Author: Wenhua Cao, Xianjin Dai, PhD, Hadis Moazami Goudarzi, Gino Lim, Miaolan Xie, Lei Xing, Lewei Zhao 👨🔬
Affiliation: University of Chicago Booth School of Business, Department of Radiation Oncology, Stanford University, Department of Industrial Engineering, University of Houston, Department of Radiation Physics, The University of Texas MD Anderson Cancer Center 🌍
Purpose: Proton Arc Therapy (PAT) delivers a continuous dose of radiation during gantry rotation. 4π is a non-coplanar technique used for advanced proton therapy delivery. This work proposes a hybrid method angular sampling boosted PAT (ASB-PAT) of integrating these two techniques with robust optimization.
Methods: This novel algorithm introduces a 4π beam angle selection adding to PAT before the actual robust spot weight optimization. The algorithm comprises two steps: pre-selection of five non-coplanar beam angles using a semi-stochastic gradient descent approach and robust optimization of spot weights to ensure dose distribution integrity under setup and range uncertainties. Robust optimization is implemented using a minimax composite objective-wise robust optimization approach, which optimizes the objective function under the worst-case scenarios of positioning inaccuracies. Disease sets, including a skull-base cases that organs at risk brainstem and spinal cord have overlap with clinic target volume (CTV), were analyzed to evaluate the dose-volume histogram (DVH). The comparison involved ASB-PAT and single-arc PAT plans, with robustness settings accounting for ±3 mm setup and ±3% range errors.
Results: For a representative skull-base case, the hybrid method ASB-PAT achieved superior dose conformity and robustness compared to single-arc PAT. The robust ASB-PAT plan achieved the highest dose coverage with CTV D98, D95, and D90 values of 64.5 Gy, 68.8 Gy, and 71.0 Gy, respectively, demonstrating superior target coverage compared to other plans. This is an indication of improved dose homogeneity and effective tumor targeting, especially under setup uncertainties and patient movements. Also, robust ASB-PAT provided better control over the brainstem dose at 61.47 Gy compared to its non-robust counterpart 63.16 Gy.
Conclusion: This hybrid 4π-PAT method achieves superior dose conformity and robustness for complex geometries, demonstrating its potential as a clinically viable solution.