Is quantitative analysis of fetal facial morphology feasible? Preliminary results from a pilot study.

Objectives Hypertelorism, micrognathia, broad nasal bridge and thin, elongated mouth may raise a suspicion of genetic condition but infrequently lead to a definitive antenatal diagnosis. We aimed to assess the feasibility of a novel post-processing imaging technique for quantitative assessment of fetal face from 3D ultrasound volumes. Methods 3D ultrasound fetal face volumes, collected for clinical indications at 32 (29–34) weeks were analysed. 3 normal and 3 abnormal cases (1 Noonan, 1 Trisomy 18 and 1 suspected genetic condition) were acquired in the midline sagittal view using a 4–8MHz probe (Voluson E8 system, GE Healthcare). Volumes were manually reconstructed (Mimics, Belgium) and cut according to standardised planes defined by the lenses and alveolar ridge. The 6 3D surface meshes generated were manually aligned and served as input for a Statistical Shape Model (SSM, www.deformetrica.org), which computed mean face shape and 3D shape variations using Principal Component Analysis (PCA). Results PCA identified four modes contributing to over 90% of face shape variability. The first mode accounted for size differences and the second highlighted shape feature changes from a proportionate face shape towards an asymmetry with wide prominent forehead, small and recessed chin. The PCA parameters for the second mode numerically distinguished between the normal and abnormal fetal case in all cases (see figure). Conclusions This study demonstrates the feasibility of objective characterisation and quantification of facial dysmorphism from 3D ultrasound raising the potential for fetal diagnosis when non-invasive and invasive tests do not provide informative results.