EchoWorld: Learning Motion-Aware World Models for Echocardiography Probe Guidance (CVPR 2025)

Authors: Yang Yue*, Yulin Wang*, Haojun Jiang, Pan Liu, Shiji Song, Gao Huang#.

*: Equal contribution, #: Corresponding author.

Overview

Echocardiography probe guidance helps sonographers capture standard heart views, but requires significant skill and experience. AI-assisted systems aim to automate this process by interpreting ultrasound images and predicting precise probe movements, making cardiac care more accessible and consistent.

EchoWorld is a motion-aware world modeling framework that learns both heart anatomy and the visual effects of probe motion. EchoWorld employs a pre-training strategy inspired by world modeling principles, where the model predicts masked anatomical regions and simulates the visual outcomes of probe adjustments. Built upon this pre-trained model, we introduce a motion-aware attention mechanism in the fine-tuning stage that effectively integrates historical visual-motion data, enabling precise and adaptive probe guidance. Trained on more than one million ultrasound images from over 200 routine scans, our model outperforms existing methods in probe guidance accuracy.

Acknowledgement

This code is developed on the top of MAE and I-JEPA

Contact

If you have any questions or concerns, please send email to yueyang22@mails.tsinghua.edu.cn

Due to data privacy and institutional restrictions, the dataset used in this study cannot be publicly released.