Diagnostic accuracy of diffusion tensor imaging for pediatric cervical spinal cord injury

Abstract

Cross-sectional non-experimental study. To examine diagnostic accuracy of diffusion tensor imaging (DTI) for pediatric spinal cord injury (SCI). Pediatric Orthopedic Hospital. Thirty-five subjects, 10 SCI and 25 controls, mean age 13.38 years underwent two scans with 3.0 T MR scanner. Fractional anisotropy (FA), axial diffusivity (AD) and radial diffusivity (RD) values were calculated. Subjects with SCI underwent examination of muscle strength, sensation and sacral sparing. Mean and s.d. values for FA, AD and RD were compared by group (controls, SCI with sacral sparing, SCI without sacral sparing) using analysis of variance for repeated measures. Comparisons were also made of DTI values at the injury site to values from cervical regions outside of the injury site. Specificity, sensitivity, receiver operating characteristics area under the curve (ROC AUC) and corresponding 95% confidence intervals were calculated. Resampling methods were used to validate the estimates from the final models. FA values differed among SCI subjects with intact sacral sparing, absent sacral sparing and controls, P<0.003 (adjusted). DTI values in combination showed the strongest diagnostic accuracy for predicting the presence of anal contraction (AD, RD; ROC AUC=0.90), deep anal pressure (FA; ROC AUC=0.88), S4-5 sensation (FA, RD; ROC AUC=0.93), motor level (FA, AD, RD; ROC AUC=0.92) and MRI level (FA, AD, RD; ROC AUC=0.92). Bootstrap and Jackknife median values indicated consistency of the parameter estimates. The predictive accuracy of DTI for sacral sparing end points and motor and MRI level of injury was good to strong.