SMuK 2023 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
AKPIK: Arbeitskreis Physik, moderne Informationstechnologie und Künstliche Intelligenz
AKPIK 11: AI Topical Day – AI in Medicine (joint session ST/AKPIK)
AKPIK 11.6: Talk
Thursday, March 23, 2023, 15:15–15:30, ZEU/0146
Thermoluminescence glow curve generation using generative adversarial networks (GANs) — •Evelin Derugin1, Olaf Nackenhorst1, Florian Mentzel1, Jens Weingarten1, Kevin Kröninger1, and Jörg Walbersloh2 — 1Department of Physics, TU Dortmund University — 2Materialprüfungsamt NRW
Personal dose monitoring is essential for a successful radiation protection program for occupationally exposed persons. The Materialprüfungsamt NRW (MPA NRW) provides thermoluminescence (TL) dosimeters based on LiF:Mg,Ti. Proof-of-concept studies to predict the day of irradiation have been successfully performed on measured TL glow curves using artificial neural networks (ANN). However, large data sets are required to train an ANN to predict the parameters of new measurements. Therefore the Department of Physics at TU Dortmund is developing multivariate methods for generating TL glow curves using generative adversarial networks (GANs). These generated glow curves will be used as training data for the irradiation day prediction model. This study trains GANs to generate glow curves using a measured data set of 4100 glow curves with 28 irradiation dates. In this talk, we present the comparison of the simulated glow curves with the measured ones and provide information about the performance and optimization of the GAN.