Diagnostic and staging accuracy of magnetic resonance imaging for the assessment of sinonasal disease.

BACKGROUND Concerns over radiation exposure with paranasal sinus computed tomography (CT) have recently evolved. Magnetic resonance imaging (MRI) may be more advantageous in this regard, but the diagnostic assessment of MRI in chronic rhinosinusitis (CRS) has not been reported. The purpose of this study was to determine the correlation between CT- and MRI-based staging and diagnosis of CRS. METHODS Paired CT and MRI scans of 89 adult patients who were imaged by both modalities within a 3-month time period for evaluation of pituitary disease were scored for sinus disease using the Lund-Mackay system in a randomized and blinded fashion. The Lund scores were compared for similarity, correlation, and diagnostic classification between modalities. RESULTS The mean Lund scores were 2.3 +/- 0.6 (95% CI) for CT-based staging and 2.1 +/- 0.5 for MRI-based staging with a median time interval between scans of 3 days. The difference means was not statistically significant (p = 0.444, paired t-test). Correlation analysis revealed a significant association between CT- and MRI-based scores (Pearson's r = 0.837, p < 0.001). Disease classification agreement analysis using published Lund score cutoffs (3 versus 4) for the likelihood of true sinus disease revealed that CT- and MRI-based scoring agreed on 76 cases (85.4%). Disagreement occurred in 13 cases (kappa: 0.557, p < 0.001). Sensitivity, specificity, positive predictive value, and negative predictive value were 66.7, 90.1, 63.2, and 91.4%, respectively. CONCLUSION Lund-Mackay staging of sinus disease by MRI is closely correlated to corresponding staging based on CT. MRI does not significantly overstage or overclassify patients with sinus disease.