Accuracy of diffusion tensor imaging for diagnosing cervical spondylotic myelopathy in patients showing spinal cord compression but otherwise unremarkable conventional MRI findings

Abstract

Objective: To assess performance of diffusion tensor imaging (DTI) for diagnosis of cervical spondylotic myelopathy (CSM) in patients who show deformed spinal cord but otherwise unremarkable conventional magnetic resonance imaging (MRI) findings and to assess the correlation between the degree of central canal stenosis (CCS) and DTI parameters.

Materials and Methods: CCS severity was graded from sagittal T2-weighted images in 64 patients with cervical spondylosis. DTI parameters (mean diffusivity [MD], longitudinal diffusivity [LD], radial diffusivity [RD] and fractional anisotropy [FA]) were measured at the most stenotic level. Correlations between the degree of CCS and the DTI parameters were assessed. The performances of MD, FA, MD∩FA (considered positive when both the MD and FA results were positive), LD∩FA, and RD∩FA were evaluated and compared in patients showing compressed spinal cord without a signal change for distinguishing whether the patient clinically has CSM (n = 4) or not (n = 29).

Results: FA and RD values were negatively (r = ？0.545) and positively (r = 0.399) correlated respectively with the degree of CCS (p < 0.001). Sensitivity, specificity, positive predictive value and negative predictive values in patients (n=33) showing compressed spinal cord without a signal change were best in LD∩FA. The calculations (percentages) were: 100, 44.8, 20, and 100 for MD; 100, 27.5, 16, and 100 for FA; 100, 58.6, 25, and 100 for MD∩FA; 100, 68.9, 30.8 and 100 for LD∩FA; and 75, 68.9, 25, and 95.2.for RD∩FA.

Conclusion: FA combined with MD, LD, and RD was a useful means to distinguish patients with and without CSM.