Diagnostic Accuracy of MRI Using the #Enzian Classification in Endometriosis: A Single-Center Retrospective Cohort Study

This retrospective, observational study was conducted at Galilee Medical Center in Nahariya, Israel. Medical records of women who underwent laparoscopic surgery for suspected endometriosis between January 2016 and December 2023 were reviewed. Inclusion criteria included age 18–50 years, laparoscopic confirmation of endometriosis, and availability of a preoperative pelvic MRI. Patients were excluded if MRI or surgical findings could not be evaluated using the #Enzian classification or if essential clinical data were missing. Institutional Review Board (IRB) approval was obtained (protocol number NHR-0184-22). Clinical and demographic data were collected from electronic medical records, including age, body mass index (BMI), smoking status, obstetric history, prior pelvic or cesarean surgeries, pelvic inflammatory disease, and comorbid inflammatory bowel diseases (e.g., Crohn’s disease and ulcerative colitis). The #Enzian classification ( Fig. 1 ) is an MRI-based extension of the surgical Enzian system, designed to stage DIE across defined anatomical compartments comprehensively. This anatomically structured system categorizes lesions by location and size across defined compartments: A: vagina and rectovaginal space, B: uterosacral ligaments, cardinal ligaments, and pelvic sidewall, C: rectum, F: extragenital sites, including: FB: urinary bladder, FU: ureters, FO: other extragenital locations. Additional compartments include: P: peritoneum; O: ovaries; FI: intestinal sites beyond the rectum (e.g., sigmoid colon, small bowel; T: tubo-ovarian adhesions, with optional assessment of tubal patency. #Enzian classification: overview with potentially affected organs and compartments. Lesions are graded by extent (1–3) within compartments P, O, T, A, B, and C. For paired structures (e.g., ovaries, tubes, ureters), severity is documented separately for each side. Missing or nonvisualized organs are denoted by suffixes m (missing) and × (unknown) [ 5 ]. This system facilitates standardized preoperative imaging interpretation, aligning radiologic and surgical findings to promote coordinated surgical planning. The primary outcome of the study was the level of agreement between MRI and laparoscopic assessments of lesion size and location using the #Enzian classification. Secondary outcomes included diagnostic performance metrics – sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy – for each anatomical compartment and lesion type. Diagnostic concordance was evaluated using a three-point scale: 0 (no match), 1 (partial match), and 2 (complete match). Descriptive statistics were used to summarize demographic and clinical characteristics. Categorical variables were reported as frequencies and percentages; continuous variables as means with standard deviations or medians with interquartile ranges, as appropriate. Agreement between MRI and surgical findings was assessed using: Spearman’s correlation coefficient for ordinal data, Wilcoxon signed-rank test for paired non-parametric data and McNemar’s or Sign test for categorical comparisons. Diagnostic accuracy indices were reported with 95% confidence intervals. A p value of <0.05 was considered statistically significant. All analyses were conducted using IBM SPSS Statistics for Windows, version 27.0.

A total of 651 women underwent laparoscopy for gynecologic indications at Galilee Medical Center between July 2016 and June 2022. Among them, 62 underwent surgery specifically for DIE. Of these, 40 patients were excluded due to either the absence of pelvic MRI before or at the time of laparoscopy or failure to meet the age criteria. The final study cohort consisted of 24 patients who had both a preoperative pelvic MRI and laparoscopic surgery. Each case was evaluated using the #Enzian classification, applied to both MRI and surgical findings. A total of 64 anatomical compartments were assessed across the study cohort. A total of 24 women met the inclusion criteria. The mean age was 34.29 years (±7.75), with a range from 19 to 49 years. The mean BMI was 23.86 kg/m 2 (±4.65). On average, patients reported 1.71 pregnancies, 1.60 births, and 0.58 miscarriages. Most had a history of previous abdominal surgeries (mean 0.86) and pelvic inflammatory disease (mean 0.56) ( Table 1 ). Baseline demographic and clinical characteristics of the study population PID, pelvic inflammatory disease. Out of 64 total anatomical compartments assessed using the #Enzian classification, MRI correctly identified the location of endometriotic lesions in 20 compartments, yielding an overall accuracy of 31.3% ( Table 2 ). The highest detection rates were in the tubo-ovarian region (T) at 60%, the rectovaginal septum (B) at 42.9%, and the ovaries (O) at 41.7%. In contrast, accuracy was notably poor in compartments such as the parametrium (P, 7.1%) and nonexistent in compartments C, F, FA, and FB. MRI accuracy in localizing endometriotic lesions by compartment according to the #Enzian classification A, rectovaginal septum/vagina; B, uterosacral/sacrouterine ligaments; C, rectum/sigmoid; O, ovaries; T, tubes/tubo-ovarian; P, peritoneum; F, extragenital organs; FA, urinary bladder; FB, ureters. MRI successfully estimated lesion size in 13 of 64 compartments (20.3%) ( Table 3 ). The best performance was again in the tubo-ovarian region (T, 60%) and rectovaginal septum (B, 35.7%). Accuracy was particularly low or absent in all other compartments, especially in the ovaries (O, 8.3%) and parametrium (P, 0%). MRI accuracy in estimating lesion size by anatomical compartment according to the #Enzian classification A, rectovaginal septum/vagina; B, uterosacral/sacrouterine ligaments; C, rectum/sigmoid; O, ovaries; T, tubes/tubo-ovarian; P, peritoneum; F, extragenital organs; FA, urinary bladder; FB, ureters. Significant correlations were observed between successful lesion localization on MRI and several patient characteristics ( Table 4 ). Patients with accurate MRI lesion localization were significantly older (mean age 38.38 vs. 32.51 years, p = 0.031). They also had higher gravidity (mean 3.00 vs. 1.00, p = 0.050) and parity (mean 2.27 vs. 1.08, p = 0.023). Additionally, the presence of soft or loose stools was more frequent among patients with successful MRI localization (38.5% vs. 7.8%, p = 0.013). The analysis revealed a trend indicating that MRI accuracy in estimating lesion size improved with increasing patient age. Correlation of MRI accuracy with clinical and demographic variables *Statistically significant ( p < 0.05).

