A Multi-Material Approach to Beam Hardening Correction and Calibration in X-Ray Microtomography

PhDX-ray microtomogaphy is a non-clinical, non-destructive, and quantitative technique for determining three-dimensional mineral concentration distributions in variably radiolucent samples with a spatial resolution on the micron scale. For reasons of practicality, particularly for longterm studies, it is often not possible or desirable to utilise a monochromatic X-ray source and so microtomography using a conventional impact-source X-ray generator to produce a polychromatic photon beam is carried out instead. The use of photons of multiple energies causes the production of projection artefacts arising from preferential absorption, which impair the greyscale accuracy of the resulting reconstruction and the material concentration measurements that are derived from the linear attenuation coefficients (LACs). The purpose of the project described in this thesis is to identify weaknesses in the current method of beam hardening correction and to develop and test a tomographic calibration and projection processing method which may demonstrably improve the quality of current beam hardening correction methods as used with the MuCAT microtomography equipment, which provides a worldclass impact-source microtomography research and production facility at Barts and The London School of Medicine and Dentistry. An overview of the physical basis of X-ray computed tomography and X-ray microtomography is given from first principles, and examples of quantitative applications of the techniques, which generally depend on accurate reconstruction of linear attenuation coefficient values, are discussed. The major sources of artefacts in X-ray microtomography are discussed, particularly those with a direct impact on reconstructed linear attenuation coefficient values. Beam hardening is identified as an error source of particular interest, with secondary research on the effects of any beam hardening correction method on the severity of Compton scatter artefacts, and a critical review is carried out of historical attempts to reduce or mitigate these artefacts, particularly the single-material parameter-optimisation approach in service at the beginning of the research project. A ‘carousel’ test piece comprising multiple attenuators of multiple materials along with attenuation optimisation software based on varying multiple system parameters in order to extend the functionality and usability of the existing correction model, and qualitative results have so far been gathered suggesting the use of this system over the pre-existing attenuation wedge and parameter-optimisation method. A study of the effects of tuning the photon energy to which calibrations are made is carried out, showing improved linear attenuation coefficient recovery at a higher energy than was previously believed to be optimal, and a significant effect arising from X-ray generator target evaporation leading to spatial changes and time-dependence of the target thickness parameter is measured, suggesting that automated calibration as a standard part of the measurement process is required. A stability experiment is carried out using this method in order to examine the possibility of inconsistency resulting from ageing of the filament cathode, which is found not to significantly impact the quality of results. An immersion tank is developed in order to ensure beam hardening correction validity in the case of dual-material specimens where a radiodensity-matching fluid can be provided and the sample is suitable for immersion. Experimental comparison using a commercial beam hardening calibration device as the specimen reveals significantly improved hydroxyapatite concentration measurement recovery. An in-scatter experiment was carried out on the immersion tank, and it was found that there was a significant scatter contribution when the tank was filled in the case where the sample thickness is much less than the tank thickness. Proposals are presented for further work to improve reconstruction quality through of scatter reduction techniques in impactsource microtomographic systems, and to utilise the immersion tank for in situ chemical erosion experiments. The effects of the improvements to the beam hardening process are demonstrated using a biological specimen to demonstrate qualitative changes in reconstruction, particularly in improved dark levels surrounding the specimen. A second experiment is carried out in order to test the reproducibility of results, which is found to be improved by approximately four times over the same dataset corrected using the pre-existingbeam-hardening calibration methodEngineering and Physical Sciences Research Council grant number EP/G007845/1