Enlarged Cardiothoracic Ratio Detection for Ebstein Anomaly Detection Using Artificial Intelligence

Authors

• Edgar A. Hernandez-Andrade
• Nathan S. Fox
• Andrei Rebarber
• Jennifer Lam-Rachlin
• Sara Garmel, Corewell Health East
• Nazbanou Heim
• Anna-Maria Coltri
• Roger Bessis
• Mia Ratsimandresy
• Phillippe Boukobza
• Malo de Boisredon
• Eric Askinazi
• Valentin Thorey
• Christophe Gardella
• Betrand Stos
• Marilyne Levy

Document Type

Conference Proceeding

Publication Date

1-2026

Publication Title

Pregnancy

Abstract

Objective: Cardiothoracic ratio (CTR) is a biomarker of altered fetal cardiac hemodynamics and may be useful to screen for Ebstein anomaly, a severe tricuspid valve malformation. However, current guidelines only provide general thresholds for CTR in the 2nd and 3rd trimesters. We aim to (1) create a nomogram of CTR for normal cases throughout pregnancy, (2) establish a gestational age (GA)-dependent threshold for enlarged CTR, (3) assess its relevance in screening for Ebstein anomaly, and (4) evaluate an AI software in detecting cases with enlarged CTR. Study Design: We evaluated 646 ultrasound exams of singleton pregnancies at the 2nd and 3rd trimester without cardiac abnormalities, with interpretable 4CH view, from 7 prenatal screening centers. Detailed area-based CTR were manually measured. A nomogram for CTR of normal cases was developed, and a GA-dependent threshold was defined as the 99th percentile (Figure 1). We evaluated this threshold on 59 fetal echocardiograms, including 41 with severe Ebstein anomaly (severe tricuspid regurgitation, functional pulmonary atresia, non functional right ventricle or hydrops) and 18 with mild variants. We then evaluated an AI software developed by BrightHeart to detect cases above the CTR threshold, based on an automated analysis of all grayscale cines in an exam. Results: Among the 646 normal exams, 5 cases had CTR above threshold (99.2% specificity, 95% CI: 98.2–99.7). CTR was above threshold in 46 of 59 exams with Ebstein anomaly, corresponding to a sensitivity of 78.0% (95% CI: 65.9–86.6) (92.7% (95% CI: 80.6–97.5) and 44.4% (95% CI: 24.6–66.3) for severe and mild cases, respectively). In the 481 cases with a 4CH view in a cine, the AI had 100% (95% CI: 92.7–100) sensitivity to detect enlarged CTR and 97.9% (95% CI: 96.1–98.9). Results were consistent across subgroups of GA (Figure 2). Conclusion: AI can accurately estimate cases with enlarged CTR, which may be an effective screening tool for Ebstein anomaly, particularly when assessed among additional markers. Such screening is especially impactful for severe cases, which typically warrant immediate post-delivery evaluation.

Volume

2

Issue

S1

First Page

49

Comments

Society of Maternal-Fetal Medicine (SMFM) 2026 Pregnancy Meeting, February 8-13, 2026, Las Vegas, NV

Last Page

50

Recommended Citation

Hernandez-Andrade EA, Fox NS, Rebarber A, Lam-Rachlin J, Garmel S, Heim N, et al. Enlarged cardiothoracic ratio detection for Ebstein anomaly detection using artificial intelligence. Pregnancy. 2026 Jan;2(S1):49-50. doi:10.1002/pmf2.70175

DOI

10.1002/pmf2.70175