Conceitos essenciais
The authors propose an Image-guided Autonomous Guidewire Navigation (IAGN) method for endovascular interventions, utilizing reinforcement learning and path planning algorithms to achieve successful navigation.
Resumo
The content discusses the development of an autonomous guidewire navigation system for endovascular interventions. The proposed method integrates image guidance, reinforcement learning, and path planning algorithms to improve safety and efficiency in surgical procedures. Experiments on a simulation platform demonstrated a 100% success rate in navigating specific arteries with reduced movement distances.
Key points include:
- Introduction to the challenges of endovascular interventions and the need for autonomous systems.
- Description of the IAGN method utilizing BDA-star path planning algorithm and real-time image observations.
- Implementation details of the autonomous guidewire navigation platform and trajectory planning.
- Reinforcement learning approach for autonomous navigation with explicit observations and reward functions.
- Results showing successful navigation rates, reduced movement distances, and trajectory convergence towards vessel centers.
Estatísticas
Experiments conducted on the aortic simulation IAGN platform demonstrated a 100% guidewire navigation success rate.
The maximum values for actions S and R are set to 20 mm and 90°.
In DBA-star, the weight of the boundary distance term is set to 2.