OP02.08: Placental volume measurements during the first trimester using artificial intelligence and virtual reality: the Rotterdam periconception cohort

Volumetric measurements of the placenta during the first trimester of pregnancy provide unique insights into development and placenta-related complications. Currently, placental volume (PV) is measured manually using Virtual Organ Computer-aided AnaLysis (VOCAL) software, which is time-consuming, prone to inter- and intrarater variability and prohibits the storage of measurements. To save time, reduce variability and enable storage of measurements, we explored two novel ways to measure the PV: virtual reality (VR) and artificial intelligence (AI). We used three-dimensional (3D) ultrasound datasets from the Rotterdam Periconception Cohort, collected between 7 and 11 weeks gestational age. VOCAL measurements were performed by tracing the gestational sac and placental outline in two-dimensional (2D) planes with a 15° rotational step. VR measurements were performed using a in-house developed VR system. Voxels attributable to the placenta were manually selected based echogenicity. We trained the AI algorithm on 90, and tested it on 24 datasets measured in VR. Table 1 shows the mean and standard deviation of the PV measured with VOCAL, and absolute volume error (cm3) between the VOCAL, VR and AI measurements. Comparing VR and AI to VOCAL, we find a comparable volume error. AI is closer to the measurements in VR, which it was trained on. Both AI and VR measure the PV based on 3D information, where VOCAL uses 2D information. AI measurements are automated and therefore favourable over VR, since no inter- and intrarater variability is introduced and it is time-saving. In summary, both novel ways to measure PV can monitor placental development. Ultimately, PV measurement will contribute to earlier and better screening, diagnostics and treatment of placenta-related pregnancy complications.