Study on Path Planning Strategy for Vascular Interventional Surgery Based on Visual Information

Vascular interventional robots have significantly advanced traditional surgical procedures by integrating autonomous navigation, real-time imaging, and AI-driven algorithms, thereby enhancing both precision and efficiency. Real-time guidewire navigation is critical for enabling autonomous operation. However, current research in this area still faces challenges such as the lack of accurate guidewire localization and the inability to execute real-time path planning commands. This study utilizes surgical visual information and proposes deep learning-based solutions to address these issues. Specifically, YOLOv11 is employed for real-time guidewire tracking, a custom coordinate localization algorithm is developed to identify the guidewire tip, and U-Net is applied for vessel segmentation to support path planning. Furthermore, a keypoint annotation mechanism is introduced to precisely characterize trajectory variations, thereby improving the accuracy and reliability of robotic navigation.