CT is one of the most important of the noninvasive imaging modalities, providing 3D representations of the x-ray attenuation coefficient with submillimeter spatial resolution.
CT is one of the most important of the noninvasive imaging modalities, providing 3D representations of the x-ray attenuation coefficient with submillimeter spatial resolution. The main limitations of CT relate to soft tissue contrast resolution and quantitative material differentiation. Administration of iodinated contrast material can improve soft tissue contrast, as well as enable vessels and parenchymatous organs to be visualized. Quantitative material differentiation requires more novel techniques.
X-ray attenuation for materials with a high atomic number (Z), such as iodine or calcium, depends strongly on beam energy during CT scanning. Only minor attenuation changes relative to water are observed in soft tissues with similar attenuation coefficients when the beam energy is altered. This is mainly due to the photoelectric effect, which depends strongly on energy and atomic number. The Compton effect is almost independent of photon energy in the energy range used for diagnostic CT. Using different x-ray spectra thereby allows high-Z materials to be differentiated, leading to new applications for CT.
Dual-energy methods were suggested as early as 1976 for enhancing the soft tissue contrast in CT.1,2 The idea was to determine the coefficients of photoelectric absorption and Compton scatter attenuation independently using two CT measurements with different tube voltages. An alternative, material decomposition, technique was also proposed. That technique used alternate sets of base functions, for example, the attenuation functions of water and bone. The latter was finally implemented into a commercially available CT scanner (Siemens Somatom DR, 1983 to 1987).3
Technical limitations of CT scanners available at this time kept dual-energy scanning from becoming a routine clinical tool. The main problems were unstable CT numbers and long scan times. In addition, x-ray tubes could not provide currents at sufficiently low voltages to achieve an output of photons adequate for reliable dual-energy imaging. This required similar levels of image noise in both image data sets, something that was not achievable at that time.4
These limitations were overcome in 2006 with the introduction of dualsource CT (DSCT) by Siemens Medical Solutions. DSCT technology solves problems of data registration and patient movement that were previously associated with dual-energy imaging. With DSCT, both low- and high-voltage data sets are obtained simultaneously. Synchronous data acquisition is particularly important in contrast-enhanced scans, where temporal changes in contrast enhancement happen extremely quickly. Any temporal offset between the two acquisitions would make dualenergy postprocessing impossible.
Early approaches to dual-energy imaging focused on bone density measurement, 5 detection of intrathoracic calcifications,6 differentiation of urinary calculi,7 and assessment of hepatic iron load.8 Brain tumor differentiation was also evaluated, but with only limited success.9 None of these techniques reached clinical relevance.
DSCT has paved the way for a broad variety of dual-energy clinical applications. Some simple but powerful tools can enhance visualization of baseline images. The dual-energy capability can also be used for two different types of clinical application: material differentiation and semiquantitative depiction of iodine distribution.
A dual-energy CT scan provides end users with two image stacks that have been acquired at different kV settings, 80 kV and 140 kV, for example. Each image series is acquired at a lower dose setting than a standard scan to keep radiation exposure to an acceptable level. The resulting images consequently have more image noise and different contrast characteristics than images from a routine CT scan. This problem may be overcome by mixing the images from each series; that is, using data from the 80 kV and 140 kV scans together. Image noise is reduced to “normal” levels and, depending on the mixing ratio, contrast characteristics can be optimized, too.
Attenuation and image noise can be optimized on a case-by-case basis for routine data evaluation. Clinical CT images are usually acquired with a 120 kV tube current. A weighting factor that uses 30% of the information from the 80 kV images and 70% from the 140 kV images produces image characteristics that are close to a 120 kV spectrum. A 50% mixing ratio, which is similar to a 110 kV spectrum, appears to be optimal for assessing CT angiography (Figure 1).10 The transmission of 80 kV photons can be so low in heavy patients that image quality is severely affected. A greater contribution from the 140 kV images may be needed in such cases.
The technology described above can be improved further by using nonlinear data blending. Starting with a predefined mixing ratio of 30% from 80 kV images, for example, the mixing ratio may be modified voxel-wise and nonlinearly for individually defined attenuation ranges. More of the 80 kV image information may be used in a selected attenuation range, with the remaining image information displayed at the routine 120 kV spectrum.
This technique is particularly helpful when looking at contrast-enhanced images. Iodine shows a higher attenuation at 80 kV when compared with 120 kV. Nonlinear attenuation-based blending of the image information may be used to enhance the contrast in a relevant region of interest.
Different materials can be distinguished using a modification of the base material decomposition algorithm. This works particularly well for heavy ions like iodine, which are known to show a particularly strong increase in attenuation with decreasing tube voltage. The most widespread clinical application using this feature is bone removal (Figure 2). Voxels containing calcium are separated from those containing iodine based on the different attenuation characteristics of calcium and iodine at different tube voltages.
