Are T1-Weighted Three-Dimensional Magnetic Resonance Images Inferior to T2-Weighted Images for Diagnosing Lumbar Foraminal Stenosis in the Fifth Lumbar Nerve Root? A Prospective, Comparative Study in Identical Patients.

ABSTRACT

Introduction: Imaging analysis of foraminal stenosis in the fifth lumbar (L5) nerve root remains to be a challenge because of the anatomical complexity of the lumbosacral transition. T2-weighted three-dimensional (3D) magnetic resonance images (MRI) have been dominantly used for diagnosis of lumbar foraminal stenosis, while the reliability of T1-weighted images (WI) has also been proven. In this study, we aim to compare the reliability and reproducibility of T1- and T2-weighted 3D MRI in diagnosing lumbar foraminal stenosis (LFS) of the L5 nerve root. Methods: In this study, 39 patients with unilateral L5 radiculopathy (20 had L4-L5 intracanal stenosis; 19 had L5-S foraminal stenosis) were enrolled, prospectively. T1- and T2-weighted 3D lumbar MRI were obtained from each patient. T1WI and T2WI were blinded and then separately reviewed twice by four examiners randomly. The examiners were instructed to answer the side of LFS or absence of LFS. The correct answer rate, sensitivity, specificity, and area under the curve were analyzed and compared between T1WI and T2WI. Also, intra- and interobserver agreements were calculated using kappa (κ)-statistics and compared in the same manner. Results: The average correct answer rate, sensitivity, specificity, and area under the curve of the T1WI/T2WI were 84.6%/80.1%, 82.9%/80.3%, 86.3%/81.3%, and 0.846/0.801, respectively. The intraobserver κ-values of the four examiners ranged from 0.692 to 0.916 (average 0.762) and from 0.669 to 0.801 (average 0.720) for T1WI and T2WI, respectively. The interobserver κ-values calculated in a round-robin manner (24 combinations in total) ranged from 0.544 to 0.790 (average 0.657) and from 0.524 to 0.828 (average 0.652), respectively. Conclusions: As per our findings, T1- and T2-weighted 3D MRI were determined to have nearly equivalent reliability and reproducibility in terms of diagnosing LFS of the L5 nerve root.