AI System Can Improve Detection of Congenital Heart Defects In Utero

TUESDAY, Feb. 4, 2025 (HealthDay News) — Artificial intelligence (AI) can improve detection of congenital heart defects (CHDs) on routine prenatal ultrasounds, according to a study presented at The Pregnancy Meeting, the annual meeting of the Society for Maternal-Fetal Medicine, held from Jan. 27 to Feb. 1 in Denver.

Jennifer Lam-Rachlin, M.D., from the Icahn School of Medicine at Mount Sinai Hospital in New York City, and colleagues examined whether an AI system can improve detection of CHDs on fetal ultrasound exams among general obstetricians and gynecologists (OB-GYNs) and maternal-fetal medicine (MFMs) subspecialists. The AI system analyzed grayscale two-dimensional ultrasound cines of an exam and detected eight morphological findings associated with severe CHDs; the presence of any finding was sufficient to refer the patient for further examination. A dataset of 200 ultrasound exams from 11 centers was obtained; 100 exams had suspicious findings. Fourteen physicians (OB-GYNS and MFMs) reviewed each exam aided and unaided by the AI system in randomized order.

The researchers found that the receiver operating characteristic area under the curve for detection of any finding was significantly higher for AI-aided versus unaided reviews (0.97 versus 0.83). The results were similar for sensitivity (0.94 versus 0.78) and specificity (0.97 versus 0.76). The mean reading time was shorter for AI-aided versus unaided review (226 ± 218 versus 274 ± 241 seconds).

“Our findings show that the AI-based software significantly improved detection of ultrasounds that were suspicious for congenital heart defects not only among OB-GYNs but also among maternal-fetal medicine subspecialists,” Lam-Rachlin said in a statement.

Press Release

More Information