Event
Proposal of a Framework for Target-Plane-Guided Robotic Neck Ultrasound
1 reports · 1 sources
Siqi Fan et al. propose a two-stage framework for target-plane probe guidance in robotic neck ultrasound: first, a latent-space conditional diffusion world model predicts future ultrasound observations based on recent context frames, probe motion, and temporal offsets; then, a target-conditioned temporal Transformer predicts ordered probe actions and is fine-tuned using rewards from the frozen world model. On a self-collected dataset, experiments show that the world model preserves action-relevant anatomical structures during target-oriented scanning; in real closed-loop experiments, the framework achieved a 70.0% success rate for carotid artery guidance and 65.0% for thyroid guidance.
Report timeline
- Action-Conditioned World Model for Robotic Neck Ultrasound Target Plane Guidance
This study proposes a local action-conditioned world model framework for probe guidance toward target planes in robotic ultrasound, using a two-stage model-based learning pipeline: first, a latent conditional diffusion world model predicts future ultrasound observations from recent context frames, probe motion, and temporal offset; second, a goal-conditioned temporal Transformer predicts ordered probe actions and is fine-tuned using rewards from the frozen world model.
arXiv cs.RO ·