Abstract
The main objective of this study has been to investigate the feasibility of using tomographic techniques for non-destructive analysis. A potentially useful technique with neutrons as probes for material characterisation is presented. The technique combines the principles of reconstructive tomography with instrumental neutron activation analysis (INAA) so that elemental distributions in a section through a specimen can be mapped. Neutron induced gamma-ray emission tomography (NIGET) technique, where prompt or delayed gamma-rays can be detected in a tomographic mode, has been developed for samples irradiated in the core of a nuclear reactor and used in studies of different biological matrices. The capabilities of the technique will be illustrated using a spatial resolution of 1 mm. The quantitative usefulness of NIGET depends on the accuracy of compensation for the effect of scattering and attenuation as well as determination of the tomographic system characteristics which contribute to the intrinsic measurement process. It will be shown how quantitative information about the induced radionuclide concentration distribution in a specimen can be obtained when compensation for scattered gamma-rays is taken into account employing a high resolution semiconductor detector and a method of scattering correction based upon the use of three energy windows to collect emission data. For attenuation correction an iterative method which combined emission and transmission measurements has been implemented and its performance was compared to the performance of a number of other attenuation correction algorithms. The work involved investigation into the role of a number of factors which influence the accuracy of data acquisition. An efficiency-resolution figure of merit as a function of collimator efficiency, system resolution and object diameter has been defined. Further, a number of reconstruction techniques were investigated and compared for accuracy, minimum number of projections required and their ability to handle noise. Reconstruction by filtered back projection was fastest to compute, but performed poorly when compared to iterative techniques.