Automatic Pedicle Screw Placement Based On Simulated Annealing Algorithm

Minimally invasive spine surgery (MISS) faces significant challenges due to limited visibility of entry points and spinal structures, necessitating precise preoperative path planning to ensure optimal surgical outcomes. Traditional methods rely on subjective measurements of vertebral parameters from CT images, which are highly dependent on surgeon experience and lack a comprehensive three-dimensional (3D) representation of the pedicle’s spatial structure. This often results in inaccuracies in angle positioning and path planning. To address these limitations, this paper presents an automated preoperative path planning and evaluation method for pedicle screw placement in thoracolumbar spine surgery. The method leverages identified and segmented pedicle regions to generate an initial path for screw placement. Next, the simulated annealing algorithm is used to optimize the path, maximizing the largest non-breakthrough diameter as the objective. The final path parameters are refined by considering bilateral path collision avoidance and medical screw specifications, ensuring a feasible and clinically viable placement strategy. Furthermore, the proposed method extends the Gertzbein-Robbins (GB) classification standard, commonly used for postoperative evaluation, to quantitatively assess the results of preoperative path planning. Experimental results demonstrate that all planned screw placements achieve an A-grade classification, validating the effectiveness of the proposed method in enhancing surgical precision and safety.