Passive Gamma Source Imaging Using Compressed Sensing Principles
MetadataShow full item record
A design concept for imaging radiation sources using compressed sensing (CS) principles was developed in this research. A proof-of-concept system model was designed using Monte Carlo N-Particle (MCNP) simulations. The resulting conceptual system comprised a collimator made of 11×11 polyvinyl chloride (PVC) pipes placed upright in the middle of a 250-gallon water-filled rectangular tank, and a detection system that consisted of six Geiger-Muller (GM) detectors. The collimator channels were modeled to be randomly filled with water or air to provide different measurement configurations. Image reconstructions were performed using l1-minimization and non-negative least squares (NNLS) methods for four cases with varying 137Cs source locations. Localization and shape identification were shown to be successful for a point source and ring-shaped sources using data from MCNP simulations. The NNLS method was selected to be applied in image reconstruction for model validation. The proof-of-concept system was physically assembled for collection of measurement data. Model validation with experimental measurements was performed by first refining the GM detector modeling. New methods of modeling GM detectors using the cell flux (F4) and the energy cell flux (*F4) tallies were proposed. The F4 method was selected to model GM detectors in this research and was validated with experimental measurements. The MCNP system model was shown to represent the experimental model based on the proven capability of localizing the gamma source. The MCNP system model was used as a baseline design model and design factors that can improve its performance were studied. These factors were found to be collimator material and geometry, number and size of collimator channels, the source-to-detector distance and the type of detector used for measurements. A set of high-level design recommendations were derived based on these design factors, which can serve as guidelines for system developers to design and optimize a CS-based radiation source imaging system. The findings from this work contribute to the diversification of counter terrorism tools, encourage future research on NNLS technique application for image reconstruction and promote cheaper radiation source imaging system development. These research findings will also help motivate higher engagement among countries in strengthening global nuclear security initiative.
Anuar, Nuraslinda (2021). Passive Gamma Source Imaging Using Compressed Sensing Principles. Doctoral dissertation, Texas A&M University. Available electronically from