetal head rotations that facilitate expulsion of the fetus from the womb. The long-term aim of the proposed research is to create a subject-specific simulator using BirthView as its basis. This simulator, that we will call “Birth4Cast”, would be capable of predicting a real childbirth outcome before the actual event. For example, if Birth4Cast predicts that a physiologic (natural) birth will not be possible then the decision could be taken to perform a planned or “elective” Caesarean Section (CS) rather than risking an emergency CS. BirthView’s ability to simulate realistic childbirth scenarios and Birth4Cast’s subject-specific adaptation enables the creation of a “digital twin” of childbirth that can predict potential adverse outcomes before the actual event. To create such a digital twin, the relevant anatomical data of the subject with the in-situ fetus need to be captured. The only medical image modality that is both safe and capable of capturing the relevant anatomical data is (fetal) MRI. The process to generate the simulator’s anatomical models from the fetal MRI data is called “image segmentation”. Once all relevant maternal and fetal body parts have been “segmented” they need to be reformatted and registered as to be rendered suitable for use with the FEM that lies at the basis of the simulator’s reproduction of the cardinal movements. Some of the challenges that come with the creation of the Birth4Cast simulator include the accuracy by which the feto-maternal geometry can be created and the estimation of simulation parameters, e.g. material properties of feto-maternal soft tissues, the force generated by the uterus etc. These simulation parameters will be estimated from the collected subject-specific fetal MRI and perinatal data using ML and AI algorithms. Due to the multidisciplinary nature of this project, it will widely benefit the underpinning engineering and computing disciplines, alongside its long-term goal of preventing potential adverse childbirth outcomes.
Andrew Fry (Co-Investigator)Beatriz De La Iglesia (Co-Investigator)Elspeth Whitby (Co-Investigator)Gavin Cawley (Co-Investigator)Paul Watton (Co-Investigator)Rudy Lapeer (Principal Investigator)Stephen Laycock (Co-Investigator)Wenjia Wang (Co-Investigator)Xinshan Li (Co-Investigator)
Plain English summaries and category classifications on this site are generated by AI and may not perfectly reflect the original research.
Is something wrong? Let us know