Molecular Quantification with Spectral Photon-Counting Computed Tomography for New Diagnostic Imaging Applications

X-ray computed tomography (CT) allows scientists and clinicians to non-invasively visualize deep tissue structures within the human body in three-dimensions and at high spatial resolution. However, conventional CT is generally limited to anatomic imaging rather than molecular or functional imaging due to limitations in spectral resolution and soft tissue contrast. Spectral photon-counting computed tomography (PCCT) enables multi-energy X-ray imaging due to advanced, energy-discriminating photon-counting detectors. Material decomposition algorithms combined with spectral PCCT image acquisition transform CT into an anatomical and molecular imaging modality. However, the accuracy of image-based material decomposition is inherently limited by underlying image quality. Image denoising can be used to reduce image noise and artifacts, but the effects on quantitative material decomposition have not been investigated. Additionally, clinical diagnostic imaging and preclinical research with spectral PCCT remains limited to applications that are already possible with conventional CT. The experimental demonstration of transformative imaging capabilities is needed to accelerate clinical and preclinical adoption of spectral PCCT. Therefore, the overall objective of this dissertation was to demonstrate computational advances and experimental capabilities of spectral PCCT in high-impact applications that are currently impossible or severely limited with current diagnostic imaging modalities.

Deep learning based denoising algorithms were demonstrated to improve the accuracy of quantitative material decomposition in spectral PCCT images while balancing preservation of local image detail that is lost by conventional denoising filters. Among potential applications, spectral PCCT enabled contrast-enhanced detection of model breast microcalcifications of varying size and hydroxyapatite (HA) density, that were targeted by bisphosphonate functionalized gold (Au) nanoparticles, using the measured Au:HA ratio from image-based material decomposition. Spectral PCCT further enabled nondestructive, longitudinal, quantitative monitoring of the degradation kinetics of hydrogel scaffolds covalently linked with Au nanoparticles and was validated by concurrent measurement with gold-standard techniques. Additionally, geometric features of complex, multi-material 3D-bioprinted hydrogel constructs were able to be discriminated by multicontrast (Au and Gd₂O₃) decomposition in spectral PCCT imaging of phantoms and after implantation in mice, despite exhibiting similar overall X-ray attenuation. Finally, spectral PCCT enabled simultaneous quantification of bone mineral and lipid content within trabecular bone cores using image-based material decomposition calibrated by a custom oil/water imaging phantom. Thus, spectral PCCT enabled molecular quantification and new imaging capabilities for innovative and clinically relevant applications which are currently not possible with other current clinical imaging modalities.