Radiation dose for external exposure to gamma-ray using artificial neural network and MC simulation

The computation of the absorbed dose in air allows the estimation of the concentrations of radionuclides in the soil and the assessment of the external exposure of the human body. The development of numerical models describing gamma ray transport in the environment provides more precise methods to analyze the pathways of external radiation dose.

A combined method using Artificial Neural Network (ANN) and Monte Carlo Simulation (MC) has been developed to calculate the absorbed dose rate in air for photon emitters from natural radionuclides. We proposed a new class of trained ANN to GEANT4 to calculate the probability, for generated photon sources, to reach the detector. Only photons with high probability were tracked in MC Simulation.

A significant reduction of computation time was reached. Unscattered flux and gamma-dose-rate conversion factors were calculated and compared to previous works.

The use of this method overcomes the problem of the long duration of computation time, obtaining a good agreement with previous works and efficient results of the dose rate conversion factor.