Radiomics and Artificial Intelligence in Ovarian Endometriosis Imaging: A Systematic Review and Critical Appraisal of Emerging Evidence

Background: Ovarian endometriosis is one of the most common manifestations of endometriosis. Although transvaginal ultrasound and magnetic resonance imaging are central to diagnosis, radiomics and artificial intelligence have recently emerged as promising tools for quantitative lesion characterization. Objective: To systematically identify, critically appraise, and qualitatively synthesize the available evidence regarding radiomics and artificial intelligence applied to imaging-based characterization of ovarian endometriosis, with particular emphasis on diagnostic performance, methodological quality, clinical applicability, and current barriers to implementation. Methods: A systematic literature search was performed in PubMed, Scopus, and Web of Science. Eligible studies investigating radiomics or artificial intelligence applied to ultrasound, computed tomography, or magnetic resonance imaging of ovarian endometriosis were included. Data regarding study characteristics, radiomics workflows, validation strategies, and diagnostic performance were extracted and synthesized qualitatively. Methodological quality was assessed using QUADAS-2. Registration note: This registration was completed retrospectively after completion of the review process to improve transparency and document the predefined methodology used during study conduct.