Toward Safe Autonomous Robotic Endovascular Interventions using World Models

arXiv cs.RO / 4/23/2026

📰 NewsDeveloper Stack & InfrastructureModels & Research

共有:

Key Points

The paper proposes a world-model-based reinforcement learning framework for autonomous endovascular navigation for mechanical thrombectomy, aiming to improve robustness to varied vascular anatomies and long navigation horizons.
It evaluates a TD-MPC2 agent on multiple simulation tasks across hold-out patient-specific vasculatures and benchmarks it against the Soft Actor-Critic (SAC) baseline.
In simulation, TD-MPC2 achieves a higher mean success rate than SAC (58% vs. 36%, p < 0.001) while keeping tip contact forces (0.15 N) well below a 1.5 N vessel rupture threshold.
In vitro tests with patient-specific vascular phantoms under fluoroscopic guidance show comparable success rates between TD-MPC2 (68%) and SAC (60%), but TD-MPC2 produces superior path ratios (p = 0.017) with longer procedure times (p < 0.001).
The authors claim this is the first demonstration of autonomous MT navigation validated across both hold-out simulation data and fluoroscopy-guided in vitro experiments, supporting the potential for safer, more generalizable AI-assisted interventions.

Abstract

Autonomous mechanical thrombectomy (MT) presents substantial challenges due to highly variable vascular geometries and the requirements for accurate, real-time control. While reinforcement learning (RL) has emerged as a promising paradigm for the automation of endovascular navigation, existing approaches often show limited robustness when faced with diverse patient anatomies or extended navigation horizons. In this work, we investigate a world-model-based framework for autonomous endovascular navigation built on TD-MPC2, a model-based RL method that integrates planning and learned dynamics. We evaluate a TD-MPC2 agent trained on multiple navigation tasks across hold out patient-specific vasculatures and benchmark its performance against the state-of-the-art Soft Actor-Critic (SAC) algorithm agent. Both approaches are further validated in vitro using patient-specific vascular phantoms under fluoroscopic guidance. In simulation, TD-MPC2 demonstrates a significantly higher mean success rate than SAC (58% vs. 36%, p < 0.001), and mean tip contact forces of 0.15 N, well below the proposed 1.5 N vessel rupture threshold. Mean success rates for TD-MPC2 (68%) were comparable to SAC (60%) in vitro, but TD-MPC2 achieved superior path ratios (p = 0.017) at the cost of longer procedure times (p < 0.001). Together, these results provide the first demonstration of autonomous MT navigation validated across both hold out in silico data and fluoroscopy-guided in vitro experiments, highlighting the promise of world models for safe and generalizable AI-assisted endovascular interventions.