Image quality evaluation and the application of spectral techniques in X-ray computed tomography

Abstract

Since its inception, X-ray computed tomography (XCT) has been increasingly adopted across various sectors, including medicine, security, archaeology, and material science. A significant challenge in its application to non-destructive evaluation (NDE) is the lack of standardization and traceability of XCT data and its quality. Additionally, the growing demands of the composite industry highlight the limitations of XCT, hindering its broader use for composite inspection. This thesis aims to address the challenges of standardisation in XCT while also exploring the potential benefits of using spectral information. Together, these two themes are proposed to meet some of the future demands of the NDE industry and to address the limitations of conventional XCT as it is currently applied. Current solutions for standardisation and image quality evaluation in XCT often specify measurements that can be made but provide little practical guidance on how to implement these solutions effectively and consistently in an industrial setting. Additionally, limited studies have been conducted to evaluate spectral methods in NDE, and even fewer have aimed to present a method that allows for widespread, tailored deployment across multiple systems and samples without incurring high costs, requiring complex calibration, or needing unrealistic prior information. In this thesis, two novel XCT methods were developed, and an investigation into the sensitivity of non-destructive imaging to detect composite damage was conducted. Firstly, a method for the image quality assessment of multi-material reference samples was developed. This process proved highly effective at automatically assessing XCT acquisition conditions and post-processing routines. With further development, it could be used to verify inspection routines and increase confidence in the analysed parts. Secondly, a method for spectral X-ray projection fusion was developed to improve contrast in weakly attenuating materials when they are in close proximity to much more attenuating materials. Radiographically, the method showed promising results by revealing structures in an object that are not discernible under a single conventional radiograph. However, optimizing the parameters for XCT was challenging, and incorrect parameter selection could suppress all material contrast, performing worse than a well-chosen linear combination of two conventional radiographs taken under two distinct X-ray spectra. Thirdly, an assessment of the sensitivity of non-destructive imaging methods was conducted to characterize in-plane (delaminations) and out-of-plane (matrix cracking/splitting) defects in open-hole carbon fibre reinforced polymer (CFRP) samples. The results indicated that information obtained during XCT scanning could be replicated through the combined capabilities of 2D planar NDE methods. Specifically, the use of a novel X-ray phase contrast imaging approach was able to detect cracking undetectable using conventional radiography, while ultrasonic C-scan visualization effectively detected delamination, which was undetectable using all 2D X-ray-based methods. Combined, these approaches provided an equivalent understanding of the damage location and extent as XCT scanning and could overcome the size limitations of XCT for large composite inspection requiring rotation.

Date of Award	31 Dec 2024
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Philip Withers (Supervisor) & Timothy Burnett (Supervisor)