Periventricular　white　matter　injury　(PWMI)　is　a　common　cause　of　spastic　cerebral　palsy　(SCP).　Diffusion　tensor　imaging　(DTI)　shows　high　sensitivity　but　moderate　specificity　for　predicting　SCP.　The　limited　specificity　may　be　due　to　the　diverse　and　extensive　brain　injuries　seen　in　infants　with　PWMI.　We　enrolled　72　infants　with　corrected　age　from　6　to　18　months　in　3　groups:　PWMI　with　SCP　(n　=　20),　non-CP　PWMI　(n　=　19),　and　control　(n　=　33)　groups.　We　compared　DTI-based　brain　network　properties　among　the　three　groups　and　evaluated　the　diagnostic　performance　of　brain　network　properties　for　SCP　in　PWMI　infants.　Our　results　show　abnormal　global　parameters　(reduced　global　and　local　efficiency,　and　increased　shortest　path　length),　and　local　parameters　(reduced　node　efficiency)　in　the　PWMI　with　SCP　group.　On　logistic　regression,　the　combined　node　efficiency　of　the　bilateral　precentral　gyrus　and　right　middle　frontal　gyrus　had　a　high　sensitivity　(90%)　and　specificity　(95%)　for　differentiating　PWMI　with　SCP　from　non-CP　PWMI,　and　significantly　correlated　with　the　Gross　Motor　Function　Classification　System　scores.　This　study　confirms　that　DTI-based　brain　network　has　great　diagnostic　performance　for　SCP　in　PWMI　infants,　and　the　combined　node　efficiency　improves　the　diagnostic　accuracy.