[2J01-08] Dose Assessment

[2J04] Development of estimation methods to improve the accuracy of Compton camera.

*Yoshiki Ohori¹, Takeo Torii¹, Tsugiko Takase¹, Katsuhiko Yamaguchi¹ (1. Fukushima univ.)

Keywords:Compton Camera, Machine learning, Estimation of Source Distribution

Compton-camera have been used to visualize radioactive materials and to determine the effectiveness of decontamination. However, when there is shielding between the radiation source and the Compton-camera, the distribution of radioactive materials spreads out, making it difficult to determine the detailed location and shape of the radiation source. Therefore, we turned our attention to machine learning, in which γ-ray energy-spectrum data are input to a neural-network (NN) to estimate the radiation source distribution. In this study, we attempted to verify to what extent the position and shape of a lead plate placed between the source and the Compton-camera could be identified by having the NN learn and analyze the γ-ray energy information and source distribution data obtained by the Compton-camera.