The Hidden K-edge Signal in K-edge Imaging
K-edge imaging is commonly used for viewing contrast pharmaceuticals in a variety of multi-energy x-ray imaging techniques, ranging from dual-energy and spectral computed tomography to fluoroscopy. When looking for the K-edge signal of a specific contrast, by taking measurements either side of the K-edge, it is found that the K-edge is not always observable for low concentrations. We have also observed that the ability to see the K-edge is unit dependent - a K-edge that is not observable in computed tomography (CT) reconstructed linear attenuation units can often be made visible by converting to Hounsfield units. This paper presents an investigation of this K-edge hiding phenomenon. We conclude that if a multi-energy x-ray measurement of any K-edge material contains a signal of any other material, then there will be a positive concentration of that K-edge material below which its K-edge cannot be observed without extracting the K-edge signal through means of basis decomposition. Mathematical descriptions of this limiting minimum concentration is provided in three cases - multi-energy radiographic projection images; and reconstructed multi-energy CT images for both linear attenuation and Hounsfield units. Two important properties of this limiting concentration are provided: the minimum concentration for Hounsfield units is always strictly less than the minimum concentration for linear attenuation; and the minimum concentration for radiographic projections will typically be much larger than the minimum concentration for reconstructed linear attenuation. Finally experimental observation of this phenomenon is presented using data collected by the MARS spectral CT scanner equipped with a CdTe Medipix-3RX camera operating in charge-summing mode, where it is also shown that the K-edge can be recovered through basis decomposition.