Purpose: To prospectively evaluate the accuracy of magnetic resonance (MR) imaging for the preoperative diagnosis of deep pelvic endometriosis and extension of the disease.
Materials and methods: One hundred ninety-five patients (mean age, 34.2 years; range, 20-71 years) who were suspected of having pelvic endometriosis were recruited at two institutions. Two experienced radiologists evaluated the MR images independently. Deep pelvic endometriosis was defined as implants or tissue masses that appeared as hypointense areas and/or hyperintense foci on T1- or T2-weighted MR images in the following locations: torus uterinus, uterosacral ligaments (USLs), vagina, rectovaginal septum, rectosigmoid, and bladder. MR imaging results were compared with surgical and pathologic findings. Sensitivity, specificity, predictive values, and accuracy of MR imaging for prediction of deep pelvic endometriosis were assessed.
Results: Pelvic endometriosis was confirmed at pathologic examination in 163 (83.6%) of 195 patients. Endometriomas, peritoneal lesions, and deep pelvic endometriosis were diagnosed on the basis of surgical findings, alone or combined with pathologic findings, in 111 (68.1%), 83 (50.9%), and 103 (63.2%) of 163 patients, respectively. Torus uterinus and USL were the most frequent sites of deep pelvic endometriosis. The sensitivity, specificity, positive and negative predictive values, and accuracy of MR imaging for deep pelvic endometriosis were 90.3% (93 of 103), 91% (84 of 92), 92.1% (93 of 101), 89% (84 of 94), and 90.8% (177 of 195), respectively. The sensitivity, specificity, and accuracy, respectively, of MR imaging for the diagnosis of endometriosis in specific sites were as follows: USL, 76% (57 of 75), 83.3% (100 of 120), and 80.5% (157 of 195); vagina, 76% (16 of 21), 95.4% (166 of 174), and 93.3% (182 of 195); rectovaginal septum, 80% (eight of 10), 97.8% (181 of 185), and 96.9% (189 of 195); rectosigmoid, 88% (53 of 60), 97.8% (132 of 135), and 94.9% (185 of 195); and bladder, 88% (14 of 16), 98.9% (177 of 179), and 97.9% (191 of 195).
Conclusion: MR imaging demonstrates high accuracy in prediction of deep pelvic endometriosis in specific locations.
Copyright RSNA, 2004