Objective: To evaluate the utility of fat-suppressed magnetic resonance imaging (MRI) in the diagnosis of endometriosis.
Design: A prospective clinical trial.
Setting: A government research hospital.
Patient(s): Forty-eight women with pelvic pain.
Intervention(s): Magnetic resonance imaging followed by surgical excision and pathologic diagnosis of endometriosis.
Main outcome measure(s): Presence and extent of endometriosis suggested by preoperative MRIs compared with surgical inspection and biopsy.
Result(s): A preoperative MRI in 46 women detected fewer endometriosis lesions than histopathology or laparoscopy (78 vs. 101 vs. 150). Few MRI lesions correlated with those identified by laparoscopy (50 of 150) or pathology (38 of 101). Of 42 women with surgically diagnosed endometriosis, 28 had at least one corresponding abnormality on MRI, 5 had abnormalities that didn't correlate with surgical findings, and 9 had normal MRIs. The sensitivity of MRI in detecting biopsy-proven endometriosis for any woman was 69% (25 of 36), and the specificity was 75%.
Conclusion(s): Although MRI identifies fewer areas of endometriosis than seen at surgery, it suggested endometriosis in 75% of those with at least mild disease. Only 67% of lesions identified at surgery contained histologic evidence of endometriosis.