Scan path optimization with/without clustering for active beam delivery in charged particle therapy.

Purpose: To compare different algorithms to optimize the scanning path in charged particle therapy with quasidiscrete scanning. We implemented a Hybrid Genetic Algorithm with Heuristics (HyGA) and combined it with clustering techniques. The performance was compared to Simulated Annealing (SA) and to commercially available treatment planning system (TPS).

Methods: Performance and clinical implications were assessed using data from 10 patients treated at CNAO (Centro Nazionale di Adroterapia Oncologica). Clinical treatments are performed relying on beam deflection, avoiding irradiation for transitions between adjacent spots larger than 2 cm. A clustering method was implemented with HyGA (HyGA_Cl), which assumes beam deflection during transition between clusters. Clinical performance was determined as the total number of particles delivered during spot transitions and the number of particles wasted due to beam deflection. Results were compared to scan paths obtained with CNAO TPS.

Results: SA and HyGA produced on average shorter paths compared to the currently available TPS. This did not result in a reduction of transit particles, due to the concomitant effect of beam deflection out of the extraction line. HyGA_Cl achieved 2% average reduction in transit particles when compared to CNAO TPS. As a drawback, wasted particles increased, due to more frequent use of beam deflection. Both the SA and HyGA algorithms reduced the number of wasted particles.

Conclusion: SA and HyGA proved to be the most cost-effective methods in reducing wasted particles, with benefits in terms of shorter scan paths. A decrease in transit particles delivered with beam deflection can be achieved using HyGA_Cl.