This technique is applied in CTA to achieve angiographic views of the examined volume. The dual-energy bone removal technique can be combined with conventional bone removal algorithms to improve results further. Overall radiation exposure to patients is reduced because only one CT scan is needed, rather than two (one precontrast and one after contrast administration). At our institution, the application has proved particularly helpful during assessments of the carotid arteries and the circle of Willis close to the skull base. Conventional bone removal algorithms used with single-energy CT are seldom adequate in these areas.
The same technique can be used to separate calcified plaques from the contrast- enhanced vessel lumen, helping to estimate the true severity of stenosis.
Applications involving two-material decomposition are not restricted to the separation of calcium and iodine. The increased carbon content of ligaments and tendons, for example, may be used to separate these structures from calcified bone or muscle in exactly the same manner. This could be helpful when assessing soft tissue involvement in trauma, or when evaluating tendons involved in suspected carpal syndrome.
Uric acid may be used as a base material when imaging gout. The selection of iron as a base material for brain imaging could help identify cerebral hemorrhage, and perhaps even determine the age of the bleed.
Semiquantitative assessment of iodine content per voxel is achieved using socalled three-material decomposition. The iodine content of every voxel is determined from the kV-dependent attenuation differences for the two main components of the relevant tissue and for iodine. The materials chosen will depend on the target region. Liver imaging may involve the decomposition of soft tissue, fat, and iodine, while in the lung it may be soft tissue, air, and iodine that are considered. Known attenuation characteristics, at 80 kV and 140 kV, for example, are then used to assign the proportions of all three materials at each voxel.
A number of algorithms have been established for semiquantitative iodine visualization. One that is used routinely is the computation of a virtual unenhanced CT image from a corresponding iodine-enhanced image. This technique replaces the need for precontrast CT scanning when looking for urinary calculi, for example, or differentiating liver tumors or assessing aortic aneurysms on contrast-enhanced CT (Figure 3).
This technique could, however, potentially be limited by an insufficient differentiation of iodine from calcium. Calcium is reliably removed from images in the material differentiation technique described previously. That is not the case here. This technique should consequently be reserved for assessments of soft tissue or organs such as the lungs. Material separation algorithms should be used in addition to the semiquantitative visualization algorithms if calcium has to be separated from iodine.
The depiction of iodine is particularly interesting in the lungs. The iodine image can help detect even small perfusion deficits that may be caused by pulmonary embolism. The use of xenon during contrast-enhanced dual-energy CT scanning of the lungs makes it feasible to perform a combined perfusion/ ventilation examination.
Another development is the computation of monoenergetic CT images at different keV settings from dual-energy images that have been acquired at two different tube voltages. This dual-energy postprocessing algorithm may be used to analyze iodine content in the myocardium. Dual-energy cardiac DSCT comes at a price, though. Improvements to temporal resolution seen in other cardiac applications are not realized because the second tube is needed to achieve the energy resolution. Cardiac DSCT with dual imaging does, nonetheless, still provide the quality and temporal resolution of a routine 64-slice cardiac CT scan.
A broad variety of dual-energy applications are now available for routine clinical use. Early studies have confirmed their feasibility.11 Little has been published on the use of dual-energy CT in daily practice, though.
Initial studies focused on the detection and differentiation of urinary calculi using virtual unenhanced CT scans derived from contrast-enhanced images. The sensitivity of the virtual unenhanced CT images for the detection of urinary stone disease was 83% and the specificity 100% when compared with actual unenhanced CT images.12
The only stones missed in this study were in obese patients. This was due to increased image noise that limited the postprocessing of data. Urinary stones less than 5 mm in diameter are known to pose problems in general as well, and some of these were missed on the dual-energy images.
Some of these limitations have now been overcome by the introduction of cross-scatter–reduction and advanced noise-reduction techniques. These techniques can lead to significant improvements in the quality of virtual unenhanced CT images.
Researchers in a separate investigation used DSCT to differentiate renal stones containing uric acid from other types of urinary calculi, including calculi composed of cystine from struvite stones.13 Calculi composed of uric acid or cystine are almost always treated conservatively. Thus, a reliable differentiation of different urinary calculi is relevant to therapy. This work indicates the potential of dual-energy imaging in uroradiology.
Dual-energy CT has also been considered as a way of detecting myocardial perfusion abnormalities in suspected coronary artery disease. One study has shown this technique to have 84% sensitivity, 94% specificity, and 92% accuracy for the detection of myocardial ischemia when compared with SPECT.14
Dual-energy ventilation CT using Xenon has been shown to be feasible.15 Further studies are needed to assess the benefit of this method as a routine clinical tool.
Other potential applications for dual-energy CT that are being investigated include musculoskeletal imaging16 and the improvement of tissue characterization in virtual autopsy.17
Dual-energy imaging is providing new opportunities and challenges for CT. The ability to differentiate iodine from calcium, improving the reliability of bone removal in clinical imaging, is of particular value. Published data on cardiac perfusion imaging and lung ventilation CT using dual-energy techniques indicate the future potential of dual-energy imaging techniques.