Problem
Research questions and friction points this paper is trying to address.
keyframe detection
multi-view echocardiography
end-systole
end-diastole
motion modeling
FrameONE: Hierarchical Motion Modeling for Universal Multi-View Echocardiographic Keyframe Detection
🤖 AI Summary
Existing methods for keyframe detection in multi-view echocardiography suffer from limited generalizability, often relying on view-specific modeling, auxiliary annotations, or complex visual architectures. This work proposes FrameONE—the first unified end-to-end framework that operates without auxiliary labels—by introducing hierarchical motion modeling to disentangle view-common and view-specific characteristics. Within each view, multi-task learning emphasizes cardiac motion features, while across views, the model learns a view-invariant representation of universal motion dynamics. Evaluated on a large-scale dataset comprising 25,872 video clips spanning four standard echocardiographic views, FrameONE achieves state-of-the-art accuracy and demonstrates substantially improved cross-view generalization performance.
📝 Abstract
Accurate detection of end-systole (ES) and end-diastole (ED) frames is fundamental to echocardiographic assessment. Existing methods are typically developed in a view-specific manner, depend on auxiliary annotations or intensive visual modeling, which limits their generalizability. In multi-view modeling, keyframe detection is driven by shared cardiac motion, yet large appearance differences and motion patterns make unified modeling challenging. To address these issues, we propose FrameONE, a unified end-to-end framework for multi-view echocardiographic keyframe detection. FrameONE introduces a Hierarchical Motion Modeling strategy: an intra-view multi-task learning reduces appearance bias and promotes motion-focused representations within each view; an inter-view general motion learning module further separates view-agnostic dynamics from view-specific patterns, enabling shared yet flexible motion representation learning across views. Extensive experiments on 25,872 videos spanning four standard views demonstrate that FrameONE achieves state-of-the-art keyframe detection accuracy with strong cross-view generalization. Code is available at https://github.com/szuboy/FrameONE.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Hierarchical Motion Modeling
Multi-View Echocardiography
Keyframe Detection
View-Agnostic Representation
End-to-End Framework
🔎 Similar Papers
No similar papers found.
