Automatically Quantified Diffuse Excessive High Signal Intensity on MRI Predicts Cognitive Development in Preterm Infants

BACKGROUND: Cognitive and language impairments constitute the majority of disabilities observed in preterm infants. It remains unclear if diffuse excessive high signal intensity on magnetic resonance imaging at term represents delayed white matter maturation or pathology. METHODS: We hypothesized that diffusion tensor imaging based objectively quantified diffuse excessive high signal intensity measures at term will be strong predictors of cognitive and language development at 2 years in a cohort of 41 extremely low birth weight (<= 1000 g) infants. Using an automated probabilistic atlas, mean diffusivity maps were used to objectively segment and quantify diffuse excessive high signal intensity volume and mean, axial, and radial diffusivity measures. Standardized neurodevelopment was assessed at 2 years of age using the Bayley Scales of Infant Development, third edition. RESULTS: Thirty-six of the 41 infants (88%) had complete developmental data at follow-up. Objectively quantified diffuse excessive high signal intensity volume correlated significantly with cognitive and language scores at 2 years (P < 0.001 for both). The sum values of the three diffusivity measures in detected diffuse excessive high signal intensity regions also correlated significantly with the Bayley scores (r(2) 34.7%; P < 0.001 for each). Infants in the highest quartile for diffuse excessive high signal intensity volumes had scores between 19 and 24 points lower than infants in the lowest quartile (P < 0.01). When diagnosed subjectively by neuroradiologists however, Bayley scores were not significantly lower in infants with extensive diffuse excessive high signal intensity. CONCLUSIONS: These findings lend further evidence that diffuse excessive high signal intensity is pathologic and that objectively quantified diffusion-based diffuse excessive high signal intensity volume at term is associated with cognitive and language impairments. Our approach could be used for risk stratification and early intervention for such high-risk extremely preterm infants.