Pelvic MRI, when interpreted through the #Enzian classification, provides a structured and anatomically detailed framework for assessing endometriosis. While it demonstrates reasonable performance in evaluating DIE, its limited accuracy in detecting superficial and parametrial disease restricts its role as a standalone diagnostic tool. Our findings align with existing literature and reaffirm the continued necessity of laparoscopy for comprehensive diagnosis and surgical planning. To enhance diagnostic consistency and reliability, future studies should prioritize the development and adoption of standardized MRI imaging protocols and radiologist training programs specifically tailored to the #Enzian classification system.

This study assessed the diagnostic performance of pelvic MRI in mapping endometriotic lesions using the #Enzian classification, comparing radiologic findings with laparoscopic observations. Our results showed that MRI correctly identified lesion location in 31.3% and lesion size in only 20.3% of anatomical compartments. These findings underscore the limitations in providing a complete preoperative map of disease distribution, especially outside DIE regions. Prior studies have reported more favorable results in specific compartments. Bazot et al. [ 7 ] found high sensitivity for rectosigmoid (77–93%) and uterosacral ligament (69%) involvement, along with high specificity. Similarly, the multicenter study by Pausch et al. [ 4 ] demonstrated substantial to excellent inter-reader agreement for compartments such as the rectovaginal septum and ovaries (Cohen’s kappa 0.75–0.96), while confirming the limited reliability of MRI for detecting peritoneal lesions [ 6 ]. Our study reinforces these observations. The highest detection rates were in the tubo-ovarian region (60%) and rectovaginal septum (42.9%), consistent with established literature. In contrast, parametrial lesions were detected in just 7.1% of cases, and compartments C, F, FA, and FB had no correct identifications, reflecting the known limitations of MRI in these areas [ 4 ]. The discrepancy in accuracy between our findings and those from larger multicenter studies may reflect differences in reader experience and variability in MRI protocols across referring centers. As suggested by Pausch et al. [ 4 ] imaging interpretation is significantly influenced by radiologist expertise and protocol standardization. Interestingly, our study identified patient-related factors associated with improved MRI accuracy. Older age, higher parity, and history of previous pregnancies were significantly linked to better lesion detection. These factors may contribute to altered pelvic anatomy or hormonal environments, making lesions more conspicuous on imaging – a phenomenon previously suggested in the literature [ 8 ]. In addition, we observed a correlation between gastrointestinal symptoms, particularly loose stools, and correct lesion localization, which may indicate deeper bowel involvement. This supports earlier findings that linked specific symptoms with DIE extent and location [ 9 ]. Although MRI is a valuable noninvasive tool for surgical planning, it remains a complementary modality rather than a replacement for laparoscopy. The American College of Radiology emphasizes that while MRI is particularly useful in preoperative evaluation of DIE when ultrasound is inconclusive, it does not reliably detect superficial peritoneal disease or evaluate key fertility-related factors such as tubal mobility or patency [ 10 ]. The #Enzian classification has proven effective in standardizing communication between radiologists and surgeons by offering a structured, compartmental approach. Studies confirm its high inter-reader reliability and diagnostic value in compartments such as the rectovaginal septum, uterosacral ligaments, and ovaries. However, accuracy for peritoneal disease remains limited and is heavily dependent on imaging quality and radiologist training [ 4 , 11 , 12 ]. The cEnzian multicenter study further supports its utility as a tool for summarizing diagnostic workups, but also highlights the variability in accuracy due to imaging modality and reader experience [ 13 ]. The retrospective design and small sample size are key limitations of this study. Additionally, MRI scans were performed across different institutions, likely introducing variability in both image quality and interpretation. As reported by Hötker et al. consistent and accurate #Enzian staging with MRI requires specialized radiologists and standardized imaging protocols [ 8 ]. Future research should focus on prospective, multicenter studies with uniform MRI protocols and targeted training programs to improve diagnostic consistency, particularly in compartments that currently show poor accuracy.

