A new formulation for polymer fricke dosimeter and an innovative application of neural network to study dose profile from spin-echo NMR data

Abstract

Dosimetric systems are used to evaluate absorbed dose and the induced effect caused by irradiation. The choice of dosimetric systems depends on their chemical and physical characteristics and in this study, the chemical Fricke dosimeter was prepared and experimentally tested with Poly(Ethylene Oxide) (PEO), a thermoplastic material. It was provided that the usage of PEO material, a cheap and clean polymer, instead of bovine gelatin, very commonly used in Fricke gel dosimeters, is a promising technological development in dosimetric systems. The improved accuracy of the dose estimation in these systems was demonstrated by comparing the results of ultraviolet–visible spectroscopy (UV-VIS) for the PEO-Fricke dosimeters and the Fricke made of bovine gel. Two sample sets of PEO-Fricke dosimeters were prepared, one set was irradiated with different absolute doses and the other set was prepared with different Fe (III) concentrations. The analyses were performed by using two different algorithms to treat Spin-Echo NMR data. The first methodology uses the single and bi-exponential functions to fit the decay data using the Nonlinear least square (NLLS) with the Levenberg-Marquardt algorithm, providing the relaxation time (T2) values as an average value. The second, more appropriate, consider the problem as a Fredholm integral equation to determine the T2 distribution functions by using Hopfield Neural Network (HNN). The results provide information about the ion-polymer interaction and spatial information of the dosimetric system.