(a) Schematic of the experimental set-up at the Munich Compact Light Source (MuCLS): X-rays are produced at the interaction point of laser photons stored in the optical cavity and relativistic electrons circulating in the storage ring. These propagate in a vacuum tube to the experiment at a source-to sample distance of 15.5 m. The detector is located at a distance of 16 m from the interaction point. (b) X-ray spectrum with the peak energy tuned to 33.49 keV measured with a KETEK AXAS-D detector, shown together with the iodine attenuation coefficient. (c) Photograph of the experimental set-up showing the iodine filter in front of a porcine coastal arch and a porcine heart.

First, unfiltered and iodine-filtered images are recorded and dark current, flatfield and flux corrected. To obtain two images only containing the high- or low-energy part of the spectrum, one needs to eliminate the high-energy part in the iodine filtered image and the low-energy part in the unfiltered image. The iodine-filtered image is weighted so that its intensity corresponds to the low-energy part of the unfiltered image, which results in the weighted iodine-filtered image (d). Subtracting image (d) from the unfiltered image (a), the high-energy image (e) is obtained, as the low-energy parts of the spectra are identical and fall away in the subtraction. To obtain the low-energy image, the high-energy part of the weighted iodine-filtered image has to be eliminated. Therefore, the unfiltered image (a) is weighted so that it corresponds to the remaining high-energy part of the weighted iodine-filtered image. The weighted unfiltered image (c) is then subtracted from the weighted iodine-filtered image (d), yielding a low-energy image (f), where the high-energy contribution is completely cancelled out. Finally, the difference image is obtained by logarithmically subtracting the low-energy image from the high-energy one.

Raw input data (a,b) and processed output image (c) associated with K-edge subtraction (KES) imaging, for a phantom containing aluminum and iodine. (a) Unfiltered reference, (b) iodine- filtered image, and (c) KES image. Both, (a) and (b), still exhibit aluminum overlaying the tube with iodine contrast agent. KES-imaging (for details see text) provides the aluminum-free pure iodine contrast image (c). The ROIs chosen for the calculation of the CNRs of iodine (1) and aluminum (2) are shown in image (a) and (c). The gray scales for the unfiltered and filtered images display the relative intensity / transmission of the X-ray beam, while the gray values in the KES image c show the negative differences in the absorption.

K-edge subtraction imaging of the left anterior descending artery of a porcine heart with injected iodine contrast agent and inhomogeneous background structures: First row: In both, the unfiltered (a) and the iodine-filtered (b) image, air artefacts and inhomogeneous absorption make the identification of small blood vessels difficult. In the KES image (c), only the contrast agent remains visible; Second row: The overlying bone structures make the identification of the small blood vessel (marked by the red arrow) impossible in images (d) and (e). Only KES imaging, combined with an energy correction, restores contrast for these blood vessels. Although in areas of lower counting statistics the image quality is compromised in the KES image (f) by noise, the visibility of the small vessels is increased. The ROIs for the calculation on the CNR of the blood vessel (1) and the bone (2) are shown in the unfiltered (d) and difference (f) image. The gray scales of the images were chosen such that the contrast for the coronary arteries is optimized. The gray scales for the unfiltered and filtered images show the relative intensity of the X-ray beam, while the gray values in the KES images show the negative differences in the absorption.