RADAI: A Large-Scale Realistic Dataset for Radiation Detection Algorithm Development

James M. Ghawaly; Daniel E. Archer; Andrew D. Nicholson; Douglas E. Peplow; Nicholas J. Prins; Tenzing H.Y. Joshi; Mark S. Bandstra; Andrew C. Jones; Brian J. Quiter; Abigael C. Nachtsheim

doi:10.1109/TNS.2026.3682654

RADAI: A Large-Scale Realistic Dataset for Radiation Detection Algorithm Development

James M. Ghawaly
, Daniel E. Archer
, Andrew D. Nicholson
, Douglas E. Peplow
, Nicholas J. Prins
, Tenzing H.Y. Joshi
, Mark S. Bandstra
, Andrew C. Jones
, Brian J. Quiter
, Abigael C. Nachtsheim

Research output: Contribution to journal › Article › peer-review

Abstract

Open, realistic datasets are essential for developing and benchmarking radiation detection algorithms, yet they remain scarce. We introduce the Radiological Anomaly Detection and Identification (RADAI) dataset, a large-scale synthetic resource that integrates high-fidelity Monte Carlo simulations with realistic urban scenarios to capture both background variability and source signatures. RADAI models construction-material NORM, people and vehicles, urban clutter, and dynamic environmental effects such as cosmic-ray and rain-induced transients, and it provides list-mode detector data with motion and response modeling suitable for algorithm training and evaluation. We publicly release three complementary datasets for this purpose (training, developer, and testing) together with an online scoring portal for standardized performance assessment and an open software toolkit that supports data access, augmentation, model development, and evaluation. These resources enable reproducible comparisons across methods and promote rigorous studies at the scale required by contemporary machine learning (ML). By grounding algorithm development in realistic, well-documented conditions, RADAI supports progress toward more robust detection, identification, and localization in complex urban environments.

Original language	English
Pages (from-to)	2045-2066
Number of pages	22
Journal	IEEE Transactions on Nuclear Science
Volume	73
Issue number	5
DOIs	https://doi.org/10.1109/TNS.2026.3682654
State	Published - May 1 2026

Funding

Rdve 28 January 2026; visedre 16 March 2026; accepted 4 April 2026. Date of publication 9 April 2026; date of current ersionv 19 May 2026. This orkw asw supported by U.S. National Nuclear Security Adminis- tration (NNSA) O ffice of Defense Nuclear Nonproliferation Research and Dtve through U.S. Department of Energy at Oak Ridge National Laboratory (ORNL) under Contract AC05-00OR2272, Lawrence Berkyele National Laboratory (LBNL) under Contract DE-AC02-05CH11231, and Los Alamos National Laboratory (LANL) under Contract 89233218CNA000001. (Corresponding author: James M. Ghawaly Jr.) James M. Ghaalyw J.r is with the Division of Computer Science and Engineering, Louisiana State U,evn Baton Rouge, LA 70726 USA (e-mail: [email protected]).wjgha Daniel E. Archer and Andrew D. Nicholson are with the Physics Division, Oak Ridge National L,a Oak Ridge, TN 37830 USA. Douglas E. Peplow and Nicholas J. Prins are with the Nuclear Energy and Fuel Cycle Division, Oak Ridge National L,a Oak Ridge, TN 37830 USA. Tg H. Y. Joshi asw with the Nuclear Science Division, Lawrence Berkyele National L,a B,yee CA 94720 USA. He is with KBd Metals, B,yee CA 94704 USA. Mark S. Bandstra, Andrew C. Jones, and Brian J. Quiter are with the Nuclear Science Division, Lawrence Berkyele National L,a B,yee CA 94720 USA. Abigael C. Nachtsheim is with the C,o Computational, and Sta- tistical Sciences Division, Los Alamos National L,a Los Alamos, NM 87545 USA. Color ersionsv of one or more figures in this article are ailableva at //doi.org /10.1109/TNS.2026.3682654. This work was supported by U.S. National Nuclear Security Administration (NNSA) Office of Defense Nuclear Nonproliferation Research and Development through U.S. Department of Energy at Oak Ridge National Laboratory (ORNL) under Contract AC05-00OR2272, Lawrence Berkeley National Laboratory (LBNL) under Contract DE-AC02-05CH11231, and Los Alamos National Laboratory (LANL) under Contract 89233218CNA000001.

