Problem
Research questions and friction points this paper is trying to address.
Automated Ultrasound Scanning
Generalization
Embodied AI
Vision-Language-Action Model
US-VLA: An Ultrasound Vision-Language-Action Model for Embodied Abdomina
🤖 AI Summary
This study addresses the limited generalizability and reliance on reward functions in existing ultrasound scanning systems by proposing US-VLA, a novel model integrating clinical semantics with real-time feedback. Through an ultrasound perception expert fusion module and sequential action generation, the framework enables autonomous probe manipulation. Additionally, we introduce US-VLA-Data, a real-world dataset comprising 32,000 steps of expert trajectories. Experimental results demonstrate superior performance in abdominal ultrasound tasks, significantly enhancing diagnostic efficiency and stability during automated scanning. These findings validate the generalization capability and practical potential of the Vision-Language-Action paradigm within medical robotics, offering a robust solution to current bottlenecks in autonomous sonography.
📝 Abstract
Artificial intelligence-assisted ultrasound scanning enhances diagnostic reliability and efficiency by providing real-time guidance for standardized image acquisition and reducing operator dependence. However, existing reinforcement learning and learning-assisted ultrasound scanning methods typically rely on carefully designed reward functions or extensive interaction data, which limits their generalization ability and stability across different devices, patient populations, and complex clinical scenarios. To address these challenges, we propose an ultrasound vision-language-action model (US-VLA) for automated ultrasound scanning that explicitly encodes clinical semantic goals and generates sequential probe manipulation actions under real-time ultrasound feedback. In particular, we first design an ultrasound-aware expert fusion module to jointly integrate ultrasound observations with auxiliary contextual information, enabling semantic ultrasound feedback to effectively guide the scanning process. Then, we construct US-VLA-Data, a real-world dataset covering liver and kidney examinations, which includes five clinically defined standard planes and comprises 320 expert scanning trajectories with approximately 80,000 synchronized timesteps. Extensive experiments demonstrate that US-VLA achieves competitive performance in ultrasound probe manipulation tasks, indicating its effectiveness and promising generalization within the evaluated abdominal ultrasound setting. The source code is available at https://github.com/VMVLab/US-VLA.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Vision-Language-Action Model
Ultrasound-Aware Expert Fusion
Semantic Feedback Guidance
US-VLA-Data
Embodied Ultrasound Scanning
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
No related jobs found.
