A novel and fast technique for evaluation of plastic rod scintillators as position sensitive gamma-ray detectors using artificial neural networks

Abstract

The problem of how to precisely estimate the radiation interaction position is an important parameter in medical and industrial imaging systems. This paper presents a new methodology for prediction of the incident position of the gamma rays based attenuation technique and multilayer perceptron (MLP) neural network system. The detection system is comprised of a gamma-ray source (60Co or 137Cs) and a plastic rod scintillator (BC400) coupled with just one PMT at one side. The experimental setup provides the required data for training and testing the network. Using this proposed method, the radiation interaction position was predicted in plastic rod scintillator with a mean absolute error less than 0.8 and 0.5 for 137Cs and 60Co sources, respectively. The mean relative error percentage was calculated less than 3.9% and 3.5%. Also, the correlation coefficient was measured 0.999 and 0.998, respectively. The results showed that the predicted interaction position using the ANN method is in good agreement with real data. The techniques and set up used in the previous position sensitive detectors were fairly complicated whereas the new set-up and proposed method are really very simple. Also, the radiation safety, cost and shielding, and electronics requirements are minimized and optimized.