Ff. Shafaie et al., Comparison of computed tomography myelography and magnetic resonance imaging in the evaluation of cervical spondylotic myelopathy and radiculopathy, SPINE, 24(17), 1999, pp. 1781-1785
Study Design. A cross-sectional retrospective radiologic study.
Objectives. To establish concordance rates between interpretations of compu
ted tomography myelography and magnetic resonance imaging in patients with
degenerative cervical spine disease.
Summary of Background Data. Observed discrepancies in interpretation of com
puted tomography myelography and magnetic resonance imaging question the re
liability of comparisons between these two methods.
Methods. This study blindly and randomly evaluated cervical computed tomogr
aphy myelography and magnetic resonance imaging in 20 patients referred for
clinically diagnosed cervical spondylotic radiculopathy, myelopathy, or bo
th. The discovertebral joints, facet joints, lateral recesses, cord size, s
pinal canal, and neural foramina also were evaluated with graded scales. Al
l results were subjected to the kappa statistic for strength of agreement.
Results. Agreement for interpretation of the discovertebral junction occurr
ed in 144 of 240 sites (60%), indicating only moderately good intermethod c
oncordance (kappa = 0.44). Intermethod agreement on the characterization of
facet joint disease was only moderately good (143 of 160 sites; 89.4%; kap
pa = 0.52), and on characterization of lateral recess disease was poor (125
of 160 sites; 78.1%; kappa = 0.20). On degree of spinal canal compromise,
there was agreement within one grade in 199 of 240 sites (82.9%; kappa = 0.
42). Intermethod agreement on neural foraminal encroachment and cord size w
as only moderately good (kappa = 0.42 and 0.46, respectively). Computed tom
ography myelography tended to upgrade the spinal canal narrowing and neural
foraminal encroachment.
Conclusions. For most parameters of interpretation, the degree of concordan
ce between computed tomography myelography and magnetic resonance imaging i
s only moderately good, with discrepancies noted especially in the differen
tiation of disc and bony pathology. These methods should be viewed as compl
ementary studies.