Automated atlas-based craniofacial biometry for 3D fetal MRI: multi-acquisition comparison of fetuses with Down syndrome and a control cohort.

We introduce the first automated atlas-based method for fetal craniofacial biometry. Using motion-corrected slice-to-volume reconstructions for 3D fetal head visualisation, an automated label propagation method extracted linear biometry across 12 measures. The optimisation process used retrospective data and no differences in automated biometry was seen between different MRI acquisition parameters. A comparison of measures made between a cohort of fetuses with Down syndrome and control fetuses, with normal development, found significant differences for the occipitofrontal skull, oral hard palate and anterior base of skull distances. This suggests a promising and meaningful method for large population-level investigation of MRI craniofacial morphology.