Data for training and testing radiation detection algorithms in an urban environment

James M Ghawaly,Brian J Quiter,Daniel E Archer,Michael J Willis,Andrew D Nicholson,Kary L Myers,Douglas E Peplow,Christine M Anderson-Cook

doi:10.1038/s41597-020-00672-2

Abstract

The detection, identification, and localization of illicit nuclear materials in urban environments is of utmost importance for national security. Most often, the process of performing these operations consists of a team of trained individuals equipped with radiation detection devices that have built-in algorithms to alert the user to the presence nuclear material and, if possible, to identify the type of nuclear material present. To encourage the development of new detection, radioisotope identification, and source localization algorithms, a dataset consisting of realistic Monte Carlo–simulated radiation detection data from a 2 in. × 4 in. × 16 in. NaI(Tl) scintillation detector moving through a simulated urban environment based on Knoxville, Tennessee, was developed and made public in the form of a Topcoder competition. The methodology used to create this dataset has been verified using experimental data collected at the Fort Indiantown Gap National Guard facility. Realistic signals from special nuclear material and industrial and medical sources are included in the data for developing and testing algorithms in a dynamic real-world background.

Highlights

Background & SummaryThe US government performs radiation detection, identification, and localization campaigns for a variety of scenarios, including emergency response, large public gatherings, and political events
These radiation detection campaigns are generally conducted by trained teams equipped with radiation detection systems that can be carried by hand, mounted on an automobile, or mounted on unmanned robotic systems
Because gamma rays emitted by different radioisotopes exhibit characteristic, discrete energies, the gamma ray spectrum can potentially be used to identify the radioisotope(s) detected

Summary

Background & Summary

The US government performs radiation detection, identification, and localization campaigns for a variety of scenarios, including emergency response, large public gatherings (e.g., concerts and sporting events), and political events (e.g., presidential inaugurations). This variation in both absolute and relative NORM concentration in different materials means that both the gross count rate and spectroscopic signal read by a detector when moving throughout a search area can change dramatically from one location to another[4,5,6,7] This is especially true in urban environments, where the composition of buildings and their resulting radioactive signatures is varied (e.g., a granite building may be placed directly next to a concrete building)[8]. The parameters used to develop this model are controlled and known with absolute certainty, which is an extremely difficult condition to obtain in real-world experimental data This level of ground truth is required to accurately assess the performance of radiation detection, identification, and localization algorithms, making this dataset a valuable asset to the algorithm development community. This dataset allowed the designers of the ARAD algorithm to evaluate new algorithms on well-controlled data and develop performance metrics, such as minimum detectable activity, receiver operator characteristic curves, and probability of detection curves[14]

Methods

Code availability