Optimising Screening For Severe Congenital Heart Defects on Prenatal Ultrasound Using Eight Suspicious Findings and Artificial Intelligence

Authors

A Rebarber
C M. Zelop
J Lam-Rachlin
A Arunamata
R Punn
S K. Behera
M Lachaud
N David
G R. DeVore
Sara Garmel, Corewell Health East

Document Type

Conference Proceeding

Publication Date

9-2025

Publication Title

Ultrasound in Obstetrics and Gynecology

Abstract

Objectives: Prenatal detection of severe congenital heart defects(CHD) is critical for improving neonatal outcomes. However, detection rates remain low, partly due to the difficulty of identifying fetal heart abnormalities during the routine anatomy scan. We aim to:1. evaluate the efficacy of eight suspicious findings designed to screen for severe CHD; and2. evaluate an artificial intelligence (AI) software designed to detect such findings. Methods: Eight ultrasound findings were designed based on guidelines and the medical literature to efficiently screen for severe CHD using a limited number of criteria (figure 1A). The presence of any of these findings warrant referral to a fetal heart specialist. We retrospectively included 280 anatomy ultrasounds and fetal echocardiograms of all major forms of severe CHD from 11 centres in the USA and France (singleton pregnancies, 18–24 weeks of gestation), with four chamber, LVOT and RVOT views featured in cine clips. A panel of expert pediatric cardiologists determined the presence or absence of the eight findings for each case. The AI software analysed all 2D grayscale ultrasound cine clips to assess the presence, absence, or inconclusiveness (in case of low confidence) of each finding. Results: Of the 280 severe CHD cases, 278 (99.3%, 95% CI:97.4–99.8%) had at least one finding present according to the pediatric cardiologists (figure 1). The AI software identified at least one finding as present in 271 cases (96.8%, 95% CI: 94.0–98.3%) and had an inconclusive output for 4 cases. Conclusions: The eight suspicious findings were strongly associated with severe CHD. Furthermore, using these findings, the AI software could accurately identify a wide range of severe CHD, and may thereby improve neonatal outcomes.

Volume

66

Issue

S1

First Page

36

Comments

35th World Congress on Ultrasound in Obstetrics and Gynecology, September 14-17, 2025, Cancun, Mexico

Last Page

36

Recommended Citation

Rebarber A, Zelop CM, Lam -Rachlin J, Arunamata A, Punn R, Behera SK, et al. [Garmel S]. Optimising screening for severe congenital heart defects on prenatal ultrasound using eight suspicious findings and artificial intelligence. Ultrasound Obstet Gynecol. 2025 Sep;66(S1):36. doi:10.1002/uog.70051

DOI

10.1002/uog.70051