Dual-Energy CT Images: Pearls and Pitfalls

Abstract

Dual-energy CT (DECT) is a tremendous innovation in CT technology that allows creation of numerous imaging datasets by enabling discrete acquisitions at more than one energy level. The wide range of images generated from a single DECT acquisition provides several benefits such as improved lesion detection and characterization, superior determination of material composition, reduction in the dose of iodine, and more robust quantification. Technological advances and the proliferation of various processing methods have led to the availability of diverse vendor-based DECT approaches, each with a different acquisition and image reconstruction process. The images generated from various DECT scanners differ from those from conventional single-energy CT because of differences in their acquisition techniques, material decomposition methods, image reconstruction algorithms, and postprocessing methods. DECT images such as virtual monochromatic images, material density images, and virtual unenhanced images have different imaging appearances, texture features, and quantitative capabilities. This heterogeneity creates challenges in their routine interpretation and has certain associated pitfalls. Some artifacts such as residual iodine on virtual unenhanced images and an appearance of pseudopneumatosis in a gas-distended bowel loop on material-density iodine images are specific to DECT, while others such as pseudoenhancement seen on virtual monochromatic images are also observed at single-energy CT. Recognizing the potential pitfalls associated with DECT is necessary for appropriate and accurate interpretation of the results of this increasingly important imaging tool.

Online supplemental material is available for this article.

^©RSNA, 2021

SA-CME LEARNING OBJECTIVES

After completing this journal-based SA-CME activity, participants will be able to:

• ■ Discuss the use and appearances of different image reconstructions available with DECT.

• ■ Describe the appearance of abnormalities and objects on DECT images.

• ■ List the pitfalls that are observed at DECT.

Introduction

Tremendous advances in the last decade have led to substantial evolution in dual-energy (DECT) techniques. Although the concept of DECT was described in the 1970s, the clinical implementation was only possible in the early 2000s because of substantial improvements in hardware and computational power.

DECT enables synthesis of a wide array of images from a single acquisition because of the availability of CT projection data at more than one polychromatic energy level. These reconstructions include a range of images that simulate reconstruction from an ideal x-ray beam at a singular energy level or virtual monochromatic (VMC) images. VMC images at different energy levels have different contrast-to-noise ratios that can be specified by a user for tailored clinical applications. Material decomposition or compositional analysis is performed on material density (MD) maps that selectively display or remove materials such as iodine, water, bone, and urate. In systematic investigations, these images have been shown to improve both subjective and objective diagnostic assessment and consequently have a positive effect on patient care (1,2).

Pitfalls to accurate interpretation of images are artifacts that can falsely conceal or mimic abnormalities, and they are encountered on images from various imaging modalities (3). Although some artifacts of single-energy CT are also observed at DECT, the different acquisition and image reconstruction algorithms introduce new pitfalls that are specific to DECT image datasets. Certain objects also pose interpretive challenges because of their particular appearance on DECT images. Because the use of DECT is increasing, radiologists must be aware of these pitfalls to prevent interpretive errors (4). In this review, we illustrate the pitfalls in interpretation of commonly used DECT reconstruction images.

Spectrum of Images Available with DECT

Currently six major approaches are used for DECT, including dual-source DECT, rapid kilovolt peak (kVp)–switching DECT, dual-layer DECT, split-filter DECT, and slow kVp–switching DECT (5). The dual-spin method is the sixth DECT technique that is available commercially, but it can also be used with any CT scanner to acquire CT projections at more than one energy level (5). All DECT platforms yield images that show similar characteristics, although they vary in nomenclature (Table). In the following sections, we describe the most common types of images reconstructed with DECT.

Image Types and Techniques with Three Commercially Available DECT Platforms

Kilovolt Peak Images

Kilovolt peak images are reconstructed from a polychromatic acquisition and have similar characteristics to those from conventional single-energy CT. Because dual-layer DECT is a detector-based technology, true 120-kVp images are only available with this DECT approach. Split-filter DECT can also be used to obtain 120-kVp images, albeit with differential hardening because of the use of gold and tin filters. At rapid kVp-switching DECT, 140-kVp images are available primarily to assist the technologist for immediate identification of the accuracy of the scanning coverage and contrast enhancement. At dual-source DECT, kilovolt peak images for the kilovolt peak pair used for image acquisition are also available for interpretation.

Kilovolt Peak–Equivalent Images

These images are available at dual-source DECT. Data acquired with two kilovolt peaks are used to create a weighted average or a blend of images in the image domain. Depending on the proportions used from the high–kilovolt peak and low–kilovolt peak data, different kilovolt peak levels, including 120 kVp, can be simulated.

VMC Images

VMC images are specific to DECT. They are available with all DECT platforms at multiple kiloelectron voltage levels starting from 40 keV, because the k-edge of iodine is 33.2 keV. Low–kiloelectron voltage (<65 keV) images aid in improving image contrast, whereas high–kiloelectron voltage (>100 keV) images are typically used to alleviate metal-related artifacts. VMC images at 65–75 keV are routinely used as surrogates for 120 kVp–equivalent portal venous phase images.

MD Images

Available on all platforms, these DECT images selectively display the material in question in gray scale or with color overlay. The most common clinically relevant uses are for images that selectively display iodine, urate, or calcium. These image maps “null” the soft-tissue background and highlight the material of interest. This improves the inherent image contrast and quantification capabilities. On virtual unenhanced (VUE) and virtual noncalcium (VNC) CT images, specified materials such as iodine and calcium are selectively suppressed (5). These images are created with material decomposition techniques that are used with projection CT data (projection domain) or on reconstructed CT data (image domain). Projection CT data are used with rapid kVp-switching DECT and dual-layer DECT, and the remaining DECT approaches are based on reconstructed CT data.

Although effective atomic number maps (6), electron density maps (7), and other MD images can also be generated, in this review we focus on iodine, urate, and calcium MD images, because they currently represent the most frequently used datasets in routine clinical practice. In the following sections, we discuss artifacts that are specifically observed with DECT image reconstructions. An overview of artifacts observed on different DECT image reconstructions is shown in Table E1.

Artifacts and Pitfalls of VMC Images

Low–kiloelectron voltage imaging is used to reduce or improve image contrast or reduce the dose of contrast media, and high–kiloelectron voltage images are used to alleviate beam-hardening artifacts from metals. Conventionally, low-energy images have high noise and high contrast, whereas high-energy images have low noise and low contrast. At dual-source DECT, the use of the Monoenergetic Plus application (Siemens Healthineers, Erlangen, Germany) reduces image noise at low–kiloelectron voltage, and noise is relatively steady throughout kiloelectron voltage levels at dual-layer DECT (8,9). In general, and especially with rapid kVp-switching DECT, an optimal kiloelectron voltage level must be selected for the clinical task to provide an optimal balance between contrast and noise (10,11). This level also varies according to the DECT platform used (12).

Pseudoenhancement

Pseudoenhancement is artifactual high attenuation observed because of a combination of beam hardening, partial volume averaging, scatter, or crosstalk. It is classically encountered in evaluation of small renal lesions scanned at peak parenchymal enhancement. VMC levels greater than or equal to 80 keV (13,14) have been suggested to help mitigate pseudoenhancement, because these images theoretically lack beam hardening. However, VMC images at 40–70 keV remain susceptible to pseudoenhancement that is similar to that on polychromatic single-energy CT images (14).

Out-of-Field Artifacts

When anatomy extends beyond the maximum field of view (50 cm), out-of-field artifacts occur on polychromatic single-energy CT images. They appear as areas of crescentic high attenuation at the periphery of the field of view. These artifacts can involve a substantial amount of area on the image, because the outer 2 cm of a 50-cm field of view accounts for approximately 15% of the field of view. This is relevant in large patients and in those who have not been properly centered on the scanning bed, because assessment for catheters, subcutaneous or omental lesions, hernias, and bowel abnormalities may be impeded. When they are severe, these artifacts can extend further inward and obscure the intraperitoneal anatomy (15). Out-of-field artifacts have been described on 65-keV images and MD water (iodine) images (ie, with water selectively displayed and iodine removed) that were acquired with the first-generation rapid kVp-switching DECT platform (Fig 1) (15). Nevertheless, these artifacts are ameliorated on MD iodine (water) and 40-keV VMC images. In our collective experience, out-of-field artifacts are not seen when second-generation rapid kVp-switching DECT scanners are used.

Figure 1a.

Out-of-field artifacts on images acquired with a first-generation rapid kVp-switching DECT platform in a 56-year-old man with a transverse diameter greater than 50 cm. (a) Axial 70-keV DECT image shows a crescentic area of hyperattenuation (arrow) that obscures the peripheral anatomy. (b) Axial MD water (iodine) DECT image shows the artifact amplified (arrow). (c) Axial MD iodine DECT image shows that this artifact can be ameliorated (arrow), and assessment of the small bowel and subcutaneous tissues is now possible.

Figure 1b.

