Reinforcement Learning for Safe Autonomous Two Device Navigation of Cerebral Vessels in Mechanical Thrombectomy

2025-03-31

AI Summary

Existing RL methods for mechanical thrombectomy are limited to carotid navigation, exhibiting poor generalizability and insufficient safety guarantees. This paper proposes a safety-constrained, dual-instrument (microcatheter/microguidewire) cooperative navigation RL framework. We first embed the mechanical constraint on guidewire tip force (<1.5 N rupture threshold) into the Soft Actor-Critic algorithm and integrate inverse RL to enable cross-patient cerebrovascular generalization. Leveraging the SOFA platform, we construct 12 patient-specific vascular models and design a physics-driven, force-informed reward function. Experiments demonstrate a 96% navigation success rate in unseen vessel simulations, with an average completion time of 7.0 seconds and a mean guidewire tip force of only 0.24 N—well below the clinical safety threshold. The framework thus achieves superior efficiency, robustness, and safety simultaneously.

Technology Category

• Natural Language Processing: Safety and Robustness
• Planning, Routing, and Scheduling: Model-Based Reasoning
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Purpose: Autonomous systems in mechanical thrombectomy (MT) hold promise for reducing procedure times, minimizing radiation exposure, and enhancing patient safety. However, current reinforcement learning (RL) methods only reach the carotid arteries, are not generalizable to other patient vasculatures, and do not consider safety. We propose a safe dual-device RL algorithm that can navigate beyond the carotid arteries to cerebral vessels. Methods: We used the Simulation Open Framework Architecture to represent the intricacies of cerebral vessels, and a modified Soft Actor-Critic RL algorithm to learn, for the first time, the navigation of micro-catheters and micro-guidewires. We incorporate patient safety metrics into our reward function by integrating guidewire tip forces. Inverse RL is used with demonstrator data on 12 patient-specific vascular cases. Results: Our simulation demonstrates successful autonomous navigation within unseen cerebral vessels, achieving a 96% success rate, 7.0s procedure time, and 0.24 N mean forces, well below the proposed 1.5 N vessel rupture threshold. Conclusion: To the best of our knowledge, our proposed autonomous system for MT two-device navigation reaches cerebral vessels, considers safety, and is generalizable to unseen patient-specific cases for the first time. We envisage future work will extend the validation to vasculatures of different complexity and on in vitro models. While our contributions pave the way towards deploying agents in clinical settings, safety and trustworthiness will be crucial elements to consider when proposing new methodology.

Problem

• Extends RL navigation beyond carotid arteries to cerebral vessels
• Ensures safety by integrating guidewire tip force metrics
• Generalizes to unseen patient-specific vascular anatomies

Innovation

• Dual-device RL algorithm for cerebral navigation
• Safety metrics integrated into reward function
• Inverse RL with patient-specific demonstrator data