Determining the lumbar vertebral segments on magnetic resonance imaging

Study Design. A study to test the ability of an additional cervicothoracic localizer scan to decrease interobserver discrepancy in the identification of vertebral segments in magnetic resonance imaging of the lumbar spine. Objectives. To investigate whether lumbar vertebral segments can be identif ied correctly from lumbosacral magnetic resonance localizer scans, the degr ee of interobserver discrepancy, and the value of an additional cervicothor acic localizer scan. Summary of Background Data. In magnetic resonance imaging of the lumbar spi ne, it may be difficult to identify transitional lumbosacral vertebral segm ents. Methods. The sagittal and coronal lumbosacral localizer scans of 141 consec utive patients referred for magnetic resonance imaging of the lumbosacral s pine were reviewed independently by two radiologists with the aim of locati ng the L5 vertebra. An additional sagittal cervicothoracic localizer scan a lso was performed in each case. The final study group consisted of 129 pati ents. The L5 vertebra was identified by counting caudally from C2 using the sagittal cervicothoracic and lumbosacral localizer scans. In the 54 most r ecently studied patients, cod liver oil capsule surface markers were placed near the thoracolumbar junction to quantify any marker shift between the t wo sagittal localizer scans. Results. The lumbar segments could be identified consistently by counting c audally using cervicothoracic and lumbosacral localizer scans. Using sagitt al lumbosacral localizer scans alone, the lumbar vertebral segments could b e identified correctly in only 80.2% of patients. Coronal lumbosacral local izer scans produced similar results (82.2%). The accuracy fell to 77.9% whe n using a combination of both sagittal and coronal lumbosacral localizer sc ans. There was a 11.6% interobserver discordance in assessment of these lev els. Lumbosacral transitional vertebrae were identified in 17 patients (13. 2%), including 8 sacralized L5 and 9 lumbarized S1 vertebrae. Apparent surf ace-marker shift between cervicothoracic and lumbosacral localizer scans wa s insignificant, averaging only 1.9 mm (range, 0.0-5.6 mm). Conclusions. The addition of a cervicothoracic localizer scan in magnetic r esonance imaging of the lumbosacral spine is highly recommended.