Out-of-field artifacts on images acquired with a first-generation rapid kVp-switching DECT platform in a 56-year-old man with a transverse diameter greater than 50 cm. (a) Axial 70-keV DECT image shows a crescentic area of hyperattenuation (arrow) that obscures the peripheral anatomy. (b) Axial MD water (iodine) DECT image shows the artifact amplified (arrow). (c) Axial MD iodine DECT image shows that this artifact can be ameliorated (arrow), and assessment of the small bowel and subcutaneous tissues is now possible.

Figure 1c.

Out-of-field artifacts on images acquired with a first-generation rapid kVp-switching DECT platform in a 56-year-old man with a transverse diameter greater than 50 cm. (a) Axial 70-keV DECT image shows a crescentic area of hyperattenuation (arrow) that obscures the peripheral anatomy. (b) Axial MD water (iodine) DECT image shows the artifact amplified (arrow). (c) Axial MD iodine DECT image shows that this artifact can be ameliorated (arrow), and assessment of the small bowel and subcutaneous tissues is now possible.

Peristalsis-related Artifacts

Bowel artifacts are seen in almost 49% of CT scans and can hinder the assessment of the mesentery and adjacent organs such as the liver (16). An evaluation of 100 patients (16) revealed that both 70-keV and MD water (iodine) images show this artifact similar to its appearance on polychromatic single-energy CT images, and the affected areas may show spuriously high or low attenuation. Although these artifacts are mitigated at single-energy CT by minimizing the rotation time (17) or using antiperistaltic medications, DECT is advantageous, because MD iodine images can be leveraged for amelioration (Fig 2). A plausible reason for the use of MD iodine images is that artifactual streaks occur with both low- and high-energy image datasets, and hence, they behave similarly to water and are thus removed from MD iodine (water) images.

Figure 2a.

Peristalsis-related artifact on images acquired with a rapid kVp-switching DECT platform in a 42-year-old woman. (a, b) Axial 65-keV (a) and MD water (iodine) (b) DECT images show a hypoattenuating streak artifact (arrow) that mimics bowel perforation and a hyperattenuating artifact (arrowhead) that simulates bleeding or a hypervascular lesion. The streak artifacts are secondary to peristalsis and obscure the mesentery. (c) Axial MD iodine DECT image shows reduced artifacts (arrow, arrowhead) and the lack of bowel abnormalities. Both barium (in the bowel loops) and iodine (in the vessels) are removed on the water (iodine) (or VUE-equivalent) image.

Figure 2b.

Peristalsis-related artifact on images acquired with a rapid kVp-switching DECT platform in a 42-year-old woman. (a, b) Axial 65-keV (a) and MD water (iodine) (b) DECT images show a hypoattenuating streak artifact (arrow) that mimics bowel perforation and a hyperattenuating artifact (arrowhead) that simulates bleeding or a hypervascular lesion. The streak artifacts are secondary to peristalsis and obscure the mesentery. (c) Axial MD iodine DECT image shows reduced artifacts (arrow, arrowhead) and the lack of bowel abnormalities. Both barium (in the bowel loops) and iodine (in the vessels) are removed on the water (iodine) (or VUE-equivalent) image.

Figure 2c.

Peristalsis-related artifact on images acquired with a rapid kVp-switching DECT platform in a 42-year-old woman. (a, b) Axial 65-keV (a) and MD water (iodine) (b) DECT images show a hypoattenuating streak artifact (arrow) that mimics bowel perforation and a hyperattenuating artifact (arrowhead) that simulates bleeding or a hypervascular lesion. The streak artifacts are secondary to peristalsis and obscure the mesentery. (c) Axial MD iodine DECT image shows reduced artifacts (arrow, arrowhead) and the lack of bowel abnormalities. Both barium (in the bowel loops) and iodine (in the vessels) are removed on the water (iodine) (or VUE-equivalent) image.

Beam Hardening

High-energy 108–149 keV VMC imaging mitigates metal artifacts by reducing beam hardening (18). However, VMC images cannot alleviate photon starvation artifacts. This artifact-reducing capability is affected by the type of alloy and the size of prosthesis because of differences in their interaction with x-ray photons. Therefore, VMC images can reduce image streaking from beam-hardening artifacts that emanate from a low-attenuation or thin high-attenuation prosthesis such as those made of titanium. A cobalt-chrome prosthesis, which is made of a higher-attenuating material, is less amenable to artifact reduction, because the cause of the artifact is photon starvation. Implant-specific and platform-specific protocols have been suggested with DECT (19). Projection-based algorithms may also be more effective, rather than just using high–kiloelectron voltage images, in a case of photon starvation (20).

Challenges of Quantification with VMC Images

Quantitatively, attenuation at CT (in Hounsfield units) is not only dependent on a structure’s effective atomic number but also is highly dependent on energy. For example, the liver parenchyma can measure 110 HU at 50 keV, 80 HU at 70 keV, and 65 HU at 140 keV (21). At the same kiloelectron voltage levels, the adrenal gland measures 174 HU, 80 HU, and 27 HU, respectively (21). Therefore, radiologists should pay attention to the kiloelectron voltage level, just as they do to the kilovolt peak level, when they interrogate a region of interest (Fig 3). Radiologists must be careful because an optimal kiloelectron voltage of choice used in clinical practice may not provide lesion attenuation values on which most societal guidelines are based. Revised cutoff attenuation values that are optimized for different kiloelectron voltage levels have been suggested by Patel et al (22) that achieved an area-under-the curve of 0.96–0.98 for detection of enhancement in renal lesions. Attenuation–kiloelectron voltage reference tables may also be a potential solution and can be useful when tube potentials other than 120 kVp are used with single-energy CT acquisitions with automated kilovolt peak selection. Variances in attenuation also exist among the DECT platforms, depending on kiloelectron voltage level and lesion type (8,23).

Figure 3a.

Variable attenuation at different kiloelectron voltage levels in an 83-year-old man. (a) Axial T2-weighted MR image shows a simple hyperintense renal cyst (*). (b–d) Axial VMC CT images obtained with a dual-source DECT platform show that the attenuation value of the cyst is 24.3 HU at 50 keV (b) and 13.2 HU at 70 keV (c) and reduces to 5.2 HU at 140 keV (d).

Figure 3b.

Variable attenuation at different kiloelectron voltage levels in an 83-year-old man. (a) Axial T2-weighted MR image shows a simple hyperintense renal cyst (*). (b–d) Axial VMC CT images obtained with a dual-source DECT platform show that the attenuation value of the cyst is 24.3 HU at 50 keV (b) and 13.2 HU at 70 keV (c) and reduces to 5.2 HU at 140 keV (d).

Figure 3c.

Variable attenuation at different kiloelectron voltage levels in an 83-year-old man. (a) Axial T2-weighted MR image shows a simple hyperintense renal cyst (*). (b–d) Axial VMC CT images obtained with a dual-source DECT platform show that the attenuation value of the cyst is 24.3 HU at 50 keV (b) and 13.2 HU at 70 keV (c) and reduces to 5.2 HU at 140 keV (d).

Figure 3d.

Variable attenuation at different kiloelectron voltage levels in an 83-year-old man. (a) Axial T2-weighted MR image shows a simple hyperintense renal cyst (*). (b–d) Axial VMC CT images obtained with a dual-source DECT platform show that the attenuation value of the cyst is 24.3 HU at 50 keV (b) and 13.2 HU at 70 keV (c) and reduces to 5.2 HU at 140 keV (d).

Artifacts and Pitfalls of MD Images

MD images highlight materials of interest that are defined by a user and are created with a material decomposition technique. Whether the material decomposition technique is based on two-material (rapid kVp-switching DECT and dual-layer DECT) or three-material (dual-source DECT and split-filter DECT) algorithms, MD images provide binary discrimination of materials (Fig 4). The presence of another material can confound assessment, because the attenuation of an additional material can be described as corresponding to the vectors of the initial materials being evaluated (24). The most common example of this is visualization of calcium-containing voxels such as bone with MD iodine and MD water or VUE images (Fig 5). Other examples include highly attenuating medication, pacemakers, and stents (Fig 6). Therefore, MD images should only be evaluated for the material pairs to which they correspond. For example, water images look different depending on the material pair used for reconstruction (eg, water [calcium] vs water [iodine] images). Radiologists must be aware of this to reduce diagnostic uncertainty (Fig 7), and a suggested schematic is presented in Figure 8 to facilitate interpretation. This affects all MD images.

Figure 4.

Material decomposition. Illustration shows how voxels are characterized by two- and three-material decomposition techniques. Rapid kilovoltage-switching and dual-layer DECT perform material analysis with a two-material decomposition technique, whereas dual-source and split-filter DECT involve three-material decomposition. With both decomposition techniques, voxels are vectorized so that the attenuation is split according to the materials being assessed. This is denoted by the application or image-pair name. SECT = single-energy CT.

Figure 5.

