This white paper evaluates mid-trimester ultrasound workflow efficiency using Live ViewAssist™, a deep learning-based tool integrated into obstetric ultrasound platforms.
The study was a prospective observational analysis conducted at a single center, including 80 singleton pregnancies undergoing routine second trimester ultrasound at 19–24 weeks of gestation. All examinations followed ISUOG guidelines for fetal anatomy and biometry assessment. Each patient underwent two examinations: one performed conventionally by manual acquisition of all required views and another performed using Live ViewAssist™.
Live ViewAssist™ automatically classifies fetal structures in real time, annotates images, performs measurements, and replaces suboptimal views during live scanning without requiring additional keystrokes. The software integrates existing tools such as ViewAssist™, BiometryAssist™, and HeartAssist™ to standardize plane acquisition and fetal biometry.
The primary outcome was total scan duration, measured from the first to the last acquired image. The median examination time was significantly longer with conventional ultrasound compared to Live ViewAssist™ (21 minutes [IQR 18–25] vs 10 minutes [IQR 8–12]; p < 0.01), representing a 52% reduction in scan time.
The findings suggest that AI-assisted mid-trimester ultrasound workflow may enhance efficiency, improve resource utilization, and support clinical practice, particularly in high-volume settings. The study focused on time efficiency rather than diagnostic accuracy and was conducted at a single center.
