Abstract
This thesis describes a series of interrelated studies that I performed to characterize two key modules of MARS spectral CT scanner: the x-ray source and the x-ray detector. Characterizing and optimizing the outputs of these two modules are steps towards utilizing the full advantages of the multi-channel spectroscopic imaging from the MARS spectral CT to enhance material resolution.
I contributed to developing a parameterized semi-analytic x-ray source model. The main body of this work and its utilization in other MARS projects have been disseminated through a submitted paper, a published paper, and three conference proceedings. I also developed a method for profile assessment and characterization. This work has given rise to a submitted paper and a conference proceeding. I contributed to characterizing the Medipix3RX performance at the pixel level through two distinct projects including characterization and calibration of per-pixel energy response, and pixel classification based on time analysis. The industrial importance of this research outcome resulted in the filing of two patents: an algorithm for generating a pixel mask, and a method for identification and correction of unstable pixel clusters.
The parameterized semi-analytic x-ray source model provides on- and off-axis x-ray spectra in the diagnostic imaging energy range of 30-120 kVp, across the field of view of the MARS spectral CT. This development was in response to a need for accurately providing the energy and position of the incident photon to a future polychromatic-based material reconstruction technique in the MARS group. Considering the polychromatic structure of the beam in data processing will enable us to make better use of the spectroscopic information that is available in the Medipix3RX ASIC.
The beam profile assessment and characterization method was motivated by the instabilities of the beam profile observed in a poorly-calibrated MARS spectral CT prototype. To monitor the beam profile, several beam profile properties were measured in the MARS spectral CT and compared with the profiles of the x-ray source model. This method is capable of identifying temporal or spatial fluctuations in the beam profile. The accurate offset parameters provided by this method are then used to calibrate the MARS source model for each scanner. This, therefore, enables us to accurately express the incident photons of the x-ray beam for the future spectral reconstruction techniques.
Characterization and calibration of the individual pixel energy response addressed degradation of spectral fidelity of the images caused by inter-pixel variation of energy response. The significance of the proposed method is to measure and calibrate per-pixel energy response when the MARS spectral CT operates in a standard acquisition mode without using any additional equipment. A by-product of this system is the measurement of the unequal effective pixel area, which can be used in the study related to sensor layer manufacturing.
A pilot study of pixel classification based on temporal pixel behavior was also conducted to an improved method. This method is capable of identifying malfunctioning and slow-drift pixels and masking them out from the data processing chain. Because of the precise threshold criteria were chosen in this method, it is unlikely to label a flickering pixel as a well-behaved pixel. This method significantly improved the signal-to-noise ratio of the reconstructed image. Furthermore, a group of malfunctioning pixels was identified as behaving correlatively. The overall response of all pixels can be treated as a well-behaved pixel, enabling reliable utilization of the low-grade sensors.