ABSTRACT
A main challenge in x-ray µCT with laboratory radiation derives from the broad spectral content, which in contrast to monochromatic synchrotron radiation gives rise to reconstruction artifacts and impedes quantitative reconstruction. Due to the low spectral brightness of these sources, monochromatization is unfavorable and parallel recording of a broad bandpath is practically indispensable. While conventional CT sums up all spectral components into a single detector value, spectral CT discriminates the data in several spectral bins. Here we show that a new generation of charge integrating and interpolating pixel detectors is ideally suited to implement spectral CT with a resolution in the range of 10 µm. We find that the information contained in several photon energy bins largely facilitates automated classification of materials, as demonstrated for of a mouse cochlea. Bones, soft tissues, background and metal implant materials are discriminated automatically. Importantly, this includes taking a better account of phase contrast effects, based on tailoring reconstruction parameters to specific energy bins.
