“…Inspired from environmental radioactivity measurements [6], decision thresholds (DT) corresponding to a false positive rate equal to 0.1% (probability of type I error) when only natural Bkg is present, were determined by Monte Carlo (MC) simulations with a dictionary comprising several γ-emitters generally used for RPM testing ( 57 Co, 60 Co, 133 Ba, 137 Cs, 241 Am) [7]. As pointed out in [5], the robustness of this approach for automatic decision-making is limited by its inability to take properly into account the sensitivity of the DTs in the case of γ-emitting radionuclides mixed with natural Bkg as well as the variability of their related counting. In order to overcome this problem, which is also sensitive to overlapping between spectral signatures of γ-emitting radionuclides, a sparsity constraint applied on identified radionuclides was implemented for decision-making along with counting estimation with spectral unmixing.…”