Illustration shows the effect of an additional material on material decomposition. When a material other than the material in question is present in a voxel, its attenuation can still be plotted. This results in erroneous representation. For example, calcium-containing voxels are seen on both MD iodine and VUE DECT images. This is illustrated by the appearance of calcified gallstones (row 1: rapid kilovoltage-switching [rs] DECT), choroid calcifications and bones (row 2: dual-source [ds] DECT), and aortic calcifications (row 3: dual-layer [dl] DECT). SECT = single-energy CT, sf = split filter.

Figure 6a.

Hyperattenuation of materials other than iodine on MD iodine images. Axial MD iodine CT images show that certain medications (arrow in a) at rapid kilovoltage-switching DECT (a), a pacemaker (arrow in b) at dual-source DECT (b), and aortic grafts (arrow in c) at dual-layer DECT (c) can appear to be hyperattenuating, although they do not contain iodine. Radiologists must be aware of this to prevent interpretive errors.

Figure 7a.

Calcium versus hemorrhage in a 34-year-old woman with a history of trauma who underwent unenhanced dual-source DECT of the brain. (a) Axial single-energy–equivalent CT image shows a punctate area of hyperattenuation in the left frontal lobe (arrow) that could be a hemorrhage or calcification. (b) Axial MD iodine CT image shows the same area of hyperattenuation (arrow). (c) Axial virtual noncalcium CT image shows a corresponding area of hypoattenuation (arrow) similar to the attenuation of the bones, which is suggestive of calcification. Note that calcium (including the bones) appears hyperattenuating on MD iodine images.

Figure 8.

Schema to guide interpretation with different DECT image datasets for characterization of an unknown material. The appearance of commonly encountered materials on VUE, MD iodine, and virtual noncalcium (VNCa) images is provided. Bright = high attenuation, dark = low attenuation, dense = enhancing.

Figure 6b.

Hyperattenuation of materials other than iodine on MD iodine images. Axial MD iodine CT images show that certain medications (arrow in a) at rapid kilovoltage-switching DECT (a), a pacemaker (arrow in b) at dual-source DECT (b), and aortic grafts (arrow in c) at dual-layer DECT (c) can appear to be hyperattenuating, although they do not contain iodine. Radiologists must be aware of this to prevent interpretive errors.

Figure 6c.

Hyperattenuation of materials other than iodine on MD iodine images. Axial MD iodine CT images show that certain medications (arrow in a) at rapid kilovoltage-switching DECT (a), a pacemaker (arrow in b) at dual-source DECT (b), and aortic grafts (arrow in c) at dual-layer DECT (c) can appear to be hyperattenuating, although they do not contain iodine. Radiologists must be aware of this to prevent interpretive errors.

Figure 7b.

Calcium versus hemorrhage in a 34-year-old woman with a history of trauma who underwent unenhanced dual-source DECT of the brain. (a) Axial single-energy–equivalent CT image shows a punctate area of hyperattenuation in the left frontal lobe (arrow) that could be a hemorrhage or calcification. (b) Axial MD iodine CT image shows the same area of hyperattenuation (arrow). (c) Axial virtual noncalcium CT image shows a corresponding area of hypoattenuation (arrow) similar to the attenuation of the bones, which is suggestive of calcification. Note that calcium (including the bones) appears hyperattenuating on MD iodine images.

Figure 7c.

Calcium versus hemorrhage in a 34-year-old woman with a history of trauma who underwent unenhanced dual-source DECT of the brain. (a) Axial single-energy–equivalent CT image shows a punctate area of hyperattenuation in the left frontal lobe (arrow) that could be a hemorrhage or calcification. (b) Axial MD iodine CT image shows the same area of hyperattenuation (arrow). (c) Axial virtual noncalcium CT image shows a corresponding area of hypoattenuation (arrow) similar to the attenuation of the bones, which is suggestive of calcification. Note that calcium (including the bones) appears hyperattenuating on MD iodine images.

Another artifact that affects all MD images is beam hardening from metal. The streaks affect decomposition analysis of structures that lie in the path of streaking and give an appearance of either an excess or lack of the material in question. High–kiloelectron voltage images can be used to mitigate the metal-related artifacts. However, iodine is suppressed on high–kiloelectron voltage images, and close comparison with prior imaging and other available kiloelectron voltage images can help in problem solving (Fig 9). In severe cases, it can also lead to incorrect characterization of materials, which may be corrected by applying metal artifact reduction algorithms (25).

Figure 9a.

Beam hardening from metallic spine hardware in a 56-year-old woman with breast cancer who was under oncologic surveillance. (a) Axial MD iodine CT image shows subtle hypoattenuation (arrow). Its visualization is hindered by streak artifacts from the spinal hardware. (b) Axial CT image shows that the lesion (arrow) is more conspicuous. (c) Axial 65-keV CT image acquired 2 months later shows that the lesion (arrow) has increased in size. This is indicative of a new enlarging hepatic metastasis.

Figure 9b.

Beam hardening from metallic spine hardware in a 56-year-old woman with breast cancer who was under oncologic surveillance. (a) Axial MD iodine CT image shows subtle hypoattenuation (arrow). Its visualization is hindered by streak artifacts from the spinal hardware. (b) Axial CT image shows that the lesion (arrow) is more conspicuous. (c) Axial 65-keV CT image acquired 2 months later shows that the lesion (arrow) has increased in size. This is indicative of a new enlarging hepatic metastasis.

Figure 9c.

Beam hardening from metallic spine hardware in a 56-year-old woman with breast cancer who was under oncologic surveillance. (a) Axial MD iodine CT image shows subtle hypoattenuation (arrow). Its visualization is hindered by streak artifacts from the spinal hardware. (b) Axial CT image shows that the lesion (arrow) is more conspicuous. (c) Axial 65-keV CT image acquired 2 months later shows that the lesion (arrow) has increased in size. This is indicative of a new enlarging hepatic metastasis.

MD Iodine CT Images

These are most commonly used to distinguish between enhancing and nonenhancing lesions and improve visualization of both hyper- and hypovascular masses.

Beam Hardening.—Beam-hardening artifacts occur because of the positioning of the arms beside the patient or the presence of metal, but they also occur because of the anatomy of the shoulder and pelvic bones and pose diagnostic challenges in lower cervical, upper thoracic, and pelvic imaging. These artifacts are more prevalent on MD iodine and MD water (iodine) images than they are on VMC images (26) and can obscure visualization of critical findings in the prostate (Fig 10) and urinary bladder and cause areas of falsely lower iodine content (Fig 11). The false appearance of high iodine content may also be seen, and it results in erroneous assessment of hyperenhancement or thickening of the bladder wall (26). Anecdotal cases, in our experience, show that these artifacts are reduced on MD iodine images acquired with second-generation rapid kVp-switching DECT when compared with the use of first-generation rapid kVp-switching DECT (Fig 10).

Figure 10a.

Beam hardening from the pelvic bones in a 77-year-old man with known prostate cancer. (a, b) Portal venous phase CT images were acquired in 2013 with a first-generation rapid kVp-switching DECT scanner for initial staging. Axial 65-keV CT image (a) shows an enhancing nodule (arrowhead) in the prostate gland. Axial MD iodine CT image (b) shows streaks (arrows) from the pelvic bones that degrade the image quality and obscure visualization of the enhancing lesion. (c, d) After the patient underwent prostatectomy, surveillance CT was performed in the portal venous phase in 2018 with a second-generation rapid kVp-switching DECT scanner. Axial 65-keV CT image (c) shows an enhancing nodule (arrowhead) at the surgical site that is suggestive of recurrence. Axial MD iodine CT image (d) shows that the streak artifacts are much less prominent when compared with those on b, and the recurrent lesion (arrowhead) is optimally seen.

Figure 11a.

Beam hardening from the shoulder bones in a 90-year-old man. (a) Axial contrast-enhanced dual-layer MD iodine DECT image shows streaks (arrows) arising from the shoulder bones and extending across the lung parenchyma that mimic the appearance of a perfusion defect. (b) Axial 120-kVp DECT image (lung window) shows no attenuation differences in the lung parenchyma and substantially fewer beam-hardening–related streaks.

Figure 10b.

Beam hardening from the pelvic bones in a 77-year-old man with known prostate cancer. (a, b) Portal venous phase CT images were acquired in 2013 with a first-generation rapid kVp-switching DECT scanner for initial staging. Axial 65-keV CT image (a) shows an enhancing nodule (arrowhead) in the prostate gland. Axial MD iodine CT image (b) shows streaks (arrows) from the pelvic bones that degrade the image quality and obscure visualization of the enhancing lesion. (c, d) After the patient underwent prostatectomy, surveillance CT was performed in the portal venous phase in 2018 with a second-generation rapid kVp-switching DECT scanner. Axial 65-keV CT image (c) shows an enhancing nodule (arrowhead) at the surgical site that is suggestive of recurrence. Axial MD iodine CT image (d) shows that the streak artifacts are much less prominent when compared with those on b, and the recurrent lesion (arrowhead) is optimally seen.

