(Sunday, 7/14/2019)
Room: ePoster Forums
Purpose: We previously reported Sensitivity-Regularized (SenR) robust optimization method for Intensity-Modulated Proton Therapy that makes a binary trade-off in robustness and dosimetry based on the choice of target volumes. Optimization based on CTV or PTV alone leads to either insufficient coverage under errors or OAR overexposure, respectively. Instead of these choices, we propose a novel probabilistic margin, to better balance between target coverage and OAR sparing.
Methods: The method was tested on a skull-base-tumor patient. The objective function includes a dose fidelity term and a sensitivity regularization term. The target volume is the CTV with a Probabilistic Margin (PMTV). In the fidelity term, voxels within the CTV are weighted equally. For each voxel in a nominal 3mm margin, a beam-specific weighting following Gaussian distribution is calculated based on its water-equivalent-distance to CTV in the plane perpendicular to the beam. Then the weightings are averaged along beams as the final weighting parameters. To account for maximal 3mm seutp uncertainty, we set the standard deviation to be 1mm. The dosimetry and robustness of PMTV-based SenR plan was compared against CTV-based and PTV-based SenR plans. The PTV is a constant 3mm expansion of CTV.
Results: The PMTV plan better spared the OARs compared to PTV-based plan, reducing [Dmean, Dmax] by [1.3, 3.1] GyRBE on average, while increasing [Dmean, Dmax] by [0.9, 0.7] GyRBE from CTV-based plan. With range uncertainties, PMTV increased the lowest CTV D95% to 97.43%, from 95.37% in CTV-based plan, while 98.21% in PTV-based plan. With setup uncertainties, PMTV increased the lowest D95% to 97.50%, from 96.43% in CTV-based plan, compared to 97.53% in PTV-based plan.
Conclusion: We developed a novel probabilistic weighting scheme on target volume for IMPT robust optimization, which allows a greater flexibility between the dosimetry and robustness than choosing either CTV or PTV in previous SenR optimization.
Funding Support, Disclosures, and Conflict of Interest: This research is supported by NIH Grants Nos. R44CA183390, R43CA183390 and R01CA188300.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
