Performance of Nonaugmented Routine Pelvic Ultrasound for Deep Endometriosis Using Static 2024 SRU Consensus Criteria: A Multireader Study

RATIONALE AND OBJECTIVES: To decrease the existing diagnostic delay of pelvic endometriosis, the Society of Radiologists in Ultrasound (SRU) has proposed consensus criteria using augmented pelvic ultrasound (APU) technique and reporting guidelines based on specific observations. We examined the feasibility of detecting pelvic endometriosis observations by retrospectively applying the static SRU criteria to nonaugmented routine pelvic ultrasound (RPU) performed in everyday practice. MATERIALS AND METHODS: This single-center retrospective institutional review board (IRB)-approved study included women (18-55 yrs), with RPU performed for chronic pelvic pain and suspicion of endometriosis, between September 2018 to March 2024. Four radiologists evaluated RPU images using the static or non-dynamic 2024 SRU consensus criteria excluding those that required dynamic maneuvers including the sliding sign. Ground truth was established by an independent expert reader using pelvic magnetic resonance imaging (MRI), laparoscopy, and histopathology. Fleiss' and quadratic weighted Kappa for inter-reader agreement and sensitivity; specificity and accuracy for each observation were calculated. RESULTS: In 118 women (median age, 37 years [Interquartile range (IQR), 30.0-43.8 years]), the sensitivity, specificity and accuracy for ovarian endometrioma (OMA) was 61-77%; 84-93%; 73-82%; p < 0.001. Inter-reader agreement for OMA was moderate (Fleiss' kappa of 0.742 [95% CI: 0.65,0.822]) and for deep pelvic endometriosis (DE) observations was low (Fleiss' kappa of 0.002 [95% CI: -0.007, 0.081]). CONCLUSION: RPU, although reliable for OMA, has limited performance for DE in a high suspicion patient even if 2024 SRU non-dynamic consensus observations are applied, highlighting the importance of adding APU to RPU and reader training, for early detection of endometriosis.