Automated Analysis of the Preterm Neonatal Cortex at Term Equivalent Age and Correlation with Cognitive Outcome at 1 Year Corrected Age

Abstract

Background: Preterm birth (< 32weeks) is associated with differences in brain anatomical structure and adverse neurodevelopmental outcome. White matter volume and IQ have been shown to be reduced in ex-preterm infants in adolescence. Subtle differences in cortical folding at term equivalent age may be useful in predicting long-term neurodevelopmental outcome.Methods: We segment grey matter (GM) and white matter (WM) using a maximum a posteriori expectation maximisation routine with adaptive priors. The GM/WM surface is extracted using a level set, and at every point we define the local shape-index, describing how cup-like or saddle-like the surface is (fig1A), and the curvedness, describing the distortion of the surface relative to that of a flat sheet. The distributions of each index may be summarised in a histogram (fig1B) and we use the histogram entropy as a potential biomarker.A: GM/WM interface coloured by shape-index B: Population average local surface histograms of curvedness and shape index overlaid on single subject GM parcellation C. Correlation between shape index & combined cognitive/langauge outcome.Results: 39 infants without obvious structural abnormalities on MRI had brain volumes and cortical folding analysed from T1-weighted images. WM volume correlates significantly with gestation at birth (p< 0.05). Shape index and curvedness did not correlate with gestation at birth. Shape index correlates with combined cognitive / language outcome at 1 year using the Bayley III scales of infant development (fig1C).Conclusion: Cortical folding analysis may be useful, possibly in combination with white matter volumes in the development of biomarkers of neurodevelopmental outcome.