Reduced Secondary Neutrons and Occupational Exposure With Optimization of Beam Shaping Device Material in Proton Therapy

Purpose/Objective(s): The goal of this study was to determine the empirical accuracy of in-vivo proton-positron activation therapeutic field longitudinal range and lateral centroid alignment characterization with respect to the intended treatment plan. Materials/Methods: The study was part of an ongoing clinical trial for treatment of pediatric brain tumors. In case of proton therapy, treatment planning was attempted to provide full target coverage with single uniform beams as part of multi-field plans. Positron emission tomography was used to acquire the proton-induced positron activation (proton-PET) distribution from 15 patients and 33 total fields. To evaluate individual proton fields the patients received single field fractions on a select proton-PET treatment days. The acquired proton-PET studies were fused to the patient’s CT planning study using a commercial radiation therapy planning (RTP) system. The RTP was used to export the following data: proton field percentile range in patient R90 along three paths, on axial planning study beam projected axis, and parallel but offset +/0.3 cm, proton-PET percentile activations (A) along the same paths A90, A80, A50, A20, A10, and the A80/A20 trend, proton-PET centroid position differentials at isocenter determined by percentile comparisons between proton lateral (PL) field defining PL50, and associated proton-PET lateral activations AL80, AL50, AL25, AL20, and the AL80/20 trend. Results: The resulting errors from geometric fitting between proton-PET and proton planned field data ranged longitudinally up to 6.0 cm for R90A90 and laterally up to 1.6 cm for PL50-AL80. The best longitudinal correlation was between the 10 percent proton-PET signal and the 90 percent planned proton dose (R90-A10) resulting in an error of 0.4 cm and standard deviation of 0.3 cm. The isocentric lateral plane proton-PET correlation was also optimal at the lowest magnitude studied, 20-25 percent proton-PET signal level (PL50-AL25, PL50-AL20) giving an error of 0.2 cm and standard deviation of 0.2 cm. Conclusions: For the empirical proton-PET/proton range as well as the proton-PET/lateral proton field centroid correlations the fits improved proportionally to lower relative proton-PET activity. The residual errors from the optimal correlations are comparable to RTP longitudinal and lateral margins providing the potential to give valuable in vivo clinical validation of planned delivery accuracy. Author Disclosure: J.B. Farr: None. Z. Li: None. D.J. Indelicato: None. R. Lukose: None. A. Gajjar: None. T.E. Merchant: None.