Physics-based calibration of photon-counting CT

Sammanfattning

Photon-Counting Computed Tomography (PCCT) represents a transformative advancement in medical imaging, offering unparalleled capabilities in spatial resolution, spectral differentiation, and dose efficiency. By leveraging photon-counting detectors, PCCT overcomes the limitations of conventional energy-integrating systems, enabling direct photon quantification and discrimination across energy bins. This thesis explores the underlying physics, technological developments, and clinical applications of PCCT, with a focus on addressing key challenges such as pile-up effects, energy bin optimization, and material decomposition accuracy.This work presents a successful implementation of a physics-based calibration method, enhanced by incorporating an exponential-polynomial term to account for variations in detector elements and a semi-nonparalyzable transformation of photon counts to combat pile-up effects. These improvements are achieved with a minimal number of free parameters, which are determined through a forward model trained on a limited set of 10 data points and evaluated on a larger dataset comprising 67 data points. Future work should aim to assess the performance of these models in the image domain and extend the approach to include a third basis material, facilitating improved characterization of contrast agents for advanced imaging applications.