Rapid simulation of X-ray transmission imaging for baggage inspection via GPU-based ray-tracing

Qian Gong, Razvan Ionut Stoian, David S. Coccarelli, Joel A. Greenberg, Esteban Vera, Michael E. Gehm

Research output: Contribution to journalArticlepeer-review

17 Scopus citations


We present a pipeline that rapidly simulates X-ray transmission imaging for arbitrary system architectures using GPU-based ray-tracing techniques. The purpose of the pipeline is to enable statistical analysis of threat detection in the context of airline baggage inspection. As a faster alternative to Monte Carlo methods, we adopt a deterministic approach for simulating photoelectric absorption-based imaging. The highly-optimized NVIDIA OptiX API is used to implement ray-tracing, greatly speeding code execution. In addition, we implement the first hierarchical representation structure to determine the interaction path length of rays traversing heterogeneous media described by layered polygons. The accuracy of the pipeline has been validated by comparing simulated data with experimental data collected using a heterogenous phantom and a laboratory X-ray imaging system. On a single computer, our approach allows us to generate over 400 2D transmission projections (125×125 pixels per frame) per hour for a bag packed with hundreds of everyday objects. By implementing our approach on cloud-based GPU computing platforms, we find that the same 2D projections of approximately 3.9 million bags can be obtained in a single day using 400 GPU instances, at a cost of only $0.001 per bag.

Original languageEnglish
Pages (from-to)100-109
Number of pages10
JournalNuclear Instruments and Methods in Physics Research, Section B: Beam Interactions with Materials and Atoms
StatePublished - 15 Jan 2018
Externally publishedYes


  • GPU-based ray-tracing
  • Inhomogeneous 3D scene
  • Model validation
  • Realistic baggage modeling
  • X-ray simulation


Dive into the research topics of 'Rapid simulation of X-ray transmission imaging for baggage inspection via GPU-based ray-tracing'. Together they form a unique fingerprint.

Cite this