ESTRO 2022 - Abstract Book

S558

Abstract book

ESTRO 2022

Zagreb, Croatia; 8 Target Systemelektronik GmbH & Co. KG, -, Wuppertal, Germany; 9 Holland Proton Therapy Centre - HollandPTC, -, Delft, The Netherlands; 10 Ru đ er Bo š kovi ć Institute, Division of Experimental Physics, Zagreb, Croatia; 11 University of Bergen, Department of Physics and Technology, Bergen, Norway Purpose or Objective Robust and sufficiently quick treatment verification for proton therapy is a well-known challenge. While detection systems for prompt gamma rays (PG) are maturing, the concept of measuring fast neutrons (FN) produced by inelastic proton interactions in the patient is relatively new. In this study, the feasibility of a novel treatment verification concept measuring both FN and PG was evaluated through MC simulations. The hypothesis was that the combination of FN and PG would increase the treatment validation sensitivity. To this end, a hybrid, beamspot-specific Range Landmark (RL) was defined and its value was correlated to deviations in the proton range. Simulations of a patient geometry revealed the upper limits of the sensitivity of the method. Materials and Methods We propose the use of a quasi-monolithic detector array (QuDA, figure 1), consisting of densely stacked and optically segmented organic bar-shaped scintillator elements with 1-2 ns decay times and pulse shape discrimination (PSD) for FN / PG identification. Fast photodetectors placed on both ends of a scintillator element measure the relative interaction position. MC simulations of a 30x20x20 cm ³ QuDA and a lung cancer patient CT from the Cancer Imaging Archive were performed with GATE / Geant4. Fifteen range-shifted versions of the patient dataset were made by removing or adding surface tissue, ranging between -5 and +5 mm. An 85 MeV posterior-anterior proton beam was stopped centrally in the lung tumor 3 cm in diameter. Consecutive PG or FN interactions in the QuDA allows for kinematic reconstruction of their respective production vertices. In the present study, to estimate the upper limit of the setup sensitivity, the MC truth production vertices of PGs and FNs interacting in the QuDA were used as a basis for the RL: calculated as the mean value plus one standard deviation of the distribution of production vertices along the beam. RL distributions were created by subsampling bootstrapping, allowing for variable proton intensity. The minimum detectable range shift was found by comparing the RL distributions of several artificial range shifts using AUROC with a detection threshold of 0.9.

Results While the PG production was isotropic (allowing for several possible QuDA positions), the FNs were produced predominantly in the forward direction. The optimal QuDA position in terms of particle detection was anterior to the patient, along the beam axis. The proton intensities for the minimum detectable range shifts are shown in figure 2. Combining FN and PG in a single detector considerably increased the sensitivity, and only 20 M protons were necessary to detect a spot-wise range shift of 1 mm.