BACKGROUND: The purposes of this study were to assess the reliability of 3-dimensional magnetic resonance (MR) imaging (3D MRI) and conventional MRI (CMRI) for detection of lumbar intra and/or extra-foraminal stenosis (LIEFS) and to compare the diagnostic accuracy of the 2 imaging modalities. METHODS: A total of 60 sets of 3D MR and CMR images from 20 healthy volunteers and 40 LIEFS patients were qualitatively rated according to defined criteria by 3 independent, blinded readers. Kappa statistics were used to characterize intra and inter-reader reliability for qualitative rating of data. Multireader, multicase analysis was used to compare lumbar foraminal stenosis detection between the 2 modalities. RESULTS: Intra-reader agreement for 3D MRI was excellent, with kappa = 0.90; that for CMRI was good, with kappa = 0.78. Average inter-reader agreement for 3D MRI was good, with kappa = 0.79, whereas that for CMRI was moderate, with kappa = 0.41. Average area under the ROC curve values (1st reading/2nd reading) for detection of lumbar foraminal stenosis using 3D MRI and CMRI were 0.99/0.99 and 0.94/0.92, respectively. Detection of LIEFS with 3D MRI was significantly better than with CMRI (P = 0.0408/0.0294). CONCLUSIONS: These results suggest that CMRI was of limited use for detection of the presence of LIEFS. Isolated imaging with CMRI may risk overlooking the presence of LIEFS. In contrast, reliability of 3D MRI for detection of LIEFS was good. Furthermore, readers' performance in the diagnosis of LIEFS can be improved by use of 3D MRI. Therefore, 3D MRI is recommended when using imaging for diagnosis of LIEFS.
BACKGROUND: The purposes of this study were to assess the reliability of 3-dimensional magnetic resonance (MR) imaging (3D MRI) and conventional MRI (CMRI) for detection of lumbar intra and/or extra-foraminal stenosis (LIEFS) and to compare the diagnostic accuracy of the 2 imaging modalities. METHODS: A total of 60 sets of 3D MR and CMR images from 20 healthy volunteers and 40 LIEFS patients were qualitatively rated according to defined criteria by 3 independent, blinded readers. Kappa statistics were used to characterize intra and inter-reader reliability for qualitative rating of data. Multireader, multicase analysis was used to compare lumbar foraminal stenosis detection between the 2 modalities. RESULTS: Intra-reader agreement for 3D MRI was excellent, with kappa = 0.90; that for CMRI was good, with kappa = 0.78. Average inter-reader agreement for 3D MRI was good, with kappa = 0.79, whereas that for CMRI was moderate, with kappa = 0.41. Average area under the ROC curve values (1st reading/2nd reading) for detection of lumbar foraminal stenosis using 3D MRI and CMRI were 0.99/0.99 and 0.94/0.92, respectively. Detection of LIEFS with 3D MRI was significantly better than with CMRI (P = 0.0408/0.0294). CONCLUSIONS: These results suggest that CMRI was of limited use for detection of the presence of LIEFS. Isolated imaging with CMRI may risk overlooking the presence of LIEFS. In contrast, reliability of 3D MRI for detection of LIEFS was good. Furthermore, readers' performance in the diagnosis of LIEFS can be improved by use of 3D MRI. Therefore, 3D MRI is recommended when using imaging for diagnosis of LIEFS.