Automated high accuracy, rapid beam hardening correction in X-Ray Computed Tomography of multi-mineral, heterogeneous core samples

Romano, Carla and Minto, James M. and Shipton, Zoe K. and Lunn, Rebecca J. (2019) Automated high accuracy, rapid beam hardening correction in X-Ray Computed Tomography of multi-mineral, heterogeneous core samples. Computers & Geosciences, 131. pp. 144-157. ISSN 0098-3004 (https://doi.org/10.1016/j.cageo.2019.06.009)

[thumbnail of Romano-etal-CG-2019-Automated-high-accuracy-rapid-beam-hardening-correction]
Preview
Text. Filename: Romano_etal_CG_2019_Automated_high_accuracy_rapid_beam_hardening_correction.pdf
Accepted Author Manuscript
License: Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 logo

Download (3MB)| Preview

Abstract

X-ray Computed Tomography scanning is an innovative procedure that allows representing the internal structure of samples. Among its several purposes, X-ray CT is widely used for investigation of petrophysical properties of porous media. To provide accurate results, it is necessary to have high quality scan images, free of artefacts. One of the most problematic artefacts is beam hardening, which, in cylindrical shapes, increases the attenuation values with increasing distance from the centre. Until now, no automatic solution has been proposed for cylindrically-shaped cores that is both computationally feasible and applicable to all geological media. A new technique is here introduced for correcting beam hardening, using a linearization procedure of the beam hardening curve applied after the reconstruction process. We have developed an automated open source plug-in, running on ImageJ software, which does not require any a priori knowledge of the material, distance from the source or the scan conditions (current, energy), nor any segmentation of phases or calibration scan on phantom data. It is suitable for expert and non-expert use, alike. We have tested the technique on μCT scan images of a plastic rod, a sample of loose sand, several heterogeneous sandstone core samples (with near-cylindrical shapes), and finally, on an internal scan of a Berea sandstone core. The Berea core was also scanned using a medical X-ray CT scanner with a fan-beam geometry, as opposed to a cone beam geometry, showing that our algorithm is equally effective in both cases. Our correction technique successfully removes the beam hardening artefact in all cases, as well as removing the cupping effect common to internal scans. For a Berea Sandstone, with a porosity of 20%, porosity calculated using the corrected scan is 20.54%, which compares to a value of 14.24% using the software provided by the manufacturer.