Dual-energy CT makes inroads in aortic and pulmonary studies

Conventional single-energy CT imaging results in an anatomic depiction of the imaged area based on depiction of differences in physical density. Dual-energy CT imaging differentiates structures of a similar density with different elemental compositions based on differing attenuations at different photon energies. Hence, dual-energy imaging moves away from imaging density toward imaging elemental, or possibly even chemical, composition (see accompanying article).

The principles of dual-energy imaging were envisioned by Hounsfield.1 Although there was considerable interest in dual-energy imaging early on, the technique was soon abandoned, mostly for technical reasons.

As a result of the advances in dual-source CT scanning, it is now possible to obtain MSCT-quality images at two different energies simultaneously during a single acquisition.2,3 The first available system (Somatom Definition, Siemens Medical Solutions) consists of two acquisition devices mounted on the rotating gantry of the scanner at a 90 degrees offset; each is equipped with an x-ray tube and corresponding 64-channel detector.3 In the axial plane, one tube/detector (A) covers the entire 50-cm-diameter scan field-of-view while the second tube/detector (B) is restricted to a central 26-cm-diameter FOV in the same z-axis position. When the A and B tubes are operated at the same energy (so-called dual-source scanning), the advantage is improved temporal resolution, which is best utilized for cardiac imaging. When the A and B tubes are operated at different kilovoltages (so-called dual-energy imaging), the advantage is material differentiation.

The greatest implications for dual-energy thoracic CT imaging are likely to be improved aortic and pulmonary arterial angiographic imaging (potentially with lower contrast volume), the use of pulmonary enhancement maps, and the potential to eliminate precontrast imaging, thereby reducing radiation exposure.

Our current thoracic imaging protocol uses the following configuration:

• slice parameters: 64 x 0.6 mm;

• rotation time: 0.33 to 0.5 sec;

• pitch: 0.5 to 0.7;

• tube A: 140 kVp, mAs 56 to 80; and

• tube B: 80 kVp, mAs 370 to 480.

Using such a configuration results in an overall radiation dose of no more than 0% to 20% more than a comparable examination performed on the same scanner with a single-source technique. Due to the greater radiation efficiency of higher channel MSCT systems, this may be equal to or less than detector systems with 16 or fewer slices.4 The use of two energies set as wide apart as possible within the CT tube/detector technical constraints maximizes the differences in attenuation of different substances. Using these parameters also results in an approximately similar radiation dose per tube/detector system. This in turn results in similar noise levels in the 80-kVp and 140-kVp systems, improving the accuracy of dual-energy calculations.

IMAGE RECONSTRUCTION

There are several reconstruction and postprocessing steps unique to dual-energy CT imaging. Image generation can be broadly categorized into three phases:

• primary reconstruction and basic postprocessing;

• material-specific imaging; and

• advanced postprocessing.

Primary reconstruction and basic postprocessing are performed simultaneously in real-time at the CT acquisition workstation and result in separate 80-kVp and 140-kVp data sets and "weighted-average image" data sets (arithmetic mean of 80-kVp and 140-kVp images). "Material-specific" imaging is performed with an FDA-approved software package on a dedicated workstation. Material-specific imaging uses a proprietary algorithm called "three-material decomposition" that calculates the proportion of three different materials within the voxels of the scanned volume. Advanced postprocessing is used to further enhance visualization of anatomy and pathology.

THORACIC IMAGING

The 80-kVp and 140-kVp primary reconstruction data sets are comparable to conventional single-energy data sets and may be visualized independently for diagnostic purposes. In general, however, these images will have higher noise compared with single-energy CT images generated with a comparable overall radiation dose. Therefore, in clinical practice, the weighted average images that combine both data sets with the least noise are used for primary interpretation.

Reviewing the 80-kVp images through an area of interest, however, may be of particular benefit. The physics involved renders these images more sensitive to calcification in nodules or masses. Markedly greater iodine attenuation is demonstrated on enhanced images. This greater iodine attenuation results in improved quality of aortic and pulmonary CT angiography studies. In an ongoing study at our institution, the review of 80-kVp images significantly reduced the number of suboptimal or uninterpretable CT pulmonary angiography (CTPA) studies. In a separate group, the review of 80-kVp images made the pulmonary arterial opacification of routinely acquired dual-energy CT examinations with normal contrast volume, flow rate, and no bolus tracking equivalent to that of dedicated CTPA studies.

Another important advantage of the 80-kVp images is the potential to reduce the total volume or rate of administration of intravenous contrast media. Figure 1 is from an ongoing study evaluating the potential for contrast volume reduction demonstrating excellent CTPA image quality with only 50 mL of intravenous contrast (300 mgI/mL).

The availability of multiple imaging energies is potentially problematic for the application of dual-energy CT to perform nodule enhancement studies. The established threshold of less than 15 HU of enhancement at 120 kVp does not necessarily imply benignity at 80 kVp or 140 kVp. New thresholds for the evaluation of significant nodule enhancement may need to be developed for these energies.

Material-specific image data sets are generated by first identifying a specific, unique target substance, then generating an image in which this substance is either demonstrated alone or removed from the image. For example, from a contrast-enhanced study, an image can be generated demonstrating the distribution of iodine alone for a "virtual contrast" image; alternatively, iodine can be removed, creating a "virtual noncontrast" image.

In thoracic imaging, one potentially useful application of virtual noncontrast imaging is to evaluate the enhancement of a nodule, node, or mass when precontrast imaging was not prospectively acquired. In stent endograft imaging, dual-energy acquisition may be able to eliminate precontrast imaging, significantly reducing radiation exposure (Figure 2). Our experience suggests that most endoleaks may be appreciable on a single phase of enhancement with 80-kVp imaging due to greater iodine sensitivity.

ADVANCED APPLICATIONS

Three-material-decomposition techniques can also be applied to the lung parenchyma, yielding a color-coded map of pulmonary parenchymal enhancement. The clear advantage of these techniques is that the pulmonary enhancement map requires no modification of acquisition technique but is an additional, "free" sequence that can be automatically generated in any thoracic dual-energy study.

Advanced postprocessed images are derived from the reconstructed primary data images (80 kVp and 140 kVp), the weighted average, and material-specific images. As these are acquired from the same volume at the same time, there is no misregistration between these data sets; they may be combined in an almost infinite arrangement of simple additions/fusions or subtractions to render images in which particular areas of anatomy or pathology are depicted.

Pulmonary enhancement maps, for example, may be superimposed on a CTPA study to simultaneously evaluate the pulmonary arteries and any potential enhancement defects (Figure 3). At our institution, several additional advanced imaging sequences have been developed independently to enhance dual-energy data evaluation in a variety of pertinent clinical applications. This image manipulation may require simple DICOM editing to make the data sets compatible for subtraction or addition on a conventional workstation (Figures 4 and 5). As more of these sequences are clinically evaluated, we anticipate that these would become available soon as standard reconstructions.

References for this article are available online at DiagnosticImaging.com.

Dr. Vlahos is an assistant professor of radiology and principal investigator of dual-energy CT, Dr. Godoy is a research fellow in chest radiology, and Dr. Naidich is a professor of radiology, all at New York University Medical Center.