Diagnostic Value of Different MRI Features in Subaxial Cervical Ligament Injury

朱锴,陈其昕,李方财,徐侃,陈维善,吴琼华
DOI: https://doi.org/10.3969/j.issn.1006-2785.2008.07.003
2008-01-01
Abstract:Objective To evaluate the reliability of different MRI features in diagnosis of subaxial cervical ligment injury. Methods MRI was performed in 132 patients with subaxial cervical spinal injury before operation. MRI features of discontinuity on sagittal T1-weighted image(T1D),longitudinal high signal intensity on sagittal T2-weighted image(T2L),transversal high or intermediate signal intensity on sagittal T2-weighted image(T2T)was used to assess the status of anterior longitudinal ligament(ALL) and posterior longitudinal ligament(PLL) by two observers. The imaging results were compared with operational findings; the inter-observer consistency was determined by the kappa value. The sensitivity, specificity, accuracy, positive and negative predictive value of different MRI features in detection of ALL and PLL injury were calculated using operative findings as the gold standard. Results Inter-observer consistency of T1D for the assessment of ALL and PLL injury was not satisfactory (Kappa value=0.112, 0.285), and the sensitivity, specificity, accuracy, positive and negative predictive value of T1D was low. For T2L, the inter-observer consistency was fair(Kappa value=0.667, 0.537), and there was a high specificity (78.8~85.9%, 90.3~97.2%), however, the sensitivity was only 39.4~50.0% and 23.8~49.2%. By contrast, the inter-observer consistency of T2T for the assessment of ALL and PLL injury was excellent (Kappa value= 0.852, 0.799), and there was high sensitivity, specificity, accuracy, positive and negative predictive value. Conclusion The T1D and T2L of MRI cannot be used as reliable indicators of ligament injury in subaxial cervical spine. However, the T2T of MRI is recommended for detection of ALL and PLL injury in subaxial cervical spine and would be helpful in assessing the stability of spine.
What problem does this paper attempt to address?