SMuK 2023 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
ST: Fachverband Strahlen- und Medizinphysik
ST 8: AI Topical Day – AI in Medicine (joint session ST/AKPIK)
ST 8.4: Vortrag
Donnerstag, 23. März 2023, 14:45–15:00, ZEU/0146
Selection of Compton events in the SiFi-CC camera using convolutional neural networks — •George Farah1, Ronja Hetzel1, Jonas Kasper1, Alexander Fenger1, Achim Stahl1, and Aleksandra Wrońska2 — 1III. Physikalisches Institut B, RWTH Aachen University — 2M. Smoluchowski Institute of Physics, Jagiellonian University Kraków, Poland
Proton therapy is a promising form of cancer treatment that uses charged protons to target and kill cancer cells. One of the main challenges in proton therapy is accurately determining the depth at which the protons will deposit their energy in the tumor.
The SiFi-CC (SiPM and scintillating Fiber-based Compton Camera) aims to enable range detection in proton therapy. It consists of multiple scintillating LYSO fibers generating signals that get read by SiPMs attached to both ends of the fibers. The camera utilizes the Compton effect and photoelectric effect to detect the prompt gamma rays produced in nuclear interactions of the protons with the nuclei in the tumor. This allows restricting the origin of the prompt gamma to a cone surface and by reconstructing many of such cones it is possible to reconstruct the source distribution of the prompt gammas.
The most recent SiFi-CC geometry has four fibers coupled to one SiPM in a shifted manner, so signals from multiple fibers get read by a single SiPM. In this talk, we present how three-dimensional neural networks can be advantageous by taking into consideration this new geometry. Hence improving the detection of Compton events, which improves the accuracy of range detection in proton therapy.