Fetal imaging with microCT (µCT) in the age of AI

Improving chemical and pharmaceutical safety while reducing animal use and development costs is a shared objective of industry, regulators, and the public. We describe a digital imaging pipeline designed to replace the method established by Wilson and Staples using external, visceral, and skeletal imaging of rat fetuses. Gestation day 21 fetuses from timed-mated Crl:CD[SD] rats were euthanized, fixed in formalin, stained with iodine-based contrast agents, embedded in agarose, and scanned using a ZEISS Xradia Versa micro-CT system. Three-dimensional images of the cranium and brain were reconstructed and segmented using Slicer and Dragonfly 3D World software. Key anatomical structures were reviewed by a trained fetal morphologist to confirm accuracy relative to conventionally prepared specimens. Immersion in 5% Lugol's solution for 72-96 h produced uniform soft-tissue contrast. The resulting 3D µCT datasets enabled high-quality external and visceral assessments that compared favorably with traditional approaches. Skeletal evaluations of unstained specimens were of diagnostic quality whereas segmented skeletal features from stained specimens were not. Overall, CT imaging combined with an explainable AI framework shows promise for streamlining developmental toxicology through quantitative, reproducible assessment while reducing labor and subjectivity. Regulatory acceptance, GLP compliance, and a favorable industry value proposition will be essential for adoption.