Material classification using basis material decomposition from spectral x-ray CT

Abstract

Purpose: Spectral CT exploits advanced MultiX ME 100 photon counting detectors (PCD) to measure a material’s spectrally resolved linear attenuation coefficient (LAC) with the simultaneous spectral acquisition at multiple energy thresholds. We present a method for material classification using spectral x-ray Computed Tomography (CT). Methods: The method employs a basis material decomposition model and estimates the effective atomic number (Zeff ) from the spectral LAC measurements. Basis material decomposition builds on the fact that the LAC of any material can be well approximated by a linear combination of equivalent thicknesses of basis materials, with known and typically Zeff values at the extremes of the relevant Zeff range. Spectral distortions of the energy spectrum due to the physical interactions between photons and the multi-energy-bin PCD such as charge sharing and photon pileup are corrected by a spectral correction algorithm. The validation of the method has been performed with experimental data acquired with a custom laboratory instrument for spectral CT, examining “real life” phantoms with materials in the range of 6 ≤ Zeff ≤ 15. Results: With optimized parameters the data was collected from 12 projections and rebinned from 128 into 15 energy bins, two basis materials were used for the decomposition. In that case the classification method gives a relative deviation of 2.4% for Zeff , while this deviation is 5.2% when spectral correction is not used. Conclusion: The classification method is now ready for use in security screening where modern spectral CT systems are employed.