Prompt-gamma (PG) imaging has the potential for monitoring proton therapy in real time. Different approaches are investigated. We focus on developing multi-slat collimators to image PG quanta, aiming at optimizing collimator performance to detect deviations in treatment delivery.
...
Prompt-gamma (PG) imaging has the potential for monitoring proton therapy in real time. Different approaches are investigated. We focus on developing multi-slat collimators to image PG quanta, aiming at optimizing collimator performance to detect deviations in treatment delivery. We investigated six different multi-slat configurations, which have either optimal (analytical) intrinsic spatial resolution at fixed efficiency, or otherwise; at different distances from the proton pencil-beam axis (15 cm–35 cm). We used Geant4 to simulate irradiations of the head (energy: 130 MeV) and pelvis (200 MeV) of an anthropomorphic phantom, with and without physiologic/morphologic or setup changes of clinical dosimetric relevance. The particles escaping the phantom were transported through each of these multi-slat configurations and the gamma counts profiles were recorded at the collimator exit. Median filtering was applied to the registered PG-profiles to mitigate the effects of septa shadowing and statistical fluctuations. Time-of-flight discrimination was used to enhance the signal-to-background ratio, which appeared crucial for 200 MeV irradiations. Visual detection of the artificially introduced changes was possible by comparing the PG to the depth-dose profiles. Moreover, 2 mm range shifts could be detected in the head irradiation case using a simple linear regression fit to the falloff of the PG-profile. The influence of changes in complex, patient-like dose distributions on the PG-profiles obtained with multi-slat collimation is first studied in this work, which further gives insight on collimator design optimization and highlights its potential and simplicity for detecting proton treatment deviations over a wide range of Bragg peak positions.
@en