Dual layer detectors simultaneously acquire high and low- energy data. The data are combined to create conventional (i.e., 120 kVp) data and used to generate true conventional images. Conventional data can be reconstructed using standard techniques such as filtered back-projection or iterative reconstruction and images are equivalent to a conventional CT image obtained from a conventional scanner (e.g., Philips Brilliance iCT) with poly-energetic X-ray spectra (Fig. 3.3). This is typically the image type used for interpretation on a routine basis. Additionally, other types of images can be generated from a dual-layer detector scanner on demand. From the combined low and high energy data, photoelectric and Compton basis pair raw data is generated, which then undergoes FBP, beam hardening correction and de-noising to generate photoelectric and Compton images respectively. Depending on how these are post-processed, additional image sets are generated.

From the basis pair of images, virtual monochromatic images can be generated by the process of linear combination, as per the approach described by Alvarez and Macovski [2]. This model is accurate in the energy range between 30 and 200 keV. Virtual monochromatic images enable the synthesis of images with attenuation properties that are similar to that of an image created with a true mono-energetic beam. These images can be generated from 40 to 200 keV with the dual-layer detector CT scanner (Fig. 3.4).

Virtual monochromatic imaging exploits the energy dependence of attenuation properties of materials. For example, generation of images at energies near the k-edge of injected contrast media can be used to boost vascular enhancement. Iodine has a k-edge at about 33 keV so lower energies produce higher levels of attenuation from iodine. Similarly, the k-edge of gold is approximately 80 keV and to enhance the contrast of gold mixtures or compounds, energies of greater than 80 keV can be selected. Virtual monochromatic imaging can also be used to target portions of the energy spectrum for greater artifact reduction. Metal and beam hardening artifacts, for example, are minimized at higher energies.

Through the material decomposition process, several types of material composition images can be generated. For example, an iodine-only image represents the presence and concentration of iodine in each voxel; this image type provides a map of iodine flow (Fig. 3.5). Iodine only images are useful for evaluating perfusion of myocardium, lungs or tumors.

In a similar way, all iodine can be removed from an image, thus generating a virtual non-contrast image. The advantage of virtually generated non-contrast images from contrast-enhanced images is the elimination of data acquisition during the non-contrast phase of a multi-phase study, thus saving radiation dose. Additionally, there may be situations where non-contrast data were not acquired but a non-contrast image would aid in the characterization of a lesion (e.g., adrenal lesion, renal lesions).

From a combination of photoelectric and Compton images, images based on effective atomic number can be generated from dual-layer data (Fig. 3.6). Discrimination of tissues on the basis of effective atomic number allows a level of discrimination beyond that provided by attenuation alone. For example, distinguishing between iodinated contrast and calcification in a vessel can be challenging because both present with the same Hounsfield unit value [4]. However, an effective atomic number image depicts the material make-up of each voxel and allows differentiation between calcium and iodine. Similarly, an effective atomic number image may be useful in the characterization of non-calcified plaques with similar attenuation properties and Hounsfield Unit values.