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

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.

An Ensemble Approach to Computationally Efficient Radiological Anomaly Detection and Isotope Identification

“…The first dataset comes from the radiological source search competition held in 2018 and is publicly available [20], [21]. Hence, it will be referred to as the competition dataset hereinafter.…”

Neural Network Approaches for Mobile Spectroscopic Gamma-Ray Source Detection

“…Note that theoretical lower bounds for MDA can be estimated for each method [6], which can be useful in assessing potential performance improvements for a given algorithm. The runs of sequential background spectra are generated from simulations of a detector system moving through an urban environment, originating from a dataset produced as part of a public data competition [23,24]. List-mode gamma-ray events were produced from a 2" × 4" × 16" NaI(Tl) detector moving through a simulated urban environment.…”

Understanding the merits of winning data competition solutions for varied sets of objectives

“…The post-competition comparison of algorithms based on a subset of the objectives described in this paper is illustrated with results from the 2019 data competition focused on radiological detection in an urban environment. The competition used simulated measurements mimicking those collected by a thallium-doped sodium iodide detector being driven along typical US urban streets [11] and challenged the competitors to find effective solutions to detect, identify, and locate several radioactive sources.…”

“…The sizes of the training and test datasets were determined to make sure sufficient data are made available to the competitors for training their algorithms to solve the general problem of interest but the majority of the data was reserved for feedback for the competitors in the public data and then to determine the winner. Data were generated using a stochastic simulation code developed at Oak Ridge National Laboratory [11] for a variety of different scenarios by manipulating internal model parameters to mimic the breadth of urban environments seen in practice.…”

“…The final ranking of the competitors' algorithms and the allocation of the prize money were determined by the private leaderboard score with a weighted average of scores based on the ability to detect, identify and locate all six radioactive sources with an acceptable FPR [11]. This combined metric encouraged competitors to value all competition objectives, but for specific operational search scenarios after the competition, this might not lead to the best choice of solution.…”