ARTIFICIAL INTELLIGENCE IN GYNECOLOGICAL ULTRASONOGRAPHY: DIAGNOSTIC CAPABILITIES, CLINICAL VALIDATION, AND LIMITATIONS

Abstract

Artificial intelligence (AI) is increasingly proposed as an adjunct  to  gynecological  ultrasonography,  including  assessment  of  adnexal  masses,  endometrial  lesions,  adenomyosis,  uterine  leiomyomas,  and  reproductive  imaging.  However, model  discrimination  reported  in  curated  datasets  does  not  automatically establish safety, calibration, transportability, or patient benefit. Objective: To critically  evaluate  diagnostic  opportunities,  methodological  weaknesses,  real-world  evidence  pathways,  and  regulatory  considerations  relevant  to  AI-assisted  gynecological  ultrasound.  Methods:  A  focused  evidence-informed  narrative  synthesis  considered  representative clinical accuracy studies, systematic  reviews, reporting standards, and  guidance  from  professional  and  regulatory  bodies  through  October  2026.