Automated resection planning for bone tumor surgery.

Planning for bone tumor resection surgery is a technically demanding and time-consuming task, reliant on manual positioning of cutting planes (CPs). This work describes an automated approach for generating bone tumor resection plans, where the volume of healthy bone collaterally resected with the tumor is minimized through optimized placement of CPs. Particle swarm optimization calculates the optimal position and orientation of the CPs by introducing a single new CP to an existing resection, then optimizing all CPs to find the global minima. The bone bounded by all CPs is collaterally resected with the tumor. The approach was compared to manual resection plans from an experienced surgeon for 20 tumor cases. It was found that a greater number of CPs reduce the collaterally resected healthy bone, with diminishing returns on this improvement after five CPs. The algorithm-generated resection plan with equivalent number of CPs resulted in a statistically significant improvement over manual plans (paired t-test, p < 0.001). The described approach has potential to improve patient outcomes by reducing loss of healthy bone in tumor surgery while offering a surgeon multiple resection plan options.