Increased material differentiation through multi-contrast x-ray imaging: a preliminary evaluation of potential applications to the detection of threat materials

TL;DR

This study explores combining phase-based dark-field x-ray imaging with dual-energy methods to enhance material discrimination in security scans, showing preliminary improvements in identifying threat materials over traditional techniques.

Contribution

It introduces a novel multi-contrast x-ray imaging approach that integrates dark-field and attenuation images for improved threat material detection.

Findings

Enhanced contrast between threat and non-threat materials

Improved discrimination when combining multiple imaging channels

Potential for better detection of materials with similar densities

Abstract

Most material discrimination in security inspections is based on dual-energy x-ray imaging, which enables the determination of a material's effective atomic number (Zeff) as well as electron density and its consequent classification as organic or inorganic. Recently phase-based "dark-field" x-ray imaging approaches have emerged that are sensitive to complementary features of a material, namely its unresolved microstructure. It can therefore be speculated that their inclusion in the security-based imaging could enhance material discrimination, for example of materials with similar electron densities and Z eff but different microstructures. In this paper, we present a preliminary evaluation of the advantages that such a combination could bear. Utilising an energy-resolved detector for a phase-based dark-field technique provides dual-energy attenuation and dark-field images simultaneously. In addition, since we use a method based on attenuating x-ray masks to generate the dark-field images, a fifth (attenuation) image at a much higher photon energy is obtained by exploiting the x-rays transmitted through the highly absorbing mask septa. In a first test, a threat material is imaged against a non-threat one, and we show how their discrimination based on maximising their relative contrast through linear combinations of two and five imaging channels leads to an improvement in the latter case. We then present a second example to show how the method can be extended to discrimination against more than one non-threat material, obtaining similar results. Albeit admittedly preliminary, these results indicate that significant margins of improvement in material discrimination are available by including additional x-ray contrasts in the scanning process.