Int J Gynaecol Obstet. 2026 Jan 9. doi: 10.1002/ijgo.70789. Online ahead of print.
ABSTRACT
BACKGROUND: The WHO recommends that all pregnant women receive an ultrasound (US) scan prior to 24 weeks gestation to encourage early identification of various conditions, such as fetal anomalies, multiple gestation, and placental abnormalities; however, global access to US remains limited. This has prompted many research groups to develop artificial intelligence (AI) approaches for obstetric US.
OBJECTIVE: The aim of this study was to update and synthesize current literature regarding the development of AI algorithms for obstetric US.
SEARCH STRATEGY: Methods were modified from Horgan et al. scoping review on the progress of AI algorithms for obstetric US. Our search, which encompassed papers published between 1991 and May 2022, adapted Horgan’s search strategy by replicating the search strings across PubMed, Cochrane Library, and clinicaltrials.gov databases, while also snowballing additional references.
SELECTION CRITERIA: Studies included both AI and obstetric US with a focus on one or more maternal and/or fetal conditions between January 2022 and January 2024. After removing duplicates, publications were screened for inclusion criteria based on their mention of both AI and obstetric US and a main objective assessing maternal and/or fetal conditions. Studies were excluded if they failed to mention the use of AI or obstetric ultrasound, discussed AI algorithm development, or consisted of expert opinions, reviews, and abstracts.
DATA COLLECTION AND ANALYSIS: We used Zotero to manage references and extracted data onto an Excel template. The remaining publications were reviewed for data extraction including-authors, dates, objectives, settings, and funding sources. Publications were categorized into seven main areas based on Horgan’s framework, with additional subcategories for emerging topics. Descriptive statistics summarized the data, with graphical visualizations depicting the geographic distribution of studies.
MAIN RESULTS: A total of 96 articles were included in the final results, revealing the rapid increase in the number of publications related to AI in obstetrics. The greatest proportion of studies were categorized as fetal biometry (25%) and anatomical evaluation of the fetus (20%). Studies took place across multiple regions with the greatest number in Asia (41%) and Europe (27%). A total of 22% were conducted in low- or middle-income countries (LMICs).
CONCLUSION: This scoping review demonstrates the growth and development of AI-enabled obstetric US applications. There is a wide variety of innovative applications on the horizon and implementation approaches and implications should be explored as these technologies become clinically available. We encourage development of algorithms that focus on parameters that identify conditions linked to the global burden of maternal and neonatal mortality and morbidity, such as gestational age and placental location.
PMID:41510613 | DOI:10.1002/ijgo.70789
