| Popis: |
In spring 2019 the international atomic energy agency (IAEA) has run a public competition to look at tomographic reconstruction algorithms for its PGET detector, used to determine activity levels in nuclear reactors spent fuel rods. The algorithm is also supposed to identify missing fuel rods. The task is particularly challenging because uranium rods have a very high cross section for gamma photons in the energy range of interest. Therefore, signal from pins at the center of the assembly is heavily attenuated. Furthermore, most photons reaching the detector will have undergone Compton scattering, originating a high background deprived of spatial information. In this paper, we describe our competition entry. Image reconstruction is based on robust, unbiased, emission tomography algorithms that have been routinely used for decades in nuclear medicine. Attenuation correction is performed by subsequent approximations. An intermediate reconstruction is used to detect missing pins. This information is used to generate the final attenuation map, which is used to generate the image used for quantification. The compton scatter background is estimated by comparing the counts in the different energy windows. Results of reconstructions of training datasets in presence of 60Co mock assemblies show a 100% accuracy in the determination of missing pins and high quantitative accuracy. |
