Muon Imaging for Illicit Cargo Detection: A Simulation-Based Study

TL;DR

This study demonstrates through simulations that muon tomography can effectively detect illicit drugs in cargo by combining rapid initial scans with detailed 3D imaging, achieving high discrimination accuracy and successful visualization of concealed contraband.

Contribution

The paper introduces a simulation-based evaluation of muon tomography for cargo inspection, proposing a two-stage protocol that balances speed and resolution for detecting hidden contraband.

Findings

Random Forest classifier achieved AUC of 0.9969 in rapid scans.

Extended scans enabled visualization of concealed cocaine with 3 sigma significance.

Muon tomography is feasible for real-world cargo inspection applications.

Abstract

This study evaluates the potential of muon tomography as a non-invasive technique for detecting concealed illicit drugs in cargo, based on detailed simulations performed using the GEANT4 toolkit. A combined analysis of muon scattering and absorption data was employed to enhance material discrimination, with a focus on realistic smuggling scenarios involving cocaine hidden within legitimate cargo. A two-stage inspection protocol is proposed to balance detection speed and resolution. In the first stage, a rapid scan lasting ~ 60 seconds is used to identify anomalous scattering and absorption rates, without requiring full tomographic reconstruction. Receiver Operating Characteristic (ROC) analysis of rapid scan data revealed that the Random Forest classifier achieved an area under the curve (AUC) of 0.9969, while the multivariate normal likelihood model attained an AUC of 0.9977, both indicating excellent discrimination between benign cargo and smuggled contraband. Upon detection of anomalies, an extended scan ~30 minutes is initiated to enable high-resolution three-dimensional imaging for accurate localization and identification of hidden materials. Simulation results demonstrate that, with a detector spatial resolution of 1~mm (FWHM), concealed contraband such as cocaine can be detected with approximately 3 sigma statistical significance during the rapid scan phase. In extended scans, cocaine packages concealed within banana boxes were successfully visualized and automatically identified using clustering algorithms such as DBSCAN applied to the tomographic reconstruction. These findings confirm the feasibility of cosmic-ray muon tomography as a passive, safe, and effective approach for contraband detection in real-world cargo inspection applications.