|dc.description.abstract||This thesis investigates the potential for quantifying soft tissues in preclinical studies by spectral computed tomography (CT) using the Medipix3 spectroscopic photon counting detector. Currently available methods for characterizing the composition of excised specimens and small animal models are insensitive, expensive or destructive. Quantitative imaging by spectral CT would provide a convenient alternative technique for preclinical studies of human diseases such as atherosclerosis and the metabolic syndrome. The work is presented as a series of interrelated studies describing the technical evaluation of the Medipix All Resolution System (MARS) spectral CT system, and the subsequent development and validation of a quantitative spectral analysis method.
The MARS-CT system and components were characterized in three stages. Firstly, the performance of a prototype MARS x-ray camera incorporating a single Medipix3 was assessed. High quality images were obtained in single pixel mode, but, in charge summing mode image quality was severely degraded by electronic instabilities and biases. Secondly, methods for calibrating the global DACs and equalizing the pixel thresholds of a quad Medipix3 array were developed and tested. These methods enabled the Medipix3 array to operate as a homogeneous large area imaging device. Thirdly, a MARS-CT system, incorporating a MARS camera with Medipix3 and silicon sensor layer, was evaluated and found to give acceptable performance for preclinical imaging of soft tissues.
A reconstruction domain material decomposition method for quantifying soft tissue components was developed and validated. Calcium, fat and water components were successfully quantified in mouse and atheroma equivalent phantoms. The material decomposition method was then validated in a preclinical study of excised mice livers. The fat mass fractions of twelve wild type and twelve transgenic mice were quantified by spectral CT and the results compared with those of biochemical analysis.
In conclusion, this thesis has developed and validated a new reconstruction image domain analysis method and has shown that quantitative soft tissue imaging of excised specimens and small animal models by spectral CT is feasible. It is a strength of the analysis method that the responses of arbitrary detector types are easily accommodated. The method is thus applicable to other areas such as functional imaging with high-Z nanoparticles.||