Lawrence Berkeley National Laboratory

The Stand-off Radiation Detection System (SORDS) program is an Advanced Technology Demonstration (ATD) project through the Department of Homeland Security’s Domestic Nuclear Detection Office (DNDO) with the goal of detection, identification and localization of weak radiological sources in the presence of large dynamic backgrounds. The Raytheon-SORDS Tri-Modal Imager (TMI) is a mobile truckbased, hybrid gamma-ray spectroscopic and imaging system able to quickly detect, identify and localize, radiation sources at standoff distances through improved sensitivity provided by multiple detection modes while minimizing the false alarm rate. Reconstruction of gamma-ray sources is performed using a combination of gamma-ray spectroscopy and two imaging modalities; coded aperture and Compton scatter imaging. The TMI consists of 35 sodium iodide (NaI) crystals (5x5x2 in3 each), arranged in a random coded aperture mask array (CA), followed by 30 position sensitive NaI iv bars (24x2.5x3 in3 each) called the detection array (DA). The CA array acts as both a coded aperture mask and scattering detector for Compton events. The large-area DA array acts as a collection detector for both Compton scattered events and coded aperture events. In this thesis, the implemented spectroscopic, coded aperture, Compton and hybrid imaging algorithms will be described along with their performance. It will be shown that multiple imaging modalities can be fused to improve detection sensitivity over a broader energy range than any mode alone. Since the TMI is a moving system, peripheral data, such as a Global Positioning System (GPS) and Inertial Navigation System (INS) must also be incorporated. A method of adapting static imaging algorithms to a moving platform has been developed. Also, algorithms were developed in parallel with detector hardware, through the use of extensive simulations performed with the Geometry and Tracking Toolkit v4 (GEANT4). Simulations have been well validated against measured data. Results of image reconstruction algorithms at various speeds and distances will be presented as well as localization capability. Utilizing imaging information will show signal-to-noise gains over spectroscopic algorithms alone.