Guidewire Tracking based on Visual Algorithm for Endovascular Interventional Robotic System

During the Minimally invasive vascular intervention surgery, deformable guidewire tracking is still a challenging task due to background clutter of the image and the complex motion of the target. However, existing researches about guidewire tracking for robot-assisted endovascular catheterization system are still limited. In this paper, scale-adaptive mean-shift method is adopted in endovascular interventional robotic system to detect the position of the guidewire tip. To evaluate the performance of this algorithm in guidewire tracking, two interventional experiment using the rigid model of cerebral vascular were designed. The experimental results show that the ratio of frames with the center location error less than 5 pixels is 97.6% in these two tasks, and the average processing speed for each frame is 1.24ms. The result shows that this algorithm with high precision and real-time has the potential to apply in endovascular interventional robotic system.