Endometriosis is a chronic, estrogen-dependent condition marked by endometrial-like tissue outside the uterus, most commonly on the ovaries and peritoneum. This ectopic tissue responds to hormonal cycles, causing pelvic pain, dysmenorrhea, dyspareunia, and infertility. The exact cause is unclear, though theories include retrograde menstruation, immune dysfunction, and genetic factors. Diagnosis is often delayed and based on clinical, imaging, and sometimes surgical findings. Treatment is tailored to symptoms and fertility goals and may involve hormonal therapy, pain management, or surgery [ 1 ]. Clinically, it often manifests as progressive dysmenorrhea and deep pelvic dyspareunia, with some patients also reporting chronic pelvic pain, which may be attributed to associated adhesions and pelvic scarring [ 2 , 3 ]. Laparoscopy has traditionally been considered the gold standard for diagnosing endometriosis. However, due to its invasive nature and the complex anatomy associated with deep infiltrating endometriosis (DIE), there has been increasing interest in noninvasive diagnostic methods to improve preoperative assessment and surgical planning. Magnetic resonance imaging (MRI) has emerged as a highly valuable tool for detecting and mapping endometriotic lesions, particularly DIE. Several studies have demonstrated its high sensitivity and specificity in evaluating disease involvement in critical locations such as the uterosacral ligaments and rectovaginal septum. For example, Pausch et al. [ 4 ] reported substantial to excellent inter-reader agreement in evaluating different compartments of endometriosis on MRI, highlighting its reliability in preoperative settings. To enhance communication between radiologists and surgeons and standardize disease description, the #Enzian classification was developed. This system provides a structured framework for categorizing superficial peritoneal, ovarian, deep infiltrating, and extragenital endometriosis, as well as pelvic adhesions, facilitating consistent documentation and treatment planning [ 5 ]. However, a 2023 prospective study by Harth et al. [ 6 ] evaluated the application of the #Enzian classification for endometriosis on MRI and found that, with sufficient experience, radiologists achieved excellent inter- and intraobserver agreement in MRI-based diagnosis of DIE and endometriomas. Specifically, for experienced readers, interobserver agreement in the assessment of compartments A, B, and C was excellent (κ ranging from 0.84 to 0.89), and intraobserver variability was also excellent (κ ranging from 0.85 to 0.95). These findings support the feasibility and reliability of integrating MRI with the #Enzian classification in clinical practice. Nonetheless, limitations persist. The sensitivity of MRI in detecting superficial peritoneal lesions remains suboptimal, and diagnostic accuracy can vary depending on the radiologist’s experience and the imaging protocols used. Pausch et al. [ 4 ] noted reduced sensitivity for peritoneal involvement, emphasizing the need for a multimodal diagnostic approach that combines imaging, clinical evaluation, and surgical findings. Given these challenges and developments, the present study aims to evaluate the utility of pelvic MRI, interpreted using the #Enzian classification, in aiding laparoscopic management of endometriosis. By correlating MRI findings with intraoperative outcomes, this study aims to evaluate the accuracy of MRI in guiding surgical decision-making and optimizing patient care.

The authors have no conflicts of interest to declare.

No sponsor or funder supported this study.

The Institutional Review Board (Helsinki Committee) of the Galilee Medical Center reviewed and approved this study protocol, approval No. (NHR-0184-22). Written informed consent was not necessary for this study due to its retrospective nature. This decision received approval from the Institutional Review Board (Helsinki Committee) of the Galilee Medical Center, approval no. (NHR-0184-22).

Ala Aiob and Dina Gumin: planning, protocol development, data management, data analysis, and manuscript writing. Amirit Luder Stinski and Lior Lowenstein: protocol development, data collection, and management. Avishalom Sharon: planning, protocol development, data management, data analysis, and manuscript writing, editing and revision. All authors interpreted the data, reviewed drafts, and approved the submitted manuscript.