Figure 10c.

Beam hardening from the pelvic bones in a 77-year-old man with known prostate cancer. (a, b) Portal venous phase CT images were acquired in 2013 with a first-generation rapid kVp-switching DECT scanner for initial staging. Axial 65-keV CT image (a) shows an enhancing nodule (arrowhead) in the prostate gland. Axial MD iodine CT image (b) shows streaks (arrows) from the pelvic bones that degrade the image quality and obscure visualization of the enhancing lesion. (c, d) After the patient underwent prostatectomy, surveillance CT was performed in the portal venous phase in 2018 with a second-generation rapid kVp-switching DECT scanner. Axial 65-keV CT image (c) shows an enhancing nodule (arrowhead) at the surgical site that is suggestive of recurrence. Axial MD iodine CT image (d) shows that the streak artifacts are much less prominent when compared with those on b, and the recurrent lesion (arrowhead) is optimally seen.

Figure 10d.

Beam hardening from the pelvic bones in a 77-year-old man with known prostate cancer. (a, b) Portal venous phase CT images were acquired in 2013 with a first-generation rapid kVp-switching DECT scanner for initial staging. Axial 65-keV CT image (a) shows an enhancing nodule (arrowhead) in the prostate gland. Axial MD iodine CT image (b) shows streaks (arrows) from the pelvic bones that degrade the image quality and obscure visualization of the enhancing lesion. (c, d) After the patient underwent prostatectomy, surveillance CT was performed in the portal venous phase in 2018 with a second-generation rapid kVp-switching DECT scanner. Axial 65-keV CT image (c) shows an enhancing nodule (arrowhead) at the surgical site that is suggestive of recurrence. Axial MD iodine CT image (d) shows that the streak artifacts are much less prominent when compared with those on b, and the recurrent lesion (arrowhead) is optimally seen.

Figure 11b.

Beam hardening from the shoulder bones in a 90-year-old man. (a) Axial contrast-enhanced dual-layer MD iodine DECT image shows streaks (arrows) arising from the shoulder bones and extending across the lung parenchyma that mimic the appearance of a perfusion defect. (b) Axial 120-kVp DECT image (lung window) shows no attenuation differences in the lung parenchyma and substantially fewer beam-hardening–related streaks.

Pseudolesions.—Radiologists should be aware of artifactual or pseudolesions on MD iodine images. They can appear as speckles of color that indicate the appearance of iodine and are most likely due to inherent noise, scatter, or imaging parameters (25). The artifactual appearance of iodine in lesions such as renal and hepatic cysts (Fig 12) raises concern for a more sinister lesion. Occasionally, artifactual defects can mimic hypoattenuating lesions (27), most likely because of photon starvation. Findings on MD iodine CT images should be correlated with those on VMC CT images and those from other examinations. If doubt persists, further evaluation with contrast-enhanced MRI may be needed.

Figure 12a.

Artifactual iodine in benign lesions due to image noise. (a) Axial color-overlay MD iodine image shows speckles of orange color in a fluid-attenuation hepatic lesion (arrow) that typically represents iodine uptake, despite an appropriate window setting. This is due to inherent image noise. (b) Axial T2-weighted MR image helps confirm the hyperintense lesion (arrow) as a benign cyst.

Figure 12b.

Artifactual iodine in benign lesions due to image noise. (a) Axial color-overlay MD iodine image shows speckles of orange color in a fluid-attenuation hepatic lesion (arrow) that typically represents iodine uptake, despite an appropriate window setting. This is due to inherent image noise. (b) Axial T2-weighted MR image helps confirm the hyperintense lesion (arrow) as a benign cyst.

Specific Interpretive Challenges of MD Iodine CT Images.—MD iodine CT images are used in the evaluation of lung perfusion defects. An acute pulmonary embolism is typically seen as a segmental/subsegmental wedge-shaped perfusion defect (28), whereas a chronic pulmonary embolism is seen as a segmental or subsegmental wedge with mottled or mosaic perfusion defects (29). Pulmonary hypertension produces heterogeneous small nonsegmental mosaic mottled or wedge-shaped defects (30,31).

A wide variety of parenchymal lesions can mimic perfusion defects on MD iodine CT images, including infection, emphysema (Fig 13), bulla (Fig 14), atelectasis, neoplasms, and pleural effusions. Hence it is important to correlate perfusion defects on iodine maps with corresponding single-energy CT–equivalent images in the lung window that depict pleuroparenchymal abnormalities in high detail. This approach is similar to interpretation of a scintigraphic ventilation-perfusion scan that is also correlated with a chest radiograph. The shape and location of the perfusion defect also provide clues about the cause of the defect. Although pulmonary embolism is the most common cause of a wedge-shaped perfusion defect, it is also seen in bullae, segmental infiltration, and tumors that compress the artery. Tumor, fluid collection, and bullae appear as circumscribed perfusion defects. Patchy defects are seen in patients with emphysema, fibrosis, or fluid collection (32).

Figure 13a.

Artifactual perfusion defect in a 66-year-old man. (a) Axial single-energy–equivalent DECT image (lung window) shows emphysematous changes in both lungs (arrows). (b) Axial color-overlay MD iodine DECT image shows corresponding patchy areas (arrowheads) that mimic perfusion defects.

Figure 14a.

Artifactual perfusion defect in a 29-year-old man. (a) Axial single-energy–equivalent DECT image (lung window) shows a large bulla in the right lung (arrow). (b) Axial color-overlay MD iodine DECT image shows a corresponding well-demarcated area without iodine uptake (arrowhead) that mimics a perfusion defect.

Figure 13b.

Artifactual perfusion defect in a 66-year-old man. (a) Axial single-energy–equivalent DECT image (lung window) shows emphysematous changes in both lungs (arrows). (b) Axial color-overlay MD iodine DECT image shows corresponding patchy areas (arrowheads) that mimic perfusion defects.

Figure 14b.

Artifactual perfusion defect in a 29-year-old man. (a) Axial single-energy–equivalent DECT image (lung window) shows a large bulla in the right lung (arrow). (b) Axial color-overlay MD iodine DECT image shows a corresponding well-demarcated area without iodine uptake (arrowhead) that mimics a perfusion defect.

Motion and beam hardening can also simulate perfusion defects. These can be identified by their location. Cardiac motion (Fig 15) is seen adjacent to the heart in the right middle lobe and lingula, whereas diaphragmatic motion (Fig 16) manifests in the lung bases (33). The effect of motion can also be clearly seen on lung-window single-energy CT–equivalent images. Dark streaks of beam hardening from the high attenuation due to contrast enhancement (Fig 17) in the superior vena cava and brachiocephalic and subclavian veins can mimic pseudodefects in the apical or upper lung lobes. Pseudodefects can also be seen in the portion of the lung that lies outside the spectral field of view on dual-source DECT images.

Figure 15.

Artifactual perfusion defect due to motion in a 58-year-old man. Axial MD iodine DECT image shows a well-demarcated area without iodine uptake in the lingular segment (arrows) because of cardiac motion.

Figure 16.

Artifactual perfusion defect due to motion in a 62-year-old woman. Coronal MD iodine DECT image shows bands of areas without iodine uptake in both lungs (arrows) because of diaphragmatic motion.

Figure 17a.

Beam hardening due to attenuation of contrast media in a 72-year-old man. (a) Axial MD iodine dual-layer DECT image shows streaks (arrows) of dense contrast media in the brachiocephalic vein. These streaks can mimic perfusion defects from pulmonary embolism. (b) Axial contrast-enhanced 120-kVp DECT image (lung window) shows no hardening-related streaks.

Figure 17b.

Beam hardening due to attenuation of contrast media in a 72-year-old man. (a) Axial MD iodine dual-layer DECT image shows streaks (arrows) of dense contrast media in the brachiocephalic vein. These streaks can mimic perfusion defects from pulmonary embolism. (b) Axial contrast-enhanced 120-kVp DECT image (lung window) shows no hardening-related streaks.

Another important pitfall is the absence of a perfusion defect in the presence of a pulmonary embolus (ie, a false-negative finding), which is most often caused by a nonocclusive clot (Fig 18) and decreases the sensitivity of CT in the diagnosis of pulmonary embolism. A perfusion defect is seen in a nonocclusive clot in only 6%–9% of patients, compared with in 82%–95% of patients with an occlusive clot (34). A combined assessment of CT angiography and pulmonary blood volume is therefore recommended to identify nonocclusive pulmonary embolisms.

Figure 18.

False-negative perfusion defect in a 47-year-old man. Coronal MD iodine DECT image shows a nonocclusive thrombus (arrow) in a segmental branch of the pulmonary artery. The lung parenchyma shows no associated perfusion defects.

