Development of energy selective techniques in x-ray computed tomography

Abstract

X-ray micro computed tomography (Micro-CT) has emerged as a powerful tool in petroleum industry for non-destructive 3D imaging of rock samples, that offers micron-scale spatial resolution images of the distribution of porosity, permeability, and fluid phases of the specimens. Micro-CT obtain the radiographic projections of a sample at different angles and use a mathematical procedure to reconstruct a 3D tomogram of the sample's X-ray attenuation coefficients. Through my thesis, the aim was to investigate and improve two main issue which micro-CT suffers from: 1) beam hardening (BH) artefacts and, 2) the requirement of material characterisation. This thesis contributes in addressing these fundamental issues by providing the "energy selective techniques" as follows. Chapter 1 provides an overview of the basics of tomography including physics of X-rays and energy dependent form of attenuation coefficient. Chapter 2 reviews the BH effects and the existing correction methods, followed by a brief review of the material characterisation methods. Chapter 3 assess the accuracy of five different linearisation BH correction models including polynomial, bimodal, power law, cubic spline and zero-order using the sample that have been imaged at ANU CT facility by measuring the BH curves directly and remapping the inverse of the models to data. Chapter 4 is based on a published conference proceeding paper [1] that applies the power law BH correction method of chapter 3 to correct the artefacts of specimens composed of concentric cylinders, e.g., a rock core within a container. Chapter 5 is based on a published paper in the Journal of Applied Physics [2] that uses dual-energy CT and the Alvarez and Macovski [3] transmitted intensity (AMTI) model to estimate the maps of density (rho) and atomic number (Z) of mineralogical samples. In this method, the attenuation coefficients are represented in the form of the two most important interactions of X-rays with atoms that is, PE and CS. This enables material discrimination as PE and CS are respectively dependent on Z and rho of materials [3]. Chapter 6 implements two simplified form of the full model of chapter 5: 1) Alvarez and Macovski polynomial (AMP) model [3], Alvarez and Macovski presented the full model but used a polynomial simplified form of it to estimate rho and Z of materials, 2) Siddiqui and Khamees (SK) model [4] that simplified the attenuation model, by assuming two monochromatic radiations. Chapter 7 presents a method to estimate the properties of sample materials from measurements of transmitted intensity and its statistical variance (TIV model). The method only requires single energy imaging, i.e., eliminates the need for requirements of dual-energy imaging for AMTI method and its simplified forms. The registered intensity on the detector is proportional to a form of "average" energy of detected quanta of X-ray spectra. The variance images can serve the same purpose as the higher energy information required in dual-energy imaging. Chapter 8 modified the TIV model of chapter 7 to apply it directly for BH correction without necessarily estimation of the properties of sample materials. The chapter also presents a simplified form of TIV model (STIV) that normalises the average intensity image. Show more