IEEE TRANSACTIONS ON NUCLEAR SCIENCE, VOL. 61, NO. 2, APRIL 2014 985 Effects of Background on Gamma-Ray Detection for Mobile Spectroscopy and Imaging Systems Timothy J. Aucott, Mark S. Bandstra, Victor Negut, Joseph C. Curtis, Daniel H. Chivers, and Kai Vetter Abstract—The presence of gamma-ray background significantly reduces detection sensitivity when searching for radioactive sources in the field, particularly in mobile systems which must contend with a variable background that is not known a priori. An extensive survey of the background was performed in the San Francisco Bay Area using both sodium iodide and high-purity germanium detectors, covering a wide variety of environments that might be encountered in an operational scenario. This data was used as a basis for source injection in a moving detector sce- nario in order to assess the effects of the background on different detection approaches. Both imaging and spectroscopic algorithms were implemented for the sodium iodide array, and their per- formances are compared for a variety of source energies and stand-off distances in the presence of the measured background. Index Terms—Gamma-ray detection, radiation imaging, spectroscopy. I. INTRODUCTION T HERE are a number of scenarios in which a weakly gamma-emitting source needs to be detected in the pres- ence of a large background. For well-controlled cases, such as those in low-count-rate laboratories, the background can be minimized and then measured with a great deal of accuracy. For many applications, however, a detector system is brought out into the field and exposed to a background that may not be well understood. This background may arise from a variety of sources, both naturally-occurring and anthropogenic, but in both cases will limit the detection of the source. Not only does this background add significant statistical noise, but there are large systematic uncertainties that arise from changes between different environments [1], [2]. Mobile detector systems, in particular, are constantly contending with a background that is changing over time and location, and which is not necessarily known or measurable a priori. Manuscript received September 29, 2013; revised January 15, 2014; accepted February 10, 2014. Date of current version April 10, 2014. This work was supported by the U.S. Department of Homeland Security under Grant Award 2011-DN-077-ARI049-03. T. J. Aucott, V. Negut, and J. C. Curtis are with the Department of Nu- clear Engineering, University of California, Berkeley, CA 94720 USA (e-mail: tjaucott@berkeley.edu). D. H. Chivers and M. Bandstra are with the Nuclear Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720 USA. K. Vetter is with the Department of Nuclear Engineering, University of Cal- ifornia, Berkeley, CA 94720 USA, and also with the Nuclear Science Division, Lawrence Berkeley National Laboratory, Berkeley, CA 94720 USA. Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TNS.2014.2306998 In any detection system, some threshold must be set as a de- cision criterion for the presence or absence of a source. The challenge faced by mobile detectors is that different environ- ments contribute different rates of background. In particular, if a threshold is chosen based on statistical variance alone, the false alarm rate of the detector will likely be much higher than ex- pected due to the systematic variance. This study focuses on scintillator-based systems, which are commonly used in these types of applications due to their relatively low cost per volume. Because of their lower energy resolution (compared to semi- conductor detectors), the effects of background are particularly pronounced, as counts in the source peak will be convolved with nearby background lines. One important goal is to mea- sure the extent of these changes in the real-world environment and to assess its impact on detection schemes that are commonly employed. Two well-established approaches for detecting these sources are gamma spectroscopy and imaging. Both methods aim to in- crease the signal-to-noise ratio of the system by looking for a feature in some reduced space; for example, spectroscopy might look for a photopeak in energy space, while imaging looks for a peak pixel in the image space. Imaging is potentially less sensi- tive to background uncertainty [3], but at the expense of lower absolute efficiency. In this study, the imager in question is a passive coded aperture, with an open fraction of 50%. Spec- troscopy, on the other hand, has a higher efficiency, but may not necessarily account for the background variability. Of course, an imaging system has the additional ability to localize the ra- diation source, and spectroscopy can identify the isotope being detected, but this work focuses on the ability to compensate for background and detect the source. These two methodologies were compared by measuring the gamma background and then injecting source photons into the background data set to mimic a stationary source some dis- tance from the path of the moving detector. This method cre- ates two distributions, one for the background alone and one for the source plus the background. By varying the alarm threshold across the range of the two distributions, a receiver operator characteristic (ROC) curve is created by plotting the probability of true positives against the false positives. The background is divided into separate runs, with a length determined by the par- ticular standoff distance. The algorithms then report a single de- cision on the presence or absence of a source in each run. In order to make a fair comparison of false alarm rate, only one isotope and standoff distance are considered at a time, which are assumed by both algorithms. In a real system, of course, the isotope and location are not known a priori; this simplification, however, results in a more controlled study. 0018-9499 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.