Intraoperative navigation methods, systems, devices, and media for vascular interventions
By performing elastic collision training and reinforcement learning on a three-dimensional vascular model, the navigation path of interventional devices is determined, and two-dimensional synthetic images are generated by combining camera pose parameters. This solves the problem of difficult operation of interventional devices in the elastic contact environment of blood vessels and enables smooth navigation of interventional devices.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN INST OF ADVANCED BIOMEDICAL ROBOT CO LTD
- Filing Date
- 2026-05-15
- Publication Date
- 2026-06-16
AI Technical Summary
In existing vascular interventional procedures, interventional instruments are difficult to operate along the planned surgical path in the elastic contact environment of blood vessels, leading to operational difficulties.
By performing elastic collision training in a three-dimensional vascular model, a reinforcement learning algorithm is used to record reward values. The trajectory of the interventional device tip corresponding to the highest reward value is determined as the navigation path. Combined with camera pose parameters, fluoroscopic projection is performed to generate a two-dimensional synthetic image for intraoperative navigation.
In a vascular elastic contact environment, interventional devices can smoothly perform surgery along the navigation path, avoiding the problem of ignoring elastic collisions in traditional path planning and providing a more reasonable operating trajectory.
Smart Images

Figure 1 
Figure 2