3109 - A Heuristic Path Optimization Framework for Proton Arc Delivery Using a Translating Checkerboard Range Shifter

Purpose/Objective(s):

 Clinical implementation of proton dynamic arc therapy is limited by prolonged energy layer switching times. A novel dynamic range shifter, consisting of a repeating pattern of four polycarbonate thickness layers, has been designed to convert a single pre-nozzle beam energy into four discrete post-nozzle energies. To achieve this functionality, the range shifter must be translated to different positions and synchronized with spot delivery. We hypothesize that a heuristic path optimization algorithm can be used to optimize and minimize range shifter motion, thereby enabling more efficient delivery of proton dynamic arc therapy.

Materials/Methods:

 A heuristic path-finding optimization algorithm was developed to determine a translational motion sequence that minimizes range shifter movement while ensuring all proton spots intercept their assigned thickness layer. The framework incorporates Greedy Best-First Search, A*-inspired heuristics, Beam Search, Hill-Climbing with Random Restarts, and a Greedy Multitarget Capture strategy. A simulated patient-specific proton arc plan, consisting of 360° gantry rotation with 180 control points, was created. Four proton beam energies were assigned at each gantry angle, with the water equivalent thickness (WET) differences matching the design of the range shifter design. The plan was optimized, and all the goals and constraints were achieved. The final plan consisted of 1,440 randomly distributed spots and was used for the validation.

Results:

 The algorithm captured all 1,440 proton spots across 180 gantry angles, with no gantry angle containing uncaptured spots. A total of 1,676 translational steps were required, corresponding to a cumulative range shifter travel distance of 3,352 cm. Assuming 50 ms per translational step, the total range shifter movement time for the full arc was 83.8 seconds. Mean spot delivery throughput was 34.37 spots per second. The full-arc optimization was completed in 9.64 seconds.

Conclusion:

 Dynamic synchronization of a translating checkerboard range shifter with proton spot delivery is feasible and computationally efficient. By replacing energy switching with controlled mechanical modulation, this framework provides a practical pathway toward continuous and dynamic proton arc therapy and improved delivery efficiency.