OP10.05: 3D craniofacial fetal MRI: formalisation of a landmarking and measurement protocol in Trisomy 21 and healthy control subjects

To assess initial experiences of novel MRI biometrics using slice to volume registration fetal MRI. A systematic literature review was performed to define anatomical landmarks appropriate for MRI in 3D space. The final protocol was agreed by consensus with 3 fetal MRI radiologists. Landmarks were propagated to subjects from a template atlas (31weeks GA) using classical affine+non-rigid registration. The centroid data of points was used to calculate biometrics, using 3D vector formulae. Biometric comparisons were made between cohorts and 10 random cases measured manually. 108 control (meanGA = 31+6) and 24 T21 subjects (meanGA = 32+4) were assessed (MRI = 1.5 or 3T). An analysis of variance between subject groups found differences in biometry for skull base, skull/midface length and nasopharyngeal areas (total 17/35 biometrics) correlating with known dysmorphology in T21 phenotypes (MANOVA, p < 0.05). The time to manually label 10 random cases was an average of 22mins/case versus 5mins/case to inspect automated labels. The manual versus automated biometry had a mean relative error of more than 10% in 11/35 biometrics and manual measurements were not possible for 22/350 due to variable image quality. Automated methodologies can assist in rapid quantitative craniofacial assessment for deep phenotyping. Since measurement error reduces precision, future work will focus on optimising MRI sequences, deep learning methods and validation in syndromic cohorts. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.