Distinguishing orally ingested material from active extravasation is crucial in the evaluation of a patient who is suspected of having a gastrointestinal bleed. True unenhanced (TUE) CT followed by multiphasic contrast-enhanced CT is used to facilitate diagnosis. Acquisition of VUE, low–kiloelectron voltage, and MD iodine CT images can reduce the radiation dose and improve detection of extravasation. High-attenuating material such as bismuth, which is often present in over-the-counter antacid formulations, appears hyperattenuating on MD iodine CT images and can be mistaken for extravasated iodine. VUE CT can allow discrimination of them because bismuth is visible on these images but iodine is suppressed. TUE CT can still be necessary to distinguish orally ingested barium from extravasated iodine because both barium and iodine have similar appearances on all DECT images (Fig 2).

The bowel wall has a specific appearance on MD iodine CT images. Pneumatosis, or air in the bowel wall, is indicative of ischemia, infection, or trauma and seen as a hypoattenuation in the wall. A hyperenhancing wall indicates ischemia, inflammation, or malignancy. On MD iodine CT images, the healthy bowel with any amount of gas (ie, partially or fully gas-distended enteric segments) can show pseudopneumatosis or pseudohyperenhancement artifacts (Fig 19) and may mimic an abnormality, which can be seen in up to 84% of CT examinations and may be seen in any bowel segment (35). Differentiation of artifacts from bowel abnormalities should be confirmed on VMC or single-energy CT–equivalent images.

Figure 19a.

Pseudopneumatosis and pseudohyperenhancement of the bowel wall in two patients. (a, b) Axial rapid kVp-switching 65-keV (a) and MD iodine (b) DECT images in a 73-year-old man show a gas-filled ileal bowel loop (* in a; arrow). The walls of this bowel loop have the appearance of pseudopneumatosis (arrow) on the MD iodine image (b). (c, d) Axial dual-source single-energy–equivalent (c) and color-overlay MD iodine (d) DECT images in an 89-year-old woman show the distal stomach (*), which is partially distended with air, and pseudohyperenhancement (white arrowhead in d) that mimics an abnormality in the anterior wall. An artifactually high uptake of iodine is seen in the adjacent liver parenchyma (black arrowhead in d).

Figure 19b.

Pseudopneumatosis and pseudohyperenhancement of the bowel wall in two patients. (a, b) Axial rapid kVp-switching 65-keV (a) and MD iodine (b) DECT images in a 73-year-old man show a gas-filled ileal bowel loop (* in a; arrow). The walls of this bowel loop have the appearance of pseudopneumatosis (arrow) on the MD iodine image (b). (c, d) Axial dual-source single-energy–equivalent (c) and color-overlay MD iodine (d) DECT images in an 89-year-old woman show the distal stomach (*), which is partially distended with air, and pseudohyperenhancement (white arrowhead in d) that mimics an abnormality in the anterior wall. An artifactually high uptake of iodine is seen in the adjacent liver parenchyma (black arrowhead in d).

Figure 19c.

Pseudopneumatosis and pseudohyperenhancement of the bowel wall in two patients. (a, b) Axial rapid kVp-switching 65-keV (a) and MD iodine (b) DECT images in a 73-year-old man show a gas-filled ileal bowel loop (* in a; arrow). The walls of this bowel loop have the appearance of pseudopneumatosis (arrow) on the MD iodine image (b). (c, d) Axial dual-source single-energy–equivalent (c) and color-overlay MD iodine (d) DECT images in an 89-year-old woman show the distal stomach (*), which is partially distended with air, and pseudohyperenhancement (white arrowhead in d) that mimics an abnormality in the anterior wall. An artifactually high uptake of iodine is seen in the adjacent liver parenchyma (black arrowhead in d).

Figure 19d.

Pseudopneumatosis and pseudohyperenhancement of the bowel wall in two patients. (a, b) Axial rapid kVp-switching 65-keV (a) and MD iodine (b) DECT images in a 73-year-old man show a gas-filled ileal bowel loop (* in a; arrow). The walls of this bowel loop have the appearance of pseudopneumatosis (arrow) on the MD iodine image (b). (c, d) Axial dual-source single-energy–equivalent (c) and color-overlay MD iodine (d) DECT images in an 89-year-old woman show the distal stomach (*), which is partially distended with air, and pseudohyperenhancement (white arrowhead in d) that mimics an abnormality in the anterior wall. An artifactually high uptake of iodine is seen in the adjacent liver parenchyma (black arrowhead in d).

Stents and catheters can be hyperattenuating or hypoattenuating on MD iodine images, depending on their composition. More importantly, the same object can look different on MD iodine images derived from different DECT approaches (Fig 20).

Figure 20a.

Varying appearance of a stent on MD iodine DECT images acquired with different platforms in a 54-year-old man. (a) Axial 120-kVp contrast-enhanced CT image shows a hyperattenuating lumen-apposing self-expandable covered metal stent (arrowhead) crossing the stomach to the hepaticojejunostomy, with a hyperattenuating coaxial pigtail (arrow) within. (b) Axial dual-layer DECT image shows the pigtail (arrow) as completely hypoattenuating. (c) Axial MD iodine rapid kV-switching DECT image shows the pigtail (arrow) as hypoattenuating and peripherally hyperattenuating. (d) Axial split-filter MD iodine DECT image shows the pigtail (arrow) as completely hyperattenuating. This is a color MD iodine map overlaid on a kilovolt peak–equivalent image; therefore, the stent has the same characteristics as those on a kilovolt peak image.

Figure 20b.

Varying appearance of a stent on MD iodine DECT images acquired with different platforms in a 54-year-old man. (a) Axial 120-kVp contrast-enhanced CT image shows a hyperattenuating lumen-apposing self-expandable covered metal stent (arrowhead) crossing the stomach to the hepaticojejunostomy, with a hyperattenuating coaxial pigtail (arrow) within. (b) Axial dual-layer DECT image shows the pigtail (arrow) as completely hypoattenuating. (c) Axial MD iodine rapid kV-switching DECT image shows the pigtail (arrow) as hypoattenuating and peripherally hyperattenuating. (d) Axial split-filter MD iodine DECT image shows the pigtail (arrow) as completely hyperattenuating. This is a color MD iodine map overlaid on a kilovolt peak–equivalent image; therefore, the stent has the same characteristics as those on a kilovolt peak image.

Figure 20c.

Varying appearance of a stent on MD iodine DECT images acquired with different platforms in a 54-year-old man. (a) Axial 120-kVp contrast-enhanced CT image shows a hyperattenuating lumen-apposing self-expandable covered metal stent (arrowhead) crossing the stomach to the hepaticojejunostomy, with a hyperattenuating coaxial pigtail (arrow) within. (b) Axial dual-layer DECT image shows the pigtail (arrow) as completely hypoattenuating. (c) Axial MD iodine rapid kV-switching DECT image shows the pigtail (arrow) as hypoattenuating and peripherally hyperattenuating. (d) Axial split-filter MD iodine DECT image shows the pigtail (arrow) as completely hyperattenuating. This is a color MD iodine map overlaid on a kilovolt peak–equivalent image; therefore, the stent has the same characteristics as those on a kilovolt peak image.

Figure 20d.

Varying appearance of a stent on MD iodine DECT images acquired with different platforms in a 54-year-old man. (a) Axial 120-kVp contrast-enhanced CT image shows a hyperattenuating lumen-apposing self-expandable covered metal stent (arrowhead) crossing the stomach to the hepaticojejunostomy, with a hyperattenuating coaxial pigtail (arrow) within. (b) Axial dual-layer DECT image shows the pigtail (arrow) as completely hypoattenuating. (c) Axial MD iodine rapid kV-switching DECT image shows the pigtail (arrow) as hypoattenuating and peripherally hyperattenuating. (d) Axial split-filter MD iodine DECT image shows the pigtail (arrow) as completely hyperattenuating. This is a color MD iodine map overlaid on a kilovolt peak–equivalent image; therefore, the stent has the same characteristics as those on a kilovolt peak image.

Challenges in Quantifying Iodine on MD Iodine DECT Images.—Excellent correlation has been shown between the expected and measured iodine concentration of MD iodine images with most of the currently available DECT platforms, with an accuracy of 10% compared with the true value (36). The lowest measurement errors have been observed on images acquired with the largest spectral separation, such as 70 or 80 kVp and 150 kVp with a tin filter at dual-source DECT or 140 kVp at dual-layer DECT (37). The limit of qualitative (0.3–0.5 mg/mL) and quantitative (0.5–1.0 mg/mL) detection of iodine is directly influenced by patient size (38).

An important challenge is to determine cutoff values to detect enhancement, because there is insufficient literature and the values can vary according to the DECT platform used (39). Independent investigations with CT examinations performed with dual-source DECT and rapid kVp-switching DECT have shown thresholds of 0.5 mg/mL and 1.3–2.0 mg/mL, respectively (39). These results suggested the need for platform-specific thresholds. Patel et al (39) investigated renal lesions on both of these platforms and found that irrespective of platform, an absolute value of 1.5 mg/mL can allow vascular lesions to be discerned from nonvascular lesions. The authors also demonstrated that a relative or normalized measurement obtained by calculating the ratio of lesion to aorta iodine concentration reduces interplatform variability and improves diagnostic performance.

