Super-resolution MRI-derived brainstem and cerebellar volumes in fetuses between 22 weeks and 32 weeks of gestation.

BACKGROUND

The posterior fossa undergoes rapid development in utero, yet normative volumetric data are limited, especially during mid-gestation when most fetal MRIs are performed.

OBJECTIVE

This study aimed to develop a segmentation method tailored to the fetal posterior fossa and establish normative growth trajectories for key infratentorial structures between 22 weeks and 32 weeks of gestation.

MATERIALS AND

METHODS

Eighty-five pregnant women were prospectively recruited at a single institution. Inclusion criteria were normal singleton pregnancies between 19-40 weeks' gestation and maternal age 18-45 years. Exclusion criteria included fetal abnormalities, MRI contraindications, or significant maternal comorbidities. Eighteen fetuses were excluded, resulting in 67 normal fetuses (43 male, 24 female). Imaging was performed at 3 T using multiplanar T2-weighted sequences, reconstructed into isotropic volumes after motion correction and brain extraction. Segmentation labels were created using age-specific atlases from the Developing Human Connectome Project. Expert annotations defined the midbrain, pons, medulla, vermis, and cerebellar hemispheres. Images were registered to the atlases using the Symmetric Normalization (SyN) algorithm, with labels manually verified by a pediatric neuroradiologist. Volumes were normalized to total brain volume and analyzed using linear regression with gestational age and sex as predictors.

RESULTS

Absolute volumes of all posterior fossa structures significantly increased with gestational age (P<0.001), showing strong linear associations (R2=0.80-0.87). The cerebellum exhibited the steepest growth (β=1,056.01, R2=0.87, P<0.001). Growth was symmetric with no significant left-right differences. Relative volumetric trends varied: the cerebellum increased proportionally (β=0.007, P<0.001), while the midbrain, pons, and medulla decreased relative to total brain volume. The vermis showed no significant association with gestational age but had sex-specific effects; males had smaller relative vermian volumes than females (β=-0.43, P=0.04), despite larger absolute posterior fossa volumes overall.

CONCLUSIONS

This study provides normative volumetric references for fetal posterior fossa structures between 22-32 weeks' gestation using a novel segmentation method. Absolute growth followed linear patterns, while relative measures revealed sex-specific variations in the vermis and cerebellum. These benchmarks may enhance diagnostic accuracy for detecting posterior fossa anomalies in clinical fetal MRI.