Parts | Days | Selection | Search | Updates | Downloads | Help
ST: Fachverband Strahlen- und Medizinphysik
ST 4: Poster session
ST 4.8: Poster
Tuesday, April 2, 2019, 14:00–16:00, Poster B1
Determination of the photon interaction position in a monolithic scintillator applied in a Compton Camera system — •Maria Kawula1, Silvia Liprandi1, Tim Binder1,2, Rita Viegas Rego1,3, Ben Hoyle1, Katia Parodi1, and Peter G. Thirolf1 — 1LMU München, Munich, Germany — 2KETEK GmbH, Munich, Germany — 3University of Coimbra, Portugal
The LMU Compton Camera is being developed for ion beam range verification during particle therapy. The system detects prompt gamma rays produced after tissue irradiation. The camera consists of a scatterer (6 layers of double-sided Si-strip detectors) and a monolithic LaBr3(Ce) scintillator as an absorber, read out by a multianode photomultiplier. To determine the photon interaction position in the scintillator the "Categorical Average Pattern" (CAP) algorithm [1] is used. This algorithm is based on the comparison of every recorded photon event with a reference library of 2D light amplitude distributions obtained by scanning the scintillator front surface with tightly collimated 60Co and 137Cs sources respectively in 10404 positions (400 photopeak events per position are acquired). A second method based on Convolutional Neural Networks (CNN) is under development. The reference library acquired for the CAP algorithm is used as training data. The architecture of the network as well as a quantitative comparison of CAP and CNN in terms of computational time, memory consumption and obtained spatial resolution will be presented. This work is supported by the DFG Cluster of Excellence Munich Centre of Advanced Photonics (MAP).