Orientation-Robust Latent Motion Trajectory Learning for Annotation-free Cardiac Phase Detection in Fetal Echocardiography
A new AI system automatically pinpoints key moments in fetal heartbeats from ultrasound videos.
Deep Dive
Researchers developed ORBIT, a self-supervised AI that detects critical phases of the fetal heartbeat in ultrasound videos without manual labeling. It works regardless of the baby's position, learning the heart's motion pattern. Trained on normal scans, it accurately identified key frames in both healthy hearts and those with congenital disease, outperforming previous methods. This reduces a major manual bottleneck in prenatal heart screening.
Why It Matters
This could make crucial prenatal heart defect screenings faster, more consistent, and more accessible.