Towards an optimization of catheter guidance in vascular surgery: A comparative analysis of the contribution of reinforcement learning

Abstract

Background: Precision in catheter guidance is essential for the success of vascular surgeries, yet current methods often need more accuracy due to the complex anatomy and dynamics of blood vessels. Methods: This study evaluates the efficacy of advanced reinforcement learning (RL) techniques to enhance catheter navigation. We compare different RL approaches within simulated vascular environments, focusing on their success rates, operational efficiency, and adaptability to varied clinical scenarios. Results: Advanced reinforcement learning techniques display exceptional performance, yielding high success rates and improved precision in catheter guidance. Integrating specific enhancements has notably increased learning speeds and strengthened operational robustness. Conclusion: The study indicates that reinforcement learning could significantly improve the precision and safety of catheter navigation in vascular surgery. By adopting these techniques, medical practices could see more accurate and less invasive procedures, enhancing patient outcomes. Future research should aim to refine these algorithms for wider clinical use and integration.

Keywords

• Artificial intelligence
• Reinforcement learning
• Vascular surgery
• Catheter guidance
• Simulation