EP13.44: The accuracy of ultrasound in detecting pelvic inflammatory disease sequelae: our experience
Year: 2026
Type: conference-abstract
Abstract: Objectives: To evaluate the diagnostic accuracy, interobserver agreement, and independent predictive value of individual Morphological Uterus Sonographic Assessment (MUSA) features for adenomyosis.Methods: Patients who underwent transvaginal ultrasound prior to hysterectomy with histopathology as the reference standard were included.Five independent observers assessed stored ultrasound examinations and recorded the presence of individual MUSA features.Diagnostic accuracy was calculated for each feature.Interobserver agreement was assessed using Fleiss' and Cohen's kappa.Mixed-effects logistic regression evaluated associations between features and histologically confirmed adenomyosis, accounting for repeated measures by observer.Results: Diagnostic performance of individual features was variable, with accuracy ranging from 34.4% to 72.7%.PPV was consistently high (70-90%), whereas NPV was low (<50%).Interobserver agreement was low for most features, with Fleiss' kappa ranging from 0.006 (hyperechogenic islands) to 0.44 (translesional vascularity), indicating poor to moderate agreement.Direct features showed particularly low reproducibility.Mixed-effects logistic regression demonstrated no significant association between any individual feature and histological adenomyosis (all p > 0.05), with wide confidence intervals.Irregular junctional zone frequently coexisted with at least one direct feature (65.9-100%), suggesting feature clustering.Conclusions: Individual MUSA features demonstrate limited diagnostic accuracy, poor reproducibility, and no independent predictive value for adenomyosis in a multiobserver setting.These findings suggest that reliance on single sonographic features is insufficient for diagnosis.Greater emphasis should be placed on composite diagnostic models, standardized interpretation, and structured training to improve reliability and clinical utility.
Institution: Carol Davila University of Medicine and Pharmacy
Language: English
Subfield: Reproductive Medicine
Field: Medicine
Domain: Health Sciences
Open Access status: bronze
OpenAlex ID: W7204789380
Cites:
Cited by:
Related to: