Latent Motion Profiling for Annotation-free Cardiac Phase Detection in Adult and Fetal Echocardiography Videos

arXiv — cs.CVMonday, November 17, 2025 at 5:00:00 AM
The paper presents an unsupervised framework for detecting cardiac phases, specifically end-diastole (ED) and end-systole (ES), in echocardiography videos. This method utilizes self-supervised learning to analyze latent cardiac motion trajectories without the need for manual annotations. The framework was evaluated using the EchoNet-Dynamic benchmark, achieving a mean absolute error of 3 frames for ED and 2 frames for ES detection, which is comparable to existing supervised methods. The approach is also extended to fetal echocardiography.
— via World Pulse Now AI Editorial System

Was this article worth reading? Share it

Recommended apps based on your readingExplore all apps

Ready to build your own newsroom?

Subscribe to unlock a personalised feed, podcasts, newsletters, and notifications tailored to the topics you actually care about