Reliability in cerebellar size assessment by 2D cranial ultrasonography in neonates.

Abstract

Cerebellar size appears to be correlated with the long-term outcome of preterm infants. Two-dimensional cranial ultrasonography (2D-cUS) is the first-line, routine tool to characterize newborn cerebral structures. Information regarding the accuracy of 2D-cUS for assessing cerebellar size is scarce. To assess the reliability of cerebellar size using 2D-cUS. Fifty infants were prospectively scanned for a variety of clinical indications. The scans were stored and then independently analyzed offline by two observers. The transverse cerebellar diameter coronally from the anterior and mastoid fontanels (TCD-AF and TCD-MF, respectively); the midsagittal cerebellar vermis height and the anterior-posterior diameter from the AF; and the width and height of the fourth ventricle coronally from the MF view were measured. Interobserver variability was assessed using the intraclass correlation (ICC) with 95% confidence interval (CI) for each of the paired measures. A total 294 images were included in the analysis. The interobserver ICCs were TCD-AF=0.85 (0.76-0.91), TCD-MF=0.9 (0.83-0.94), vermis height=0.79 (0.66-0.88), anterior-posterior vermis diameter=0.71 (0.5-0.84), width of the fourth ventricle=0.72 (0.33-0.9), height of the fourth ventricle=0.86 (0.7-0.92). The ICC for the TCD comparing the AF and the MF approach was ≥0.9 for both observers. The reliability of the offline 2D-cUS cerebellar size assessments is good, and TCD-MF is the most consistent measure. Further studies should address the predictive value for the neurodevelopmental outcome of serial cerebellar size assessments by 2D-cUS for preterm infants.