Autonomous Catheterization with Open-source Simulator and Expert Trajectory

TL;DR

This work introduces CathSim, the first open-source, real-time MuJoCo-based simulator for endovascular interventions, designed to accelerate autonomous catheterization research by providing anatomically accurate phantoms, realistic force feedback, and AR/VR compatibility. Building on CathSim, the authors develop a multimodal expert navigation network (ENN) that leverages image, segmentation, and kinematic/force cues to learn navigation policies via SAC, and demonstrate downstream benefits in imitation learning and force prediction. Validation against a real robot shows similar force distributions, while extensive experiments reveal that ENN with multi-modal inputs improves efficiency (shorter paths and episodes) and accuracy, though humans exhibit higher safety in some scenarios. The work highlights the potential of open, ML-friendly simulators to enable rapid prototyping, sim-to-real transfer, and broader exploration of autonomous endovascular strategies, while outlining clear paths to address soft-tissue dynamics and real-world applicability.

Abstract

Endovascular robots have been actively developed in both academia and industry. However, progress toward autonomous catheterization is often hampered by the widespread use of closed-source simulators and physical phantoms. Additionally, the acquisition of large-scale datasets for training machine learning algorithms with endovascular robots is usually infeasible due to expensive medical procedures. In this chapter, we introduce CathSim, the first open-source simulator for endovascular intervention to address these limitations. CathSim emphasizes real-time performance to enable rapid development and testing of learning algorithms. We validate CathSim against the real robot and show that our simulator can successfully mimic the behavior of the real robot. Based on CathSim, we develop a multimodal expert navigation network and demonstrate its effectiveness in downstream endovascular navigation tasks. The intensive experimental results suggest that CathSim has the potential to significantly accelerate research in the autonomous catheterization field. Our project is publicly available at https://github.com/airvlab/cathsim.

Figures (12)

• Figure 1: An overview of CathSim.
• Figure 2: The design architecture of CathSim.
• Figure 3: The visualization of the aorta, guidewire and blood in our simulator.
• Figure 4: Aortic Models.
• Figure 5: Schematic representation of the CathBot's follower mechanism kundrat2021mr (a) alongside a visualization of our simulated model (b).