AI Accurately Detects Placenta Accreta Spectrum in Ultrasound Images

A new artificial intelligence (AI) model demonstrates promising accuracy in identifying placenta accreta spectrum (PAS) during pregnancy, potentially improving prenatal diagnosis of this serious condition. The analysis, presented at The Pregnancy Meeting on February 8-13, 2026, in Las Vegas, suggests AI could become a valuable tool for clinicians.

Understanding Placenta Accreta Spectrum

PAS is a complication where the placenta abnormally attaches to the uterine wall, often linked to prior cesarean deliveries. According to data from the American College of Obstetricians and Gynecologists (ACOG), the prevalence of PAS has risen significantly in recent decades. A 2016 study indicated a rate of one in 272 for women with a birth-related hospital discharge diagnosis, compared to one in 533 from 1982 to 2002.

AI Model Performance

Researchers, led by Alexandra L. Hammerquist, MD, a maternal-fetal medicine fellow at Baylor College of Medicine, developed an AI model to interpret 756 2D obstetric ultrasound images – a total of 38,907 grayscale frames – from 113 patients at risk for PAS between 2018 and 2025. The model accurately predicted the presence or absence of PAS in 88% of cases. Importantly, the model demonstrated 100% sensitivity, meaning it identified all actual cases of PAS without any false negatives.

Did You Know? PAS is a leading cause of maternal morbidity and mortality in the U.S., yet only 30% of cases are diagnosed before delivery.

The model also achieved a specificity of 75%, a positive predictive value of 81.8%, and a negative predictive value of 100%. The area under the receiver operating characteristic curve was 0.972. Researchers noted that the model’s key identifying features focused on the placental interface, supporting its biological plausibility.

Future Implications

Dr. Hammerquist emphasized that while the study’s findings are promising, the AI model requires further validation before clinical implementation. She clarified that the model is intended to assist, not replace, trained sonographers and physicians. “Ideally, models such as ours would allow a physician to better synthesize a patient’s clinical diagnosis and establish a more sure diagnosis,” she stated.

Expert Insight: The high sensitivity of this AI model is particularly noteworthy, as it suggests a potential to significantly reduce the number of undiagnosed PAS cases, which are associated with increased maternal risk.

Frequently Asked Questions

What is placenta accreta spectrum?

PAS is a condition in which the placenta abnormally attaches to the uterine wall, often following prior uterine surgical procedures like cesarean delivery.

How accurate was the AI model?

The AI model accurately predicted the presence or absence of PAS in 88% of cases, with 100% sensitivity and 75% specificity.

Is this AI model ready for use in clinics?

According to Dr. Hammerquist, the model requires further validation before it can be used in clinical practice. It is designed to support, not replace, the expertise of medical professionals.

As AI continues to evolve, how might these technologies reshape the landscape of prenatal care and improve outcomes for both mothers and babies?