Keywords

Benchmark dataset
design of experiments (DoE)
machine learning (ML)
radiation detection
radiation detection algorithm
simulation

Access to Document

10.1109/TNS.2026.3682654

Cite this

@article{11b85274632344dc887a4fa369247858,

title = "RADAI: A Large-Scale Realistic Dataset for Radiation Detection Algorithm Development",

abstract = "Open, realistic datasets are essential for developing and benchmarking radiation detection algorithms, yet they remain scarce. We introduce the Radiological Anomaly Detection and Identification (RADAI) dataset, a large-scale synthetic resource that integrates high-fidelity Monte Carlo simulations with realistic urban scenarios to capture both background variability and source signatures. RADAI models construction-material NORM, people and vehicles, urban clutter, and dynamic environmental effects such as cosmic-ray and rain-induced transients, and it provides list-mode detector data with motion and response modeling suitable for algorithm training and evaluation. We publicly release three complementary datasets for this purpose (training, developer, and testing) together with an online scoring portal for standardized performance assessment and an open software toolkit that supports data access, augmentation, model development, and evaluation. These resources enable reproducible comparisons across methods and promote rigorous studies at the scale required by contemporary machine learning (ML). By grounding algorithm development in realistic, well-documented conditions, RADAI supports progress toward more robust detection, identification, and localization in complex urban environments.",

keywords = "Benchmark dataset, design of experiments (DoE), machine learning (ML), radiation detection, radiation detection algorithm, simulation",

author = "Ghawaly, \{James M.\} and Archer, \{Daniel E.\} and Nicholson, \{Andrew D.\} and Peplow, \{Douglas E.\} and Prins, \{Nicholas J.\} and Joshi, \{Tenzing H.Y.\} and Bandstra, \{Mark S.\} and Jones, \{Andrew C.\} and Quiter, \{Brian J.\} and Nachtsheim, \{Abigael C.\}",

note = "Publisher Copyright: {\textcopyright} 1963-2012 IEEE.",

year = "2026",

month = may,

day = "1",

doi = "10.1109/TNS.2026.3682654",

language = "English",

volume = "73",

pages = "2045--2066",

journal = "IEEE Transactions on Nuclear Science",

issn = "0018-9499",

publisher = "Institute of Electrical and Electronics Engineers",

number = "5",

}

TY - JOUR

T1 - RADAI

T2 - A Large-Scale Realistic Dataset for Radiation Detection Algorithm Development

AU - Ghawaly, James M.

AU - Archer, Daniel E.

AU - Nicholson, Andrew D.

AU - Peplow, Douglas E.

AU - Prins, Nicholas J.

AU - Joshi, Tenzing H.Y.

AU - Bandstra, Mark S.

AU - Jones, Andrew C.

AU - Quiter, Brian J.

AU - Nachtsheim, Abigael C.

PY - 2026/5/1

Y1 - 2026/5/1

N2 - Open, realistic datasets are essential for developing and benchmarking radiation detection algorithms, yet they remain scarce. We introduce the Radiological Anomaly Detection and Identification (RADAI) dataset, a large-scale synthetic resource that integrates high-fidelity Monte Carlo simulations with realistic urban scenarios to capture both background variability and source signatures. RADAI models construction-material NORM, people and vehicles, urban clutter, and dynamic environmental effects such as cosmic-ray and rain-induced transients, and it provides list-mode detector data with motion and response modeling suitable for algorithm training and evaluation. We publicly release three complementary datasets for this purpose (training, developer, and testing) together with an online scoring portal for standardized performance assessment and an open software toolkit that supports data access, augmentation, model development, and evaluation. These resources enable reproducible comparisons across methods and promote rigorous studies at the scale required by contemporary machine learning (ML). By grounding algorithm development in realistic, well-documented conditions, RADAI supports progress toward more robust detection, identification, and localization in complex urban environments.

AB - Open, realistic datasets are essential for developing and benchmarking radiation detection algorithms, yet they remain scarce. We introduce the Radiological Anomaly Detection and Identification (RADAI) dataset, a large-scale synthetic resource that integrates high-fidelity Monte Carlo simulations with realistic urban scenarios to capture both background variability and source signatures. RADAI models construction-material NORM, people and vehicles, urban clutter, and dynamic environmental effects such as cosmic-ray and rain-induced transients, and it provides list-mode detector data with motion and response modeling suitable for algorithm training and evaluation. We publicly release three complementary datasets for this purpose (training, developer, and testing) together with an online scoring portal for standardized performance assessment and an open software toolkit that supports data access, augmentation, model development, and evaluation. These resources enable reproducible comparisons across methods and promote rigorous studies at the scale required by contemporary machine learning (ML). By grounding algorithm development in realistic, well-documented conditions, RADAI supports progress toward more robust detection, identification, and localization in complex urban environments.

KW - Benchmark dataset

KW - design of experiments (DoE)

KW - machine learning (ML)

KW - radiation detection

KW - radiation detection algorithm

KW - simulation

UR - https://www.scopus.com/pages/publications/105036432657

U2 - 10.1109/TNS.2026.3682654

DO - 10.1109/TNS.2026.3682654

M3 - Article

AN - SCOPUS:105036432657

SN - 0018-9499

VL - 73

SP - 2045

EP - 2066

JO - IEEE Transactions on Nuclear Science

JF - IEEE Transactions on Nuclear Science

IS - 5

ER -

RADAI: A Large-Scale Realistic Dataset for Radiation Detection Algorithm Development

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this