Effect of Phantom Geometry on Measured Prompt Gamma Distribution for 150 MeV Proton Beam

Accurately determining the distal dose edge, where the dose decreases to 80% of its peak dose directly after the Bragg peak in the patient, is very important not only for successful treatment, but also for the safety of the patient in proton therapy. Recently it is proved that the proton dose distribution has clear relationship with the distribution of the prompt gammas, generated from the proton-induced nuclear reactions in the passage of the proton beam. A proof-of-principle measurement system, which could measure the prompt gammas distribution by scanning method with a single CsI(Tl) scintillation detector and a single-slit collimation system, showed that the location of the distal dose edge or the proton beam range in the water phantom can be accurately determined despite the high level background of neutrons and neutron capture gammas. However, the prototype measurement system developed only to prove the principle cannot be used in real clinical situations, principally due to the very large and heavy (160 cm x 100 cm x 82 cm; 500 kg) scanning system. Moreover, the scanning process used in the study is not suitable for 'spot scanning' or any other scanning techniques, because the spot of the proton beam continuously moves during treatment. For the clinical purpose, a small, array-type prompt gamma detection system incorporating a linear array of multiple CsI(Tl) scintillation detectors, a multi-slit collimation system, and multi-channel DAQ is under development to obviate the problematic scanning process. It is also fundament for the clinical application to assess the prompt gamma distribution according to the phantom size, material, shape, and location (considered the human characteristics) in proton therapy for the accurate determination of the distal dose edge. In the present study, the prompt gamma distributions according to the phantom-to-phantom variation were assessed based on the background fraction, indicating the distribution index as the ratio of the average background gamma counts to the peak gamma counts, by MCNPX code using the Monte Carlo method. The computational burden with the repeated calculation with the various phantoms was reduced with the employment of the parameterized source term and with the use of the 97-node computer cluster