(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: quantify the dosimetric impact of transitioning to a Monte Carlo dose algorithm for a proton pencil beam scanning beam delivery system in the treatment of CNS patients.
Methods: IRB-approved retrospective study was conducted on 30 CNS patient treatment plans. Patients were treated using an IBA pencil beam with range shifters (either 4 or 7.5 cm). Previously delivered treatment plans were anonymized and copied to a research database. The plans were re-calculated with the latest beam models in the current planning system (Raystation 9a) using the pencil beam algorithm (PBA). These plans were then copied and re-calculated using the latest Monte Carlo (MC) algorithm in the same planning system. A dosimetric comparison between the two planning algorithms was done on a patient-by-patient basis. The dose covering 99, 98, 95, 50, 2 and 1% of the GTV, CTV and PTV was evaluated as well as the average dose to each of those structures.
Results: systematic dose difference was seen for all cases. The trend was independent of target (GTV, CTV or PTV), and coverage level (1% to 99%). Average dose to the CTV had the most consistent ratio, with MC calculations 1.5 to 4% lower than PBA. Dose differences were independent of tumor size which ranged from GTV’s of less than 5cc to greater than 90cc. Peripheral tumors (reaching the skull) showed an average dose difference of 2.5% while centrally located tumors had an average dose difference of 1.5%.
Conclusion: accuracy of PBA dose calculations is affected by the in-room range shifter. Thicker range shifters (for peripheral tumors) create a larger effect than thinner ones for deeper tumors. Even for relatively homogeneous CNS targets switching from PBA to MC for plan optimization will result in systematically higher doses than what most clinical experience is based on.
Keywords
Protons, Monte Carlo, Treatment Planning
Taxonomy
TH- External Beam- Particle/high LET therapy: Proton therapy – computational dosimetry-Monte Carlo
Contact Email
