Application of Diffusion Tensor Imaging in Forecasting Neurological Injury and Recovery after Human Cervical Spinal Cord Injury.

The aim of this study is to determine the strength and accuracy of diffusion tensor imaging (DTI) parameters to predict neurological injury and recovery in adult cervical spinal cord injury (SCI). DTI magnetic resonance imaging (MRI) was performed on 23 acute cervical SCI patients within 12 h after injury and on 45 controls utilizing a rapid DTI sequence (∼5 min). Neurological assessments were conducted from within 24 h of injury up to 6 months utilizing detailed International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI) examinations. Spearman correlation and receiver operating characteristic (ROC) analysis were used to identify relationships between the DTI parameters at the lesion epicenter and adjacent regions, with injury severity and recovery. In proximity to the anatomical injury (within one level above and below epicenter), there was significant reduction in fractional anisotropy (FA), and elevation in mean diffusivity (MD) and radial diffusivity (RD). DTI values measured one level rostral to the injury epicenter showed stronger correlations with multiple clinical features at several time-points. Area under the curve (AUC) obtained from ROC analysis showed FA (AUC = 0.77) measured at lesion epicenter, and FA (0.83), MD (0.76), and RD (0.83) values measured immediately rostral (one level above) to epicenter discriminate injury severity. Further, MD (0.78) measured at lesion epicenter, and MD (0.79) and RD (0.74) values measured immediately rostral to epicenter discriminate neurological recovery. DTI indices measured immediately rostral to the anatomical level of injury consistently showed better correlation (moderate to strong) and accuracy in predicting neurological injury (FA, r = -0.51 and RD, r = 0.54) and recovery (MD, r = -0.51) than indices measured at the epicenter. There was weak to moderate correlation of all measures at lesion epicenter in predicting neurological injury (FA: r = -0.48; MD: r = 0.23; RD: r = 0.34; axial diffusivity [AD]: r = 0.02) and recovery (FA: r = 0.27; MD: r = -0.44; RD: r = -0.35; AD: r = -0.34).