Quantitative analysis of dynamic contrast enhancement MRI between orbital lymphoma and inflammatory mass based on different regions of interest selection

Abstract

To assess the value of dynamic contrast enhanced MRI (DCE-MRI) quantitative parameters in differentiating orbital lymphoma from inflammatory mass based on different regions of interest selection. The study population consists of 83 patients with histologically proven orbital masses (39 lymphomas and 44 inflammatory masses). Two methods of selecting regions of interest (ROI) were compared to obtain quantitative parameters: (a) a round or ovoid region of 8 mm2 covering the most enhanced area, and (b) covering the lesion but excluding areas of hemorrhage or necrosis on the slice with the largest tumor area. Parametric maps were obtained for quantitative parameters including Ktrans, Ve and Kep. Quantitative parameters that distinguish orbital lymphoma from orbital inflammatory mass were analyzed by receiver operating characteristic curves (ROC). Kep values (ROI selection method a) were significantly higher in orbital lymphoma compared with orbital inflammatory mass (p = 0.003) and Ve values (ROI selection method a and b) of orbital lymphoma were significantly lower than orbital inflammatory mass (p = 0.01, p = 0.006, respectively). Kep (ROI selection method a) could distinguish lymphoma from inflammatory mass with sensitivity (97.4%, 38/39), specificity (52.3%, 23/44), positive predictive value (64.4%, 38/59), negative predictive value (95.8%, 23/24) and accuracy (73.5%, 61/83). Ve (ROI selection method a) could distinguish lymphoma from inflammatory mass with sensitivity (54.5%, 24/44), specificity (79.5%, 31/39), positive predictive value (75%, 24/32), negative predictive value (60.8%, 31/51) and accuracy (66.3%, 55/83). Ve (ROI selection method b) distinguished lymphoma from inflammatory mass with sensitivity (48.7%, 21/44), specificity (86.4%, 34/39), positive predictive value (80.8%, 21/26), negative predictive value (59.6%, 34/57), and accuracy (66.3%, 55/83). The pairwise comparison of the AUCs among these significant quantitative parameters showed no difference in distinguishing orbital lymphoma from orbital inflammatory mass (Z = 1.066, p = 0.286, Z = 0.998, p = 0.318 and Z = 0.370, p = 0.711, respectively). Quantitative parameters can differentiate orbital lymphoma from inflammatory mass. ROI selection methods had no definite effect on the diagnostic performance.