Nuisance Source Population Modeling for Radiation Detection System Analysis
A major challenge facing the prospective deployment of radiation detection systems for homeland security applications is the discrimination of radiological or nuclear 'threat sources' from radioactive, but benign, 'nuisance sources'. Common examples of such nuisance sources include naturally occurring radioactive material (NORM), medical patients who have received radioactive drugs for either diagnostics or treatment, and industrial sources. A sensitive detector that cannot distinguish between 'threat' and 'benign' classes will generate false positives which, if sufficiently frequent, will preclude it from being operationally deployed. In this report, we describe a first-principles physics-based modeling approach that is used to approximate the physical properties and corresponding gamma ray spectral signatures of real nuisance sources. Specific models are proposed for the three nuisance source classes - NORM, medical and industrial. The models can be validated against measured data - that is, energy spectra generated with the model can be compared to actual nuisance source data. We show by example how this is done for NORM and medical sources, using data sets obtained from spectroscopic detector deployments for cargo container screening and urban area traffic screening, respectively. In addition to capturing the range of radioactive signatures of individual nuisance sources, a nuisance source population model must generate sources with a frequency of occurrence consistent with that found in actual movement of goods and people. Measured radiation detection data can indicate these frequencies, but, at present, such data are available only for a very limited set of locations and time periods. In this report, we make more general estimates of frequencies for NORM and medical sources using a range of data sources such as shipping manifests and medical treatment statistics. We also identify potential data sources for industrial source frequencies, but leave the task of estimating these frequencies for future work. Modeling of nuisance source populations is only useful if it helps in understanding detector system performance in real operational environments. Examples of previous studies in which nuisance source models played a key role are briefly discussed. These include screening of in-bound urban traffic and monitoring of shipping containers in transit to U.S. ports.
- Research Organization:
- Lawrence Livermore National Lab. (LLNL), Livermore, CA (United States)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- W-7405-ENG-48
- OSTI ID:
- 971799
- Report Number(s):
- LLNL-TR-418342; TRN: US1001952
- Country of Publication:
- United States
- Language:
- English
Similar Records
Development of Gamma Background Radiation Digital Twin with Machine Learning Algorithms: Application of Unsupervised Machine Learning to Detection of Anomalies and Nuisances in Gamma Background Radiation Environmental Screening Data
Expectation-propagation for weak radionuclide identification at radiation portal monitors