InterPore2026

Mitigating Beam Hardening for Accurate Density and Atomic Number Estimation in µCT: A Dual-Energy Inversion Approach

20 May 2026, 14:35

15m

Oral Presentation (MS10) Advances in imaging porous media: techniques, software and case studies MS10

Speaker

Cinar Turhan (The University of Texas at Austin)

Description

Beam hardening (BH) artifacts in polychromatic CT and micro-CT (µCT) of geological materials hinder quantitative analysis of density and effective atomic number. This study develops and validates a dual-energy µCT workflow in 2D parallel-beam geometry to correct for BH and provide a robust link between measured attenuation and density and effective atomic number. We systematically investigate how BH is affected by material composition, particle size, and resolution, and compare the accuracy of two primary inversion models: the Basis-Vector Model (BVM) and the Parametric Fit Model (PFM).

A digital CT simulation of a phantom containing common rock minerals of varying sizes and volumetric ratios was scanned using a sequential dual-energy µCT protocol in 2D parallel-beam geometry (80/140 kV) at two resolutions. Data were processed using two distinct inversion approaches: (1) a projection-space Basis-Vector Model (BVM) that inherently corrects for beam hardening, and (2) a standard image-based Parametric Fit Model (PFM) with a standard image-based beam hardening correction. The accuracy of each method was evaluated against the known ground truth composition of the phantoms.

The projection-space BVM successfully mitigated beam hardening artifacts and resulted in physically consistent inversions. In contrast, the PFM inversion was strongly biased by uncorrected beam hardening; therefore, it had significant uncertainty in the inversion results. Incorporating a standard beam hardening correction before PFM produced non-physical parameters and inaccurate results. The BVM proved to be a more robust method for preserving the physical correlations between attenuation and material properties in heterogeneous samples.

This study provides a validated, physics-based workflow for extracting accurate, quantitative mineralogical data from lab-based 2D dual-energy parallel-beam µCT systems. By demonstrating the superiority of the Basis-Vector Model for correcting beam hardening without compromising the underlying physics, this work provides a way to extend the use cases of CT and prevent the requirement of sample-destructive complementary methods.