MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Secondary Optimization of VMAT Plans Using Monte Carlo Beamlets

J Mathews1,2*, S French3 , S Bhagroo1,2 , D Nazareth1,2 , (1) Roswell Park Comprehensive Cancer Center, Buffalo, NY, (2) University at Buffalo, SUNY, Buffalo, NY, (3) Piedmont Healthcare, Fayetteville, GA

Presentations

(Wednesday, 7/17/2019) 1:45 PM - 3:45 PM

Room: 301

Purpose: To develop a secondary optimization process for improving VMAT plans based on beamlets produced using a Monte Carlo (MC) dose calculation algorithm.

Methods: For a Varian Trilogy VMAT plan, the treatment parameter information (e.g. MLC leaf positions, dose rate, gantry angles) is extracted from DICOM files exported from Eclipse. These parameters are used to generate input files for use in an open source MC dose calculation algorithm (EGSnrc user codes DOSXYZnrc and BEAMnrc). The full plan is simulated in two ways: in its entirety to obtain a single (base) dose matrix, and separately as the sum of individual control points. A beamlet is calculated by modifying a single leaf’s position in a control point by moving it either in or out by 0.5 cm (projected to isocenter) and then simulating that modified control point. The difference in dose matrices between a modified and unmodified control point is considered as the dose distribution corresponding to a single beamlet. In order to reduce extraneous calculations, an algorithm was written to exclude beamlets that do not contribute dose to the PTV or would violate linac deliverability constraints. In addition, the beamlet calculation times were reduced with parallelization. A greedy search algorithm is employed, which utilizes a dose-volume-based objective function, to identify beamlets which improve the VMAT plan. This method was evaluated on a prostate plan using clinical DVH objectives.

Results: The greedy search algorithm was able to improve the objective score by 16.2%. In addition, the modified plan’s DVH plots for the PTV and OARs indicated favorable dose distributions, particularly to the bladder, rectum, and right femoral head.

Conclusion: We have developed a secondary optimization method for VMAT plans which applies fine changes to original TPS MLC leaf positions. Based on MC-calculated beamlets, the method can produce clinically-viable improved VMAT plans.

Keywords

Monte Carlo, Optimization

Taxonomy

Not Applicable / None Entered.

Contact Email