Machine-learning to characterise neonatal functional connectivity in the preterm brain

摘要: Brain development is adversely affected by preterm birth. Magnetic resonance image analysis has revealed a complex fusion of structural alterations across all tissue compartments that are apparent term-equivalent age, persistent into adolescence and adulthood, associated with wide-ranging neurodevelopment disorders. Although functional MRI the relatively advanced organisational state neonatal brain, full extent nature disruptions following birth remain unclear. In this study, we apply machine-learning methods to compare whole-brain connectivity in infants at age healthy term-born neonates order test hypothesis results specific age. Functional networks were estimated 105 26 term controls using group-independent component graphical lasso model. A random forest-based feature selection method was used identify discriminative edges within each network nonlinear support vector machine classify subjects based on alone. We achieved 80% cross-validated classification accuracy informed small set edges. These connected number nodes subcortical cortical grey matter, most stronger compared those born preterm. Half one or more basal ganglia. demonstrate brain significantly altered confirming previous reports between structures higher-level association cortex