Abstract
The quality of the spectral images acquired using a spectral imaging system relies on the performance quality of its energy discriminating detector. This thesis reports on the development of pragmatic techniques and theoretical models for characterization and performance evaluation of Medipix3RXbased detectors used in the MARS spectral scanner. My research facilitates detector optimizations and suggests algorithmic compensation techniques to advance the spectral imaging development. The impact of my work is reflected in the outcomes of my research and those of my colleagues.
An efficient pixel classification and masking technique was developed and implemented in the MARS scanner. The outcome of this study has already been used by the MARS users over the past several years. It has reduced image noise and artefacts, therefore improved classification and quantification of materials. This thesis also reports on development of techniques for detector energy calibration. An automated energy calibration procedure was investigated through measurement of x-ray fluorescence signals as the physical energy references. Theoretical models were developed to provide optimizations for the scan geometry. Another calibration technique with a particular focus on calibrating the energy response on a per-pixel basis was developed. This study quantified the energy response dispersion across the detector array. A noticeable energy resolution improvement was achieved in the experimental evaluation of the technique. I also developed statistical models to quantify the degradations caused by the pileup effect. The efficiency of the models in simulating the experimental measurements was validated with the detector operation under high flux rates. By knowing the contribution of pileup to count loss and energy distortion, future reconstruction calculations will cope with these distortions and thus allow higher flux rates and in turn shorter imaging times.
To summarize, this thesis has developed strategies to take full advantage of spectral capabilities of the MARS camera and improve spectral image quality.