Application of artificial neural network in non-destructive Compton scattering densitometry

This study investigates the use of artificial neural networks (ANN) in Compton scattering densitometry. Samples with different densities were irradiated by gamma rays and the spectra of photons, scattered at 90°, were recorded by a NaI scintillator. These data were used to train the network and to validate its performance. After various training functions with different structures of layers were examined, by comparing the ANN predicted results with the experimental ones, the best algorithm was adopted for the ANN.