Measurement of Thoracic Inlet Alignment on MRI: Reliability and the Influence of Body Position.

Jun Qiao,Feng Zhu,Zhen Liu,Leilei Xu,Zezhang Zhu,Bangping Qian,Xu Sun,Qing Jiang,Yong Qiu
DOI: https://doi.org/10.1097/BSD.0000000000000306
2017-01-01
Abstract:Study design: A retrospective radiographic study. Objective: To investigate the reliability of using magnetic resonance imaging (MRI) to measure thoracic outlet parameters and whether supine MRI could supersede upright x-ray in measuring these parameters. Summary of Background Data: On x-ray, overlapped soft and bony tissues make sternum contour invisible. Thus, measuring thoracic inlet (TI) alignment on x-ray may be inaccurate. MRI could clearly show anatomy around TI. Methods: A total of 124 adolescents with idiopathic thoracolumbar/lumbar scoliosis were recruited. The visibility of T1 upper endplate and the upper end of sternum on the lateral standing x-ray films was rated. For patients with moderate or good clarity forT1 upper endplate and the upper end of sternum on x-ray films, thoracic inlet angle (TIA), T1 slope (T1S), and neck tilt (NT) were measured by 3 observers on standing x-ray films and supine MR images, respectively. Intraclass correlation coefficients (ICC) were used to determine the intraobserver, interobserver, and method reliability. The paired t test was performed to compare the measurements between the 2 methods. The correlation coefficients between the 2 methods were analyzed by regression analysis. Results: The visibility of TI region on x-ray films was unsatisfactory, whereas all the landmarks were clear on MR images. Only 81 patients' x-ray films were graded as moderate or good visibility in the TI region. Both the intraobserver ICC and interobserver ICC were better in the MRI set than in the x-ray set among all the observers for 3 TI parameters. The paired t test demonstrated that no significant difference was noted in terms of TIA, T1S, and NT (P = 0.572, 0.203, and 0.637, respectively). Regression analysis demonstrated high correlation coefficients for TIA, T1S, and NT (R = 0.612, 0.629, and 0.722, respectively). Conclusions: MRI serves as a good substitute for x-ray scans with regard to the measurement of TI alignment, with superior reliability.
What problem does this paper attempt to address?