Non-invasive diagnosis of endometriosis: From expert imaging to molecular and artificial intelligence-enabled pathways.

Authors

Abstract

Endometriosis affects approximately one in 10 women and people assigned female at birth of reproductive age, yet diagnosis is still delayed by years, historically because confirmation required laparoscopy. This review synthesizes which non-invasive tools-expert ultrasound and magnetic resonance imaging (MRI), molecular and microRNA (miRNA) biomarkers, and artificial intelligence (AI)-are ready for clinical use and which have so far been evaluated only in research settings and proposes a pragmatic clinical pathway. This is a narrative review and not a systematic review. Major guidelines now endorse a clinical and imaging-based diagnosis without mandatory laparoscopy. Expert transvaginal ultrasound and MRI identify ovarian endometrioma and deep endometriosis with high specificity and map disease for surgery, but a normal scan does not exclude endometriosis because superficial peritoneal endometriosis remains largely invisible. No blood, urinary, or endometrial biomarker meets criteria for clinical use; circulating miRNA panels have failed independent validation, and a salivary miRNA signature still needs independent, lower-prevalence validation. Radiomics and AI remain investigational pending external validation, calibration, and demonstrated clinical utility. Non-invasive diagnosis is best framed not as one replacement test but as a multimodal, staged pathway in which each tool is used according to how far its evidence has progressed, which shortens diagnostic delay, targets specialist care, and interprets a negative scan safely.

Tags

Topics