Multi-Material Discrimination Using Photon Counting Spectral Computed Tomography

Quantitative volumetric mapping of multiple materials with spectral computed tomography (CT) has applications in many areas including biomedical imaging, defense and security, geophysical imaging of rock composition and in materials and chemical imaging. Dual or multi-kVp x-ray exposure when using an energy-integrating detector has been proposed and demonstrated in the past for biomedical imaging. X-ray dose and imaging time, along with insufficient spectral separation, limits dual and multi-kVp applications and their ability for accurate quantitation. When using photon-counting spectral detectors (PCDs), some of these limitations can be overcome. However, low dose and computationally efficient mechanisms to yield volumetric maps of more than two or three materials remain as a significant challenge. Recently, our group has proposed a multi-step method for virtual discernment between the materials of an object termed material decomposition for short. Experimental implementation of this method adds new challenges including reliable detector spectral corrections. This work presents initial simulation studies, experimental validation and detailed methods to successfully implement this multi-material decomposition technique. Here we show examples with virtual separation of up to six materials in simulations and five materials in experiments on our benchtop spectral CT system. For comparison, a conventional single-step decomposition was also performed on the same synthetic and experimental data. Results show a significant reduction in decomposition errors with low noise over the single-step approach. Finally, a biological specimen of a chicken heart was injected with tantalum and gadolinium (likely candidate contrast agent materials) and multi-step decomposition was also successfully conducted on this sample. These studies offer validations required for robust utility of the method in imaging applications requiring separation of multiple materials.