Sample-Efficient Learning with Online Expert Correction for Autonomous Catheter Steering in Endovascular Bifurcation Navigation
Hao Wang, Tianliang Yao, Bo LU, Zhiqiang Pei, dong liu, Lei Ma, Peng Qi
AI summary
Problem
Conventional reinforcement learning for autonomous catheter navigation suffers from sparse rewards, poor sample efficiency, and limited adaptability to real-time intraoperative changes, hindering reliable bifurcation traversal.
Approach
The authors propose a hybrid SAC-GAIL framework that integrates real-time image-based pose estimation, expert-in-the-loop correction, and fuzzy control to dynamically guide policy learning and stabilize navigation at vascular junctions.
Key results
- Converges in 123 training episodes, a 25.9% reduction over baseline SAC
- Reduces average positional error to 83.8% of the baseline algorithm
- Successfully maps expert demonstrations to real-time fuzzy control rules for bifurcation navigation
- Validated on a physical robotic platform using a transparent vascular phantom
Why it matters
Provides a clinically viable pathway for safe, sample-efficient autonomous navigation in minimally invasive endovascular procedures, reducing operator fatigue and radiation exposure.
Abstract
Robot-assisted endovascular intervention offers a safe and effective solution for remote catheter manipulation, reducing radiation exposure while enabling precise naviga- tion. Reinforcement learning (RL) has recently emerged as a promising approach for autonomous catheter steering; however, conventional methods suffer from sparse reward design and reliance on static vascular models, limiting their sample effi- ciency and generalization to intraoperative variations. To over- come these challenges, this paper introduces a sample-efficient RL framework with online expert correction for autonomous catheter steering in endovascular bifurcation navigation. The proposed framework integrates three key components: (1) A segmentation-based pose estimation module for accurate real- time state feedback, (2) A fuzzy controller for bifurcation- aware orientation adjustment, and (3) A structured reward generator incorporating expert priors to guide policy learn- ing. By leveraging online expert correction, the framework reduces exploration inefficiency and enhances policy robustness in complex vascular structures. Experimental validation on a robotic platform using a transparent vascular phantom demonstrates that the proposed approach achieves convergence in 123 training episodes—a 25.9% reduction compared to the baseline Soft Actor-Critic (SAC) algorithm—while reducing average positional error to 83.8% of the baseline. These results indicate that combining sample-efficient RL with online expert correction enables reliable and accurate catheter steering, particularly in anatomically challenging bifurcation scenarios critical for endovascular navigation.