LAUSR

Background: Craniosynostosis is typically diagnosed postnatally. Prenatal diagnosis would allow for improved parental counseling and facilitate timely intervention. Our purpose was to determine whether prenatal ultrasound can be used to diagnose nonsyndromic craniosynostosis. Methods: The authors reviewed 22 prenatal ultrasounds of infants known to have nonsyndromic craniosynostosis and 22 age-matched controls. Cross-sectional images at the plane used to measure biparietal diameter were selected and cranial shape of each participant was parameterized and used to discriminate affected patients from controls. The results from quantitative shape analysis were compared with results from a blinded visual inspection alone. Results: Among the 22 patients, the most common diagnosis was sagittal synostosis (n = 11), followed by metopic synostosis (n = 6). The average gestational age at time of ultrasound of controls and synostotic patients was 26 weeks and 6.8 days at the junction of the second and third trimesters. The controls and synostotic cases segregated into statistically different populations by their shape profiles (p < 0.001). An automatic shape classifier using leave-one-out cross-validation correctly classified the 44 images as normal versus synostotic 85 percent of the time (sensitivity, 82 percent; specificity, 87 percent). Cephalic index was a poor indicator of sagittal synostosis (45 percent sensitivity). Visual inspection alone demonstrated only a fair level of accuracy (40 to 50 percent agreement) in identifying cases of synostosis (kappa, 0.09 to 0.23). Conclusions: Craniosynostosis can be identified on prenatal ultrasound with good sensitivity using formal shape analysis. Cephalic index and visual inspection alone performed poorly. CLINICAL QUESTION/LEVEL OF EVIDENCE: Diagnostic, II.