Is multidetector computed tomography comparable to magnetic resonance imaging for assessment of lumbar foraminal stenosis?

Background Both multidetector computed tomography (MDCT) and magnetic resonance imaging (MRI) are used for assessment of lumbar foraminal stenosis (LFS). Therefore, it is relevant to assess agreement between these imaging modalities. Purpose To determine intermodality, inter-, and intra-observer agreement for assessment of LFS on MDCT and MRI. Material and Methods A total of 120 foramina in 20 patients who visited our institution in January and February 2014 were evaluated by six radiologists with different levels of experience. Radiologists evaluated presence and severity of LFS on sagittal CT and MR images according to a previously published LFS grading system. Intermodality agreement was analyzed by using weighted kappa statistics, while inter- and intra-observer agreement were analyzed by using intraclass correlation coefficients (ICCs) and kappa statistics. Results Overall intermodality agreement was moderate to good (kappa, 0.478-0.765). In particular, two professors and one fellow tended to overestimate the degree of LFS on CT compared with MRI. For inter-observer agreement of all six observers, ICCs indicated excellent agreement for both CT (0.774) and MRI (0.771), while Fleiss' kappa values showed moderate agreement for CT (0.482) and MRI (0.575). There was better agreement between professors and fellows compared with residents. For intra-observer agreement, ICCs indicated excellent agreement, while kappa values showed good to excellent agreement for both CT and MRI. Conclusion MDCT was comparable to MRI for diagnosis and assessment of LFS, especially for experienced observers. However, there was a tendency to overestimate the degree of LFS on MDCT compared with MRI.