Purpose: To determine whether magnetic resonance (MR) imaging with diffusion-weighted (DW) imaging can help discriminate between radiologically indeterminate benign and malignant orbital masses and to identify optimal apparent diffusion coefficient (ADC) thresholds for such discrimination.
Materials and methods: Informed consent was waived for this HIPAA-compliant institutional review board-approved retrospective study. Forty-seven orbital masses imaged with echo-planar DW imaging were identified in 47 patients (25 female patients, 22 male patients; average age, 35 years). A fellowship-trained orbital surgeon determined reference-standard diagnoses on the basis of chart review, and a neuroradiology fellow and senior neuroradiologist who were blinded to the diagnoses selected a region of interest for each lesion by consensus. ADC was calculated from signal intensity on DW images obtained with b = 1000 and b = 0 sec/mm(2). Lesion ADC was also compared with that of normal-appearing white matter (ADC ratio). The Student t test was used to compare groups. Receiver operating characteristic analysis was performed. Intraobserver agreement was assessed with a repeat data collection.
Results: Malignant lesions had lower ADCs than benign lesions, irrespective of patient age (P < .02) and in adults specifically (P < .05). Lymphomas had lower ADCs than pseudotumors (P < .001). An ADC of less than 1.0 x 10(-3) mm(2)/sec and an ADC ratio of less than 1.2 were optimal for predicting malignancy (sensitivity, 63% for both; specificity, 84% and 90%, respectively; and accuracy, 77% and 81%, respectively). Lymphoma was differentiated from pseudotumor with 100% accuracy (in 16 of 16 cases) by using these values. Infiltrative lesions that were hypointense on T2-weighted images were better characterized with DW imaging than lesions that were hyperintense or well defined.
Conclusion: Echo-planar DW MR imaging can help characterize indeterminate orbital masses.