MRI based radiomics enhances prediction of neurodevelopmental outcome in very preterm neonates

This study evaluated whether MRI-derived radiomic features could predict adverse neurodevelopmental outcomes in 166 very preterm neonates. Clinical variables and radiomic features extracted from T1- and T2-weighted images were assessed against motor, cognitive, and language outcomes at 18 months, 33 months, and 4.5 years. Radiomic features outperformed clinical variables for several adverse motor outcomes, while combining clinical and radiomic information produced the strongest model for motor outcome at 4.5 years. The findings support radiomics as a useful complement to clinical predictors in this population.

Scientific Reports, 12(1), 11872