(Wednesday, 7/17/2019) 9:30 AM - 10:00 AM
Room: Exhibit Hall | Forum 4
Purpose: The beam angle flexibility introduced by SBRT increases the complexity of determining good beam profiles for high quality treatment plans. Also, the beamlet based-dose calculation is time-consuming and has discrepancy comparing to actual aperture-based dose. We propose a multi-objective direct aperture optimization framework to generate high quality SBRT treatment plans and couple with warm-start beam angle and fluence map optimization to improve plan quality.
Methods: Multiple objectives include PTV conformity, total OAR dose, total MU and total dose-fall off. Preemptive programming is employed where each objective is solved for the direct aperture optimization. The mixed integer programming (MIP) model determines the optimal gantry angle during the preemptive steps. Constraints include PTV coverage and conformity and OAR DVH, we utilize the geometric information of the PTV and OAR to assist in the column generation process and apply a fast heuristic to generate improving apertures while by passing costly beamlet-based dose generation.
Results: CT and unit dose calculation and clinical guidelines are used for 5 prostate cancer patients, 5 lung cancer patients and 5 intracranial cancer patients. In all cases our method produces high quality deliverable plans satisfying all clinical requirements. The quality outperforms single-objective and the clinical plans. We achieve a 1.12 conformity for all prostate cases, 1.09 for all lung cases, and < 1.2 for all intracranial cases. The column generation for each objective converges mostly within 7 iterations, showing that our heuristic algorithm is high efficient for generating aperture.
Conclusion: This work developed an efficient multi-objective direct aperture optimization model that optimizes each treatment objective preemptively based on their clinical priority. This is the first model that optimize beam angle and aperture weight simultaneously in a direct aperture optimization framework.
Keywords
Optimization, Treatment Planning
Taxonomy
Not Applicable / None Entered.
Contact Email