In addition to platform and patient size, lesion attenuation on unenhanced images and location have also been shown to influence iodine quantification. Meyer et al (40) found that 1.0–4.0–cm unenhancing hyperattenuating (>20 HU) lesions may still show detectable iodine (Fig 21). Large-scale prospective studies are still needed to make generalizable conclusions, and radiologists must be cognizant of these fallacies.

Figure 21a.

Quantitation on MD iodine DECT images and iodine in a hyperattenuating cyst in a 67-year-old man with an incidental renal lesion. (a) Axial portal venous phase 70-keV CT image shows a 10-mm left renal lesion (arrow) that measures 90 HU. (b) Axial portal venous phase MD iodine DECT image shows the lesion (arrow) with an iodine concentration of 1.9 mg/mL. (c) Axial TUE DECT image shows a hyperattenuating (66 HU) benign hemorrhagic renal cyst (arrow). (d, e) Follow-up axial unenhanced T1-weighted MR image (d) shows a hyperintense lesion (arrow), and axial contrast-enhanced T1-weighted MR image (e) shows no internal enhancement (arrow), which allows confirmation of the diagnosis of a hemorrhagic cyst.

Figure 21b.

Quantitation on MD iodine DECT images and iodine in a hyperattenuating cyst in a 67-year-old man with an incidental renal lesion. (a) Axial portal venous phase 70-keV CT image shows a 10-mm left renal lesion (arrow) that measures 90 HU. (b) Axial portal venous phase MD iodine DECT image shows the lesion (arrow) with an iodine concentration of 1.9 mg/mL. (c) Axial TUE DECT image shows a hyperattenuating (66 HU) benign hemorrhagic renal cyst (arrow). (d, e) Follow-up axial unenhanced T1-weighted MR image (d) shows a hyperintense lesion (arrow), and axial contrast-enhanced T1-weighted MR image (e) shows no internal enhancement (arrow), which allows confirmation of the diagnosis of a hemorrhagic cyst.

Figure 21c.

Quantitation on MD iodine DECT images and iodine in a hyperattenuating cyst in a 67-year-old man with an incidental renal lesion. (a) Axial portal venous phase 70-keV CT image shows a 10-mm left renal lesion (arrow) that measures 90 HU. (b) Axial portal venous phase MD iodine DECT image shows the lesion (arrow) with an iodine concentration of 1.9 mg/mL. (c) Axial TUE DECT image shows a hyperattenuating (66 HU) benign hemorrhagic renal cyst (arrow). (d, e) Follow-up axial unenhanced T1-weighted MR image (d) shows a hyperintense lesion (arrow), and axial contrast-enhanced T1-weighted MR image (e) shows no internal enhancement (arrow), which allows confirmation of the diagnosis of a hemorrhagic cyst.

Figure 21d.

Quantitation on MD iodine DECT images and iodine in a hyperattenuating cyst in a 67-year-old man with an incidental renal lesion. (a) Axial portal venous phase 70-keV CT image shows a 10-mm left renal lesion (arrow) that measures 90 HU. (b) Axial portal venous phase MD iodine DECT image shows the lesion (arrow) with an iodine concentration of 1.9 mg/mL. (c) Axial TUE DECT image shows a hyperattenuating (66 HU) benign hemorrhagic renal cyst (arrow). (d, e) Follow-up axial unenhanced T1-weighted MR image (d) shows a hyperintense lesion (arrow), and axial contrast-enhanced T1-weighted MR image (e) shows no internal enhancement (arrow), which allows confirmation of the diagnosis of a hemorrhagic cyst.

Figure 21e.

Quantitation on MD iodine DECT images and iodine in a hyperattenuating cyst in a 67-year-old man with an incidental renal lesion. (a) Axial portal venous phase 70-keV CT image shows a 10-mm left renal lesion (arrow) that measures 90 HU. (b) Axial portal venous phase MD iodine DECT image shows the lesion (arrow) with an iodine concentration of 1.9 mg/mL. (c) Axial TUE DECT image shows a hyperattenuating (66 HU) benign hemorrhagic renal cyst (arrow). (d, e) Follow-up axial unenhanced T1-weighted MR image (d) shows a hyperintense lesion (arrow), and axial contrast-enhanced T1-weighted MR image (e) shows no internal enhancement (arrow), which allows confirmation of the diagnosis of a hemorrhagic cyst.

Iodine concentration cannot be measured on the color-overlay images in a picture archiving and communications system (PACS) because attenuation cannot be processed on color images with Digital Imaging and Communications in Medicine (DICOM) software. Measurement on these images can only be obtained by using proprietary vendor software. On the PACS, color images can be used only for qualitative purposes, while both qualitative and quantitative analysis can be performed on gray-scale images. The standard deviation of iodine measurements is also not quantifiable at dual-source DECT and with split-filter DECT platforms, but it is available at rapid kVp-switching DECT and dual-layer DECT.

MD Urate Images

These images are used for evaluating the presence of monosodium urate crystals in patients with gout and for determining the composition of urinary stones to differentiate uric acid from non–uric acid stones. Use of DECT for detection of monosodium urate crystals is part of the clinical guidelines (4).

In addition to the motion-related and image noise artifacts mentioned previously, MD urate CT images can show beam-hardening artifacts from an adjacent prosthesis, staples, jewelry, or dense cortical bone (40). By far, the most common artifacts, which are seen in up to 88% of patients, are green pixelations that mimic urate at the nail beds (Fig 21), skin, calluses (Fig 22), and tendons (41). The postulated reasoning for this is that the decomposition ratios for uric acid and keratin are similar (42). False-positive interpretation can be reduced by creating multiplanar reformation images to correlate anatomic planes and by improving spectral separation (eg, 80 kVp/tin 150 kVp vs 80/140 kVp) (43). Choosing appropriate postprocessing parameters such as air and bone distances, which are vendor-specific adjustable parameters available at dual-source DECT and split-filter DECT that represent the minimum voxel distance among urate and air and bone, can also help. A high air distance of 10 mm and a low bone distance of 5 mm have been shown to mitigate this artifact (41). An initial flare, low volume, or concentration of tophi deposits can lead to a false-negative interpretation and affect sensitivity (42,44).

Figure 22a.

Artifactual urate deposition in the nail bed and skin in two patients. (a) Axial color-overlay MD urate DECT image of the foot in a 40-year-old man shows green pixelations (arrow) in the nail bed. (b) Volume-rendered MD urate image in a 43-year-old man shows green pixelations in the skin along the plantar (arrows) and dorsal (arrowheads) aspects of the foot. The green pixelations mimic deposition of gout crystals and are due to the similar attenuation of keratin and monosodium urate.

Figure 22b.

Artifactual urate deposition in the nail bed and skin in two patients. (a) Axial color-overlay MD urate DECT image of the foot in a 40-year-old man shows green pixelations (arrow) in the nail bed. (b) Volume-rendered MD urate image in a 43-year-old man shows green pixelations in the skin along the plantar (arrows) and dorsal (arrowheads) aspects of the foot. The green pixelations mimic deposition of gout crystals and are due to the similar attenuation of keratin and monosodium urate.

A debatable artifact is the appearance of uratein vascular walls. It is hypothesized that urate has a role in vascular homeostasis, and investigators (45) have demonstrated urate deposits with DECT along the vascular lumen and cardiac valves in patients with gout. However, no urate deposition that is demonstrable at DECT was found in carotid vessels of patients with asymptomatic hyperuricemia and metabolic syndrome (46). Motion-related, beam-hardening, and partial averaging artifacts; image noise; and low serum uric acid levels have been attributed to these discrepant findings, and therefore, large-scale clinical studies with adequate pathologic correlates are needed (47).

To avoid missing or misinterpreting small urinary stones subjacent to ureteral stents in patients with urolithiasis, radiologists should note that silicone stents appear red (similar to urate stones) and polyurethane stents appear blue (similar to nonurate stones) with default settings on both dual-source DECT and split-filter DECT images (48).

Artifacts and Pitfalls of VUE CT Images

VUE CT images are generated by subtracting iodine from contrast-enhanced CT images. Acquisition of VUE images can preclude acquisition of TUE images, thus reducing the radiation dose and improving workflow (49). However, certain artifacts are specific to VUE CT images.

Incomplete Iodine Removal

All DECT platforms can show residual iodine voxels on VUE images. Incomplete iodine removal is most commonly observed when the iodine concentration is very high (50). Frequent examples include (a) residual iodine in arteries or portal veins on arterial phase VUE CT images (Fig 23); (b) iodinated contrast media in the renal collecting system on delayed or excretory phase VUE CT images (Fig 24); (c) incomplete subtraction of a high-concentration, high-attenuating contrast media bolus in the subclavian veins (Fig 25), superior vena cava, or right atrium; and (d) inhomogeneous subtraction of high-attenuating enteric contrast media. This phenomenon can be explained as the saturation effect, in which attenuation values of a highly concentrated iodine column reach close to the maximum attenuation values on the Hounsfield scale, minimizing the difference in attenuation between low- and high-kVp CT images. As a result, iodine with saturated attenuation values is categorized as noniodine and is still displayed. Incomplete iodine removal at VUE CT may lead to a difficulty in differentiating between iodine and calcium in the absence of TUE CT (Fig 24) (51). Streaks from the high-attenuating iodinated contrast media bolus can also impair assessment of pulmonary perfusion and can lead to a false-positive or false-negative result (52).

Figure 23a.

Residual iodine in vessels on VUE DECT images in a 75-year-old man. (a) Axial contrast-enhanced rapid kVp-switching 50-keV DECT image shows heterogeneous opacification of the portal vein (arrow). (b) Axial arterial phase VUE DECT image shows residual iodine in the portal vein (arrow) that appears slightly hyperattenuating and can mimic a thrombus.

Figure 24a.

Residual iodine in the collecting system on VUE DECT images in a 72-year-old man. (a, b) Axial dual-layer MD iodine (a) and VUE (b) DECT images show hyperattenuation (arrow) in the right renal pelvis. (c) Axial TUE DECT image shows no corresponding hyperattenuation, which is suggestive of residual contrast media in the collecting system on b due to a saturation effect. (d–f) Axial MD iodine (d), VUE (e), and TUE (f) DECT images show hyperattenuation in the right ureter (arrow), which is suggestive of calculus.

Figure 25a.

Residual iodine in veins on contrast-enhanced dual-source DECT and VUE CT images of the chest in a 36-year-old man. (a) Axial contrast-enhanced single-energy–equivalent dual-source CT image shows a contrast-enhanced right subclavian vein (arrow) and carotid vessels (arrowhead). (b) Axial VUE CT image shows residual iodine in the right subclavian vein (arrow) and subtraction of contrast media in the carotid vessels (arrowhead).

Figure 23b.

Residual iodine in vessels on VUE DECT images in a 75-year-old man. (a) Axial contrast-enhanced rapid kVp-switching 50-keV DECT image shows heterogeneous opacification of the portal vein (arrow). (b) Axial arterial phase VUE DECT image shows residual iodine in the portal vein (arrow) that appears slightly hyperattenuating and can mimic a thrombus.

Figure 24b.

Residual iodine in the collecting system on VUE DECT images in a 72-year-old man. (a, b) Axial dual-layer MD iodine (a) and VUE (b) DECT images show hyperattenuation (arrow) in the right renal pelvis. (c) Axial TUE DECT image shows no corresponding hyperattenuation, which is suggestive of residual contrast media in the collecting system on b due to a saturation effect. (d–f) Axial MD iodine (d), VUE (e), and TUE (f) DECT images show hyperattenuation in the right ureter (arrow), which is suggestive of calculus.

Figure 24c.

Residual iodine in the collecting system on VUE DECT images in a 72-year-old man. (a, b) Axial dual-layer MD iodine (a) and VUE (b) DECT images show hyperattenuation (arrow) in the right renal pelvis. (c) Axial TUE DECT image shows no corresponding hyperattenuation, which is suggestive of residual contrast media in the collecting system on b due to a saturation effect. (d–f) Axial MD iodine (d), VUE (e), and TUE (f) DECT images show hyperattenuation in the right ureter (arrow), which is suggestive of calculus.

Figure 24d.

Residual iodine in the collecting system on VUE DECT images in a 72-year-old man. (a, b) Axial dual-layer MD iodine (a) and VUE (b) DECT images show hyperattenuation (arrow) in the right renal pelvis. (c) Axial TUE DECT image shows no corresponding hyperattenuation, which is suggestive of residual contrast media in the collecting system on b due to a saturation effect. (d–f) Axial MD iodine (d), VUE (e), and TUE (f) DECT images show hyperattenuation in the right ureter (arrow), which is suggestive of calculus.

Figure 24e.

Residual iodine in the collecting system on VUE DECT images in a 72-year-old man. (a, b) Axial dual-layer MD iodine (a) and VUE (b) DECT images show hyperattenuation (arrow) in the right renal pelvis. (c) Axial TUE DECT image shows no corresponding hyperattenuation, which is suggestive of residual contrast media in the collecting system on b due to a saturation effect. (d–f) Axial MD iodine (d), VUE (e), and TUE (f) DECT images show hyperattenuation in the right ureter (arrow), which is suggestive of calculus.

Figure 24f.

Residual iodine in the collecting system on VUE DECT images in a 72-year-old man. (a, b) Axial dual-layer MD iodine (a) and VUE (b) DECT images show hyperattenuation (arrow) in the right renal pelvis. (c) Axial TUE DECT image shows no corresponding hyperattenuation, which is suggestive of residual contrast media in the collecting system on b due to a saturation effect. (d–f) Axial MD iodine (d), VUE (e), and TUE (f) DECT images show hyperattenuation in the right ureter (arrow), which is suggestive of calculus.

Figure 25b.

Residual iodine in veins on contrast-enhanced dual-source DECT and VUE CT images of the chest in a 36-year-old man. (a) Axial contrast-enhanced single-energy–equivalent dual-source CT image shows a contrast-enhanced right subclavian vein (arrow) and carotid vessels (arrowhead). (b) Axial VUE CT image shows residual iodine in the right subclavian vein (arrow) and subtraction of contrast media in the carotid vessels (arrowhead).

Lipiodol is a drug delivery vehicle that is commonly used in transarterial chemoembolization procedures. It is an iodized poppy seed oil (480 mg of iodine per milliliter) that preferentially accumulates in treated hepatocellular carcinomas and correlates with treatment response and recurrence. The lack of visualization of lipiodol at VUE CT is not an artifact, because iodine is rightfully being removed from the image. However, the concentration of lipiodol varies widely after transarterial chemoembolization, so that lower concentrations of lipiodol can be removed at VUE CT, while it remains in areas with higher concentrations (53). This heterogeneous subtraction of lipiodol may hinder the use of VUE CT images for evaluation after transarterial chemoembolization (Fig 26).

Figure 26a.

Lipiodol on VUE DECT images in a 61-year-old man with hepatocellular carcinoma who was treated with transarterial chemoembolization with lipiodol. (a) Axial dual-layer 120-kVp contrast-enhanced portal venous phase DECT image shows an inhomogeneous hyperattenuating exophytic hepatic lesion. (b) Axial MD iodine DECT image shows the lesion as hyperattenuating because of the presence of iodine in lipiodol. Because lipiodol is visualized on MD iodine CT images, detection of arterially enhancing foci that suggest a recurrent or residual tumor would be hard to discern on these images. (c) Axial VUE DECT image shows that lipiodol in the treated lesion is inhomogeneously suppressed, with a couple of peripheral hyperattenuating specks (arrowheads) still visible. (d) Axial TUE DECT image shows that lipiodol remains hyperattenuating. Thus, the VUE image may not be a suitable surrogate for evaluation of hepatocellular carcinoma treated with transarterial chemoembolization.

Figure 26b.

Lipiodol on VUE DECT images in a 61-year-old man with hepatocellular carcinoma who was treated with transarterial chemoembolization with lipiodol. (a) Axial dual-layer 120-kVp contrast-enhanced portal venous phase DECT image shows an inhomogeneous hyperattenuating exophytic hepatic lesion. (b) Axial MD iodine DECT image shows the lesion as hyperattenuating because of the presence of iodine in lipiodol. Because lipiodol is visualized on MD iodine CT images, detection of arterially enhancing foci that suggest a recurrent or residual tumor would be hard to discern on these images. (c) Axial VUE DECT image shows that lipiodol in the treated lesion is inhomogeneously suppressed, with a couple of peripheral hyperattenuating specks (arrowheads) still visible. (d) Axial TUE DECT image shows that lipiodol remains hyperattenuating. Thus, the VUE image may not be a suitable surrogate for evaluation of hepatocellular carcinoma treated with transarterial chemoembolization.

Figure 26c.

Lipiodol on VUE DECT images in a 61-year-old man with hepatocellular carcinoma who was treated with transarterial chemoembolization with lipiodol. (a) Axial dual-layer 120-kVp contrast-enhanced portal venous phase DECT image shows an inhomogeneous hyperattenuating exophytic hepatic lesion. (b) Axial MD iodine DECT image shows the lesion as hyperattenuating because of the presence of iodine in lipiodol. Because lipiodol is visualized on MD iodine CT images, detection of arterially enhancing foci that suggest a recurrent or residual tumor would be hard to discern on these images. (c) Axial VUE DECT image shows that lipiodol in the treated lesion is inhomogeneously suppressed, with a couple of peripheral hyperattenuating specks (arrowheads) still visible. (d) Axial TUE DECT image shows that lipiodol remains hyperattenuating. Thus, the VUE image may not be a suitable surrogate for evaluation of hepatocellular carcinoma treated with transarterial chemoembolization.

Figure 26d.

Lipiodol on VUE DECT images in a 61-year-old man with hepatocellular carcinoma who was treated with transarterial chemoembolization with lipiodol. (a) Axial dual-layer 120-kVp contrast-enhanced portal venous phase DECT image shows an inhomogeneous hyperattenuating exophytic hepatic lesion. (b) Axial MD iodine DECT image shows the lesion as hyperattenuating because of the presence of iodine in lipiodol. Because lipiodol is visualized on MD iodine CT images, detection of arterially enhancing foci that suggest a recurrent or residual tumor would be hard to discern on these images. (c) Axial VUE DECT image shows that lipiodol in the treated lesion is inhomogeneously suppressed, with a couple of peripheral hyperattenuating specks (arrowheads) still visible. (d) Axial TUE DECT image shows that lipiodol remains hyperattenuating. Thus, the VUE image may not be a suitable surrogate for evaluation of hepatocellular carcinoma treated with transarterial chemoembolization.

Removal of Calcium

As described earlier, calcium is decomposed on both MD iodine and VUE CT images. This means that calcium-containing voxels are seen as hyperattenuating regions on both MD iodine and VUE CT images, and iodine-containing voxels are seen as hyperattenuating regions on MD iodine and virtual noncalcium CT images. In clinical routine, voxels that display high attenuation on both contrast-enhanced and VUE CT images are recognized as calcium. However, when a calcific object is too small or not highly attenuating, the VUE reconstruction process can result in removal or reduced attenuation of calcium (eg, small urinary stones or calcification in a tumor). Studies (54–56) have demonstrated that urinary calculi smaller than 2.8 mm or with attenuation values less than 387 HU are highly likely to be erased on VUE CT images, especially on noisy images or on images of obese patients. This can also lead to ambiguity in assessment for endoleaks in patients with a calcified intramural thrombus. Even if calcium is visualized, the reduced blooming of calcium and metallic clips (Fig 27) (57,58) on VUE CT images can affect sensitivity for detection of gallstones, urinary stones (Fig 27), and calcified parenchymal lesions or can lead to a discrepancy in the size of the stones or lesions between TUE and VUE images (59).

Figure 27a.

Reduced attenuation of metal and removal of calcium on VUE DECT images in a 31-year-old woman. (a) Axial TUE rapid kVp-switching DECT image shows metallic clips (arrows) along the left kidney from a prior partial nephrectomy and punctate calculus in the right kidney (arrowhead). (b) Axial rapid kVp-switching VUE DECT image shows reduced attenuation of the metallic clips (arrows) and erasure of calculus (arrowhead) in comparison with the TUE CT image.

Figure 27b.

Reduced attenuation of metal and removal of calcium on VUE DECT images in a 31-year-old woman. (a) Axial TUE rapid kVp-switching DECT image shows metallic clips (arrows) along the left kidney from a prior partial nephrectomy and punctate calculus in the right kidney (arrowhead). (b) Axial rapid kVp-switching VUE DECT image shows reduced attenuation of the metallic clips (arrows) and erasure of calculus (arrowhead) in comparison with the TUE CT image.

Removal of Foreign Materials

The integrity and placement of stents and grafts is evaluated on unenhanced images because they display high attenuation. However, some stents or grafts may be erroneously removed (Fig 28) or show reduced attenuation or blooming (Fig 29) on VUE CT images. To our knowledge, studies have anecdotally reported this finding (60), but the performance of VUE CT for assessment of stent integrity has not been evaluated.

Figure 28a.

Removal of foreign materials on VUE images in a 25-year-old man. (a) Axial dual-source TUE DECT image shows a hyperattenuating Hobbs pancreatic stent (arrow) and a hyperattenuating biliary stent (arrowhead). (b) Axial dual-source VUE DECT image shows that the pancreatic stent (arrow), which is made of soft pliable material, is no longer visible, whereas the biliary stent (arrowhead) still appears hyperattenuating.

Figure 29a.

Reduced attenuation of foreign materials on VUE images in a 57-year-old man. (a) Axial TUE dual-source DECT image of the chest shows a hyperattenuating Valsalva Dacron graft (arrow) placed as part of aortic root repair. (b) Axial VUE dual-source DECT image shows reduced attenuation of the graft (arrow).

Figure 28b.

Removal of foreign materials on VUE images in a 25-year-old man. (a) Axial dual-source TUE DECT image shows a hyperattenuating Hobbs pancreatic stent (arrow) and a hyperattenuating biliary stent (arrowhead). (b) Axial dual-source VUE DECT image shows that the pancreatic stent (arrow), which is made of soft pliable material, is no longer visible, whereas the biliary stent (arrowhead) still appears hyperattenuating.

Figure 29b.

Reduced attenuation of foreign materials on VUE images in a 57-year-old man. (a) Axial TUE dual-source DECT image of the chest shows a hyperattenuating Valsalva Dacron graft (arrow) placed as part of aortic root repair. (b) Axial VUE dual-source DECT image shows reduced attenuation of the graft (arrow).

Challenges in Quantifying Attenuation on VUE Images

Reliable and valid estimation of attenuation on unenhanced images is important in several clinical settings, especially for characterization of renal and adrenal lesions and evaluation of fatty liver. Good correlation has been found between attenuation on TUE and VUE CT images. However, differences in attenuation between TUE and VUE CT images can become substantial based on factors such as body habitus and phase used to derive VUE CT images (61–67). Overall, attenuation on VUE CT images may be overestimated by 10–15 HU (49). This can affect characterization of lesions with borderline attenuation values. Overestimation of adrenal lesion attenuation can reduce sensitivity but not affect specificity (68). For identification of fatty liver, the discrepancy in attenuation between TUE and VUE CT images could result in divergent diagnoses (Fig 30). A potential clue to assess the validity of attenuation on VUE CT images is to look for residual iodine on the images.

Figure 30a.

Quantitative challenge on VUE images in a 40-year-old man. (a) Axial dual-layer TUE DECT image shows hepatic and splenic attenuation values of 35 HU and 47 HU, respectively (spleen-liver attenuation, 12 HU). (b) Axial dual-layer VUE CT image shows liver attenuation of 48 HU and spleen attenuation of 51 HU (spleen-liver attenuation, 3 HU). This patient would receive a diagnosis of a fatty liver on the basis of TUE DECT images, but not on the basis of VUE DECT images (fatty liver criteria: liver attenuation, <40 HU; spleen-liver attenuation, >10 HU).

Figure 30b.

Quantitative challenge on VUE images in a 40-year-old man. (a) Axial dual-layer TUE DECT image shows hepatic and splenic attenuation values of 35 HU and 47 HU, respectively (spleen-liver attenuation, 12 HU). (b) Axial dual-layer VUE CT image shows liver attenuation of 48 HU and spleen attenuation of 51 HU (spleen-liver attenuation, 3 HU). This patient would receive a diagnosis of a fatty liver on the basis of TUE DECT images, but not on the basis of VUE DECT images (fatty liver criteria: liver attenuation, <40 HU; spleen-liver attenuation, >10 HU).

Artifacts and Pitfalls of Virtual Noncalcium CT Images

Virtual noncalcium CT images have been used most commonly for identification of bone marrow edema, marrow malignancy, and disk displacements. Incomplete masking and filtering can limit assessment of regions of less than 2 mm in dense cortical bone, including subcortical areas and vertebrae less than 4 mm in height or with more than 50% sclerosis and gas on virtual noncalcium CT images (18,69). Choosing an appropriate calcium suppression index during postprocessing of dual-layer DECT images can facilitate visualization of the subcortical region (70).

At rapid kVp-switching DECT, MD water (hydroxyapatite) images are shown to be susceptible to scatter artifacts, because the artifactual voxels are classified as water attenuation and are thus accentuated on these images (71). On the other hand, MD water (cortical bone) CT images have been shown to mitigate artifacts seen at the bone–soft-tissue interface (72).

At dual-source DECT, virtual noncalcium images are based on decomposition of voxels into yellow marrow, red marrow, and bone. Because marrow types are the basis of material decomposition on this platform (in addition to the acquisition parameters and decomposition ratio), skeletal site, age, and prior treatment with radiation can influence qualitative and quantitative interpretation because of associated changes in bone marrow composition (73).

Conclusion

DECT has evolved from the research realm into a widely accepted advanced CT technique for numerous clinical applications. Although some pitfalls are common to both conventional and DECT images, the physical basis and decomposition process used to generate DECT reconstruction images introduce pitfalls to imaging interpretation and appearance that are specific to DECT with all commercially available DECT technologies. Recognizing the potential pitfalls of DECT is necessary for appropriate and accurate interpretation of the results of this increasingly important imaging tool.