Nuclide identification of radioactive sources from gamma spectra using artificial neural networks

Abstract

Gamma spectroscopy is commonly used to identify the radionuclides present in samples or materials, by using the existing knowledge on the gamma ray energies and intensities for each radionuclide. However, when dealing with samples where the composition, internal configuration and shielding materials are unknown, as is the case, for instance, in nuclear security applications, the task can become challenging. Furthermore, gamma detection systems in field applications often do not have the high resolution typical of controlled laboratory conditions. In this work, we apply artificial intelligence techniques for automated identification of radioactive sources from gamma spectra obtained with a LaBr₃(Ce) detector with 3.6 % resolution at 662 keV. Combinations of up to 10 sources in each spectrum were used to train and test the artificial neural network developed. We report on the results, which show effective nuclide identification of radioactive sources from gamma spectra using ANNs.