MRI-based assessment of fetal brain gyrification: Reliability across two methods

To explore the reliability of atlas-based assessment of fetal gyrification on MRI, given the absence of a standardized method to assess cerebral gyrification on fetal imaging and the potential of recently published gestational-age specific fetal brain atlases including high-resolution anatomical detail.

Fetal MRI scans (n = 20; gestational age 26 + 0 to 33 + 1 weeks) were retrospectively randomly selected, including 10 healthy controls and 10 cases with early-onset fetal growth restriction, of which 5 were suspected of delayed fetal gyrification based on comparison with an atlas from 2004. Two experienced pediatric neuroradiologists, independently and blinded to clinical condition, assessed each scan using two approaches: (1) a 2D template from the 3D Gholipour atlas, and including measurements of Sylvian fissure angle and Pistorius scoring, and (2) a full 3D atlas-based evaluation.

Intraclass correlation coefficients (ICCs) between radiologists' estimates of GA and the actual GA were above 0.940 for both methods. Inter-rater ICCs were 0.937 for method 1 and 0.932 for method 2. In contrast, ICCs for the Sylvian fissure angle or the Pistorius score for the left and right side were low (0.393, 0.649, 0.098 and 0.131, respectively). Determining delayed gyrification proved challenging, with Cohen's Kappa values between 0.22 and 0.64 between methods and radiologists.

Gestational age based on gyrification can be estimated with high reliability, both in agreement with actual GA and between raters, using both 2D and 3D MRI templates from the Gholipour fetal brain atlas. These findings support the reliability of standardized fetal brain templates for assessment of fetal gyrification within this single-center cohort. Reproducibility across centers, scanners and clinical settings remains to be established.