322 - An Omni-Optimizer of Range Modulation, Scanning Path, Beam Current, and Spot Intensity for Proton FLASH
Presenter(s)
H. Gao1, Y. N. Zhu1, J. Han1, J. Ma2, J. Liu3, A. Wang1,4, W. Li1, J. Setianegara1, M. Tang2, Y. Lin1, and R. C. Chen1; 1Department of Radiation Oncology, University of Kansas Medical Center, Kansas City, KS, 2Institute of Natural Sciences and School of Mathematics, Shanghai Jiao Tong University, Shanghai, China, 3LSEC, Institute of Computational Mathematics and Scientific/Engineering Computing, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing, China, 4Department of Biomedical Engineering, Huazhong University of Science and Technology, Wuhan, China
Purpose/Objective(s): Proton therapy enables FLASH delivery with ultra-high dose rate (UHDR) at depth. Clinical implementation of proton FLASH relies on a patient-specific range modulator (PSRM) to maintain multi-energy dose-shaping capabilities while achieving UHDR through single-high-energy delivery. This work introduces a first-of-its-kind omni-optimizer for proton FLASH, optimizing dose distributions while preserving UHDR in patients by jointly optimizing range modulation, scanning path, beam current, and spot intensity.
Materials/Methods: Conventional proton FLASH planning optimizes a multi-energy IMPT plan via spot-intensity-only optimization, then converts it into a single-energy proton FLASH plan using PSRM-only optimization—overlooking the inherent interdependence between spot intensity and PSRM. In contrast, our proposed joint spot-intensity and PSRM optimization (JSPO) integrates these steps into a single process, analytically modeling one-to-many spot-pin correspondence. Additionally, the spot scanning path is optimized via the genetic algorithm to maximize dose rates, as scanning path optimization (SPO) can significantly impact the achievable UHDR; the beam current is optimized by solving the minimum-monitor-unit problem via stochastic splitting optimization (SSP), to jointly optimize dose and dose rate distributions. The overall optimization framework employs iterative convex relaxation and alternating optimization.
Results: Each component—JSPO, SPO, and SSP—was independently validated. JSPO outperformed the conventional approach in terms of improved plan quality and improved robustness to fewer spots and misalignments between spots and pins. For example, JSPO had an improvement of optimization objective value from 5.28 to 4.23, maximum target dose from 133.5% to 109.8% and conformity index (CI) from 0.804 to 0.838 for a prostate case. Experimental results also showed that the measured integrated depth dose (IDD) curve aligned perfectly with the planned IDD, validating JSPO's effectiveness. SPO substantially enhanced the FLASH-dose-rate coverage. For example, in a head-and-neck case, SPO elevated FLASH-dose-rate coverage from 68% to 86%, and at the same time decreased the objective function value from 3.79 to 3.21, e.g., reduced the brainstem volume receiving =18Gy from 0.43cc to 0.35cc. SSP was shown to improve both dose and dose rate distribution. For example, for a lung case, SSP improved the CI from 0.68 to 0.76, and improved FLASH-dose-rate coverage from 92% to 100%.
Conclusion: A novel omni-optimizer of range modulation, scanning path, beam current, and spot intensity is developed for proton FLASH, with demonstrated improvements in dose and dose rate distributions.