Deformable Medial Modeling To Generate Novel Shape Features Of The Placenta Using Automated Versus Manual Segmentations
Tuesday, April 9, 2024
4:42pm – 4:49pm
Location: 410
Authors: Gabriel Arenas, University of Pennsylvania Perelman School of Medicine Alison Pouch, University of Pennsylvania Ipek Oguz, Vanderbilt University Baris Oguz, University of Pennsylvania Brett Byram, Vanderbilt University Xing Yao, Vanderbilt University Nadav Schwartz, University of Pennsylvania, Perelman School of Medicine
Deformable medial modeling is a novel, standardized approach that evaluates for differences beyond volume and Dice overlap score between manual and automated segmentations of the placenta. This method produces a detailed analysis in identifying specific automated segmentation errors that relies on placental morphometry. Thus, deformable medial modeling may improve validations efforts between manual and automated segmentations of the placenta. More importantly, this tool affords clinicians and investigators the opportunity to standardize and characterize in vivo placental morphology as it relates to adverse maternal and perinatal outcomes. Deformable medial modeling can serve as a platform through which placental shape and function are better understood.