Program Information
Comparability of Three Output Prediction Models for a Compact Passively Scattered Proton Therapy System
S Ferguson1*, Y Chen1 , C Ferreira1 , M Islam1 , V Keeling2 , A Lau1 , S Ahmad1 , H Jin1 , (1) Oklahoma Univ. Health Science Ctr., Oklahoma City, Oklahoma, (2) CARTI, Inc., Little Rock, AR
Presentations
SA-B-BRD-3 (Saturday, March 5, 2016) 10:30 AM - 12:30 PM Room: Grand Ballroom D
Purpose: To investigate comparability of three output (cGy/MU) prediction models for a compact passively scattered proton therapy system.
Methods: Two published output prediction models were commissioned for our Mevion S250 proton therapy system: model (1) is a correction-based model (Sahoo et al, Med Phys, 35, 5088-5097, 2008) and model (2) is a semi-analytical model as a function of r=(R-M)/M (Kooy et al, PMB, 50, 5487-5456, 2005). The function r was modified to ((R’-0.31)-0.81xM’)/(0.81xM’) to convert the theoretical range R (distal 100% dose) and modulation M (distal 100% dose-to-proximal 100% dose) to the Mevion definition (R’: distal 90% dose, M’: distal 90% dose-to-proximal 95% dose). In addition, another quartic polynomial model (3) was newly developed based on the r. The outputs of 112 combinations of range R and modulation M covering the 24 options were measured. Each model’s predicted output was compared to the measured output. In addition, outputs predicted by each model were also compared against each other using the Student’s t-test.
Results: For the total data set, the percent differences between predicted (P) using the three different models and measured (M) outputs ((P-M)/Mx100%) were within ±3%. The average differences (±SD) were -0.02±1.06%, 0.03±1.16%, and -0.17±1.27% for model (1), (2), and (3), respectively. The p-values of the t-test were 0.5869 (model (1) vs. (2)), 0.0725 (model (1) vs. (3)), and 0.0294 (model (2) vs. (3)). In general, the difference is greater as r becomes smaller for all the models.
Conclusion: For all the options, all three models have clinically acceptable prediction. The difference between model (1) and model (2) (or model (3)) is statistically insignificant (p>0.07). Care should be taken when a small r of R and M combination is used. It is concluded that the models can comparably be used for the compact passively scattered proton therapy system.
Contact Email: