Diagnosis of presence and extent of deep infiltrating endometriosis by preoperative MRI-improvement of staging accuracy by expert training.

Abstract Introduction This study evaluates the clinical utility of magnetic resonance imaging (MRI) for the determination of presence and extent of DIE with special emphasis on effects of MRI reporting training Material and methods Data from 80 patients with clinically suspected DIE presented at our certified endometriosis center between 2015 and 2018 were analyzed. For all patients an ENZIAN score (describing DIE related to individual anatomical localizations) was obtained based on the preoperative MRI findings. The intraoperatively determined ENZIAN score served as the reference for assessment of diagnostic performance of the MRI. Results Overall, sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for the diagnosis of DIE by MRI were 76.9%, 53.3%, 87.7% and 34.8%, respectively. Analysis by compartment revealed a sensitivity, specificity, PPV and NPV of 59.5%, 88.2%, 86.2% and 63.9%, respectively, for compartment A, with similar values for compartment B, and 50.0%, 88.9%, 64.7% and 81.4%, respectively, for the less often affected compartment C. Expert training (n = 32 before, n = 48 after) led to a considerable increase in sensitivities for the overall detection of DIE (84.6% vs. 65.4%, p = 0.071) and for the detection of DIE in compartment A (71.4% vs. 35.7%, p = 0.026), compartment B (66.7% vs. 37.5%, p = 0.057) and compartment C (75.0% vs. 20.0%, p = 0.010), without significant loss in specificity (all p > 0.50). Discussion After expert training, MRI has a good sensitivity with fair specificity regarding preoperative assessment of presence, location and extent of DIE.