Correlation between degree of subvoxel spinal cord compression measured with super-resolution tract density imaging and neurological impairment in cervical spondylotic myelopathy

Abstract

OBJECT: The purpose of this study was to explore the use of super-resolution tract density images derived from probabilistic diffusion tensor imaging (DTI) tractography of the spinal cord as an imaging surrogate for microstructural integrity and functional impairment in patients with cervical spondylosis. METHODS: Structural MRI and DTI images were collected for 27 patients with cervical spondylosis with (n = 21) and without (n = 6) functional impairment as defined by the modified Japanese Orthopaedic Association Scale (mJOA). DTI was performed axially through the site of compression in a total of 20 directions with 10 averages. Probabilistic tractography was performed at 0.5-mm isotropic spatial resolution using the streamline technique combined with constrained spherical deconvolution. The following measurements were calculated for each patient: maximum tract density at the site of compression, average tract density in rostral normal-appearing spinal cord, and the ratio of maximum density to normal density. RESULTS: Compared with normal tissue, the site of compression exhibited elevated fiber tract density in all patients, and a higher fiber tract density was also noted in focal areas at the site of compression in patients with functional impairment. There was a strong negative correlation between maximum tract density and mJOA score (R2 = 0.6324, p < 0.0001) and the ratio of maximum tract density to normal tract density (R2 = 0.6647, p < 0.0001). When grouped according to severity of neurological impairment (asymptomatic, mJOA score of 18; mild, mJOA score of 15-17; moderate, mJOA score of 11-14; and severe, mJOA score < 11), the results showed a significant difference in the ratio between severe and both no impairment (p = 0.0009) and any impairment (p = 0.036). A ratio of maximum fiber tract density at the site of compression to fiber tract density at C-2 greater than 1.45 had 82% sensitivity and 70% specificity for identifying patients with moderate to severe impairment (ROC AUC = 0.8882, p = 0.0009). CONCLUSIONS: These results support the use of DTI as a surrogate for determining spinal cord integrity in patients with cervical spondylosis. Probabilistic tractography provides spinal cord microstructural information that can help discern clinical status in cervical spondylosis patients with varying degrees of neurological impairment.