Determination of saturation depth in Compton scattering using Artificial Neural Network

The intensity of Compton scattered γ-ray photons provide useful information about the electron density distribution of a test sample. Because of photon attenuation, the application of this method is limited to a certain depth of the sample (saturation depth). The saturation depth value depends on the energy and intensity of primary photons and on the material of the sample. In this study, we measured the energy spectrum of the scattered photons of 662 keV at 90° with a NaI(Tl) scintillator; and determined the saturation depth of the sample by the Artificial Neural Network (ANN) algorithm. Two sets of samples with known and unknown density were used to train and test the network, respectively. The highest precision (with 0.15% relative error) was achieved by using the Levenbert-Marquardt algorithm with five hidden layers