Physicians in the Raquel and Jaime Gilinski Department of Obstetrics, Gynecology and Reproductive Science at Mount Sinai in 2025 became the first in New York City to implement an artificial intelligence (AI) tool that enhances fetal ultrasounds on a large scale—resulting in more accurate detection of congenital heart defects and better care for babies and families.
The AI tool was deployed at Carnegie Imaging for Women, a Mount Sinai-affiliated advanced imaging practice with three locations in Manhattan, becoming the first centers in New York City to use a U.S. Food and Drug Administration-cleared AI software tool from the company BrightHeart to make ultrasounds more accurate and efficient for the detection of congenital heart defects in the fetus.
In a study led by Mount Sinai West physicians, and published in January 2026 in Obstetrics & Gynecology, researchers used the AI technology to improve the detection rates of ultrasound findings suspicious for major congenital heart defects to more than 97 percent, with an 18 percent reduction in reading time and 19 percent improvement in confidence score.
“AI assistance in prenatal diagnosis offers not only improved detection, but has the potential to offer significant improvement in workflow and efficiency benefits,” said corresponding author Jennifer Lam-Rachlin, MD, Assistant Clinical Professor of Obstetrics, Gynecology and Reproductive Science at the Icahn School of Medicine at Mount Sinai and the Director of Fetal Echocardiography at Carnegie Imaging for Women. “We, as clinicians, should embrace innovation and technology that is available, in order to maximize quality patient care. This technology allows for ‘leveling’ of the field of prenatal diagnosis to offer close to expert-level review of fetal ultrasounds, particularly in centers or geographical locations without fetal heart experts.”
A. In this multireader, multicase study, obstetrician-gynecologists and maternal-fetal medicine specialists interpreted entire second-trimester fetal ultrasound examinations to identify the presence or absence of eight morphologic findings suspicious for severe congenital heart defects (CHDs), along with confidence scores, both aided and unaided by the AI-based software. B. Example annotated cine clip generated by the AI-based software. The AI-based software detects each finding as present (orange), absent (green), or inconclusive (white) and indicates times in the cine clip at when the finding can be seen as present or absent. Vertical bar represents the current frame. CTR, cardiothoracic ratio; RV/LV, right to left ventricular; TV/MV, tricuspid to mitral valve; Abn. OT, abnormal outflow tract; PV/AV, pulmonary to aortic valve.
Congenital heart defects are one of the most common birth abnormalities affecting nearly one percent of births per year in the United States About one in four heart defects is classified as a severe or critical congenital heart defect that requires urgent medical or surgical intervention for survival, according to the National Institutes of Health.
In the retrospective study, the researchers examined a dataset of 200 deidentified fetal ultrasound examinations between 18 and 24 weeks of gestation from 11 medical centers across two countries, including 100 with at least one suspicious finding for heart defect. The study aimed to evaluate the association between the use of AI-based software and reader performance in identifying second-trimester ultrasound examinations suspicious for severe congenital heart defects. Seven obstetrician-gynecologists and seven maternal-fetal medicine specialists (experts in high-risk pregnancies) reviewed each examination in randomized order, both with and without AI assistance, and assessed the presence or absence of each finding suspicious for congenital heart defects with confidence scores. Outcomes included readers' performance in identifying the presence of any finding and each finding at the examination level, as measured by the area under the receiver operating characteristic curve (AUROC), sensitivity, and specificity.
The detection of any suspicious finding significantly improved for AI-aided compared with unaided readers, with a significantly higher AUROC (97 percent vs. 82 percent). The study also demonstrated the ability of AI-based software to improve overall confidence and time efficiency in interpreting these scans.
“Our study should prompt and encourage future research into AI-assisted software’s ability to improve detection rates, once integrated into clinical workflows, to reduce the variability and inequity of detection of congenital heart defects globally,” says co-author Andrei Rebarber, MD, Director of the Division of Maternal-Fetal Medicine at Mount Sinai West and Clinical Professor of Obstetrics, Gynecology and Reproductive Science at the Icahn School of Medicine. “The future for prenatal diagnostic imaging is bright when AI software is employed as an adjunct to physician interpretation.”
BrightHeart, developer of the AI-based software, funded the study. Researchers from the following institutions contributed to this study: the Division of Pediatric Cardiology at the Icahn School of Medicine at Mount Sinai; New York University School of Medicine; Maternal Fetal Medicine Associates in New York City; Pediatrics-Cardiology at Stanford University School of Medicine in Palo Alto, California; Palo Alto Medical Foundation, Sutter Health, in Palo Alto, California; the Fetal Diagnostic Center of Pasadena in Pasadena, California; Université Grenoble Alpes, Centre Hospitalier Universitaire-Grenoble Alpes (CHU Grenoble Alpes) in Grenoble, France; Medical Training Center in Rouen, France; Centre d’Echographie de l’Odéon in Paris; UE3C-Unité d’Explorations Cardiologiques-Cardiopathies Congénitales in Paris; Cardiologie Pédiatrique-Hôpital Necker-Enfants Maladies in Paris; Michigan Perinatal Associates, Corewell Health East, in Dearborn, Michigan; Wayne State University School of Medicine in Detroit; Fetal Echocardiography and Perinatal Research-Valley Health System in Paramus, New Jersey; Division of Maternal Fetal Medicine, Pennsylvania Hospital, University of Pennsylvania in Philadelphia; and Maternal Fetal Medicine, Perinatal Specialists of the Palm Beaches in West Palm Beach, Florida.
For more Mount Sinai artificial intelligence news, visit: https://icahn.mssm.edu/about/artificial-intelligence.
Featured
Jennifer Lam-Rachlin, MD
Assistant Clinical Professor of Obstetrics, Gynecology and Reproductive Science, and Director of Fetal Echocardiography at Carnegie Imaging for Women
Andrei Rebarber, MD
Director of the Division of Maternal-Fetal Medicine at Mount Sinai West and Clinical Professor of Obstetrics, Gynecology and Reproductive Science