A Nearly-Exact Beam Angle Optimization Algorithm Via Angle Generation Method
H Shen1,2, G Zhang2, Y Lin1, R Chen1, Y Long2, H Gao1*, (1) University of Kansas Medical Center, Kansas City, KS (2) Shanghai Jiao Tong University, Shanghai, China
Presentations
MO-H345-IePD-F4-6 (Monday, 7/11/2022) 3:45 PM - 4:15 PM [Eastern Time (GMT-4)]
Exhibit Hall | Forum 4
Purpose: In treatment planning, beam angle optimization (BAO) refers to the selection of a subset from all available angles that provides the best plan quality, for a given number of beam angles. BAO is a NP-hard combinatorial problem. Although exhaustive search (ES) can solve BAO by exploring all possible combinations, ES is very time-consuming and practically infeasible. To the best of our knowledge, (1) no optimization method has been demonstrated to solve BAO exactly, and (2) no study has attempted to validate an optimization method by benchmarking with the optimal solution from ES, both of which will be addressed by this work.
Methods: This work considers BAO for spot-scanning proton therapy, i.e., the selection of 2-4 beam angles for IMPT. First, ES is solved as the validation benchmark. Second, a new BAO algorithm, so-called angle generation (AG) method, is proposed, which will be demonstrated to provide nearly-optimal solutions for BAO in benchmark with ES solutions. AG iteratively updates the set of candidate angles via group-sparsity (GS) regularization, until the plan-quality objective does not decrease further.
Results: While BAO can be solved by GS directly, AG was better than GS since (1) AG achieved nearly-optimal solutions for BAO and (2) N-optimal angular set does not have to be a subset of N+1-optimal angular set for AG. Specifically, AG and GS were evaluated for 2-angle brain, 3-angle lung, and 4-angle brain cases, in comparison to ES that solved and ranked optimized plan objectives from 276, 2024, 10626 IMPT solutions respectively: AG had nearly-optimal solutions, i.e., with the ranking (1/276, 1/2024, 4/10626), while the GS solution had the ranking (42/276, 279/2024, 4328/10626).
Conclusion: A nearly-exact algorithm for solving BAO via AG is proposed and validated against ES.
Funding Support, Disclosures, and Conflict of Interest: This research is partially supported by the NIH Grant No. R37CA250921 and a KUCC physicist-scientist recruiting grant.
Keywords
Optimization, Treatment Planning, Inverse Planning
Taxonomy
TH- External Beam- Particle/high LET therapy: Proton therapy – dose optimization
Contact Email