Available dual-energy technologies
Different technological implementations of DECT are commercially available and in
clinical use. [Fig. 1] provides a schematic overview of available DECT concepts and technologies. While
different terminologies for the several DECT platforms have been suggested and used
in the literature, the Society for Body Computed Tomography and Magnetic Resonance
(SBCTMR) recently provided systematic nomenclature to facilitate communication, which
will be used throughout this article [6].
Fig. 1 Schematic overview of different CT technologies currently available, including conventional,
dual-energy, and multi-energy CT
To understand the underlying technological concepts of DECT, it is important to acknowledge
that X-rays as used in medical imaging refer to a spectrum of photons (or X-ray quanta)
of different energies (unit: kiloelectron volt, keV) with the maximum being defined
by the tube voltage (kVp). In this spectrum, the energies are not linearly distributed
but follow a specific pattern that depends on the X-ray tube materials and other system-inherited
characteristics, e. g., filters that are used. Alteration of the X-ray tube voltage
will therefore result in a different X-ray spectrum, which is the underlying principle
of emission-based DECT. While techniques such as scan-scan and rotate-rotate DECT
are available from some vendors, they are severely limited by motion effects and temporal
misregistration. The following technical implementations of DECT are commercially
available and relevant to routine imaging [7]
[8]:
Dual-source DECT (dsDECT): Two tube detector pairs offset by 90* to each other allow for simultaneous acquisition
of low- and high-energy attenuation spectra. The reconstructed images from both tubes
can be fused to obtain a conventional CT image or used as input for the reconstruction
of DECT results.
Rapid kVp switching DECT (rsDECT): Within a single rotation, the tube voltage is changed frequently in a succession
of as short as 0.5 ms, enabling acquisition of low- and high-energy attenuation spectra.
After angular correction, raw data can be fused to obtain a conventional CT image
or be used to reconstruct DECT results.
Twin-beam DECT (tbDECT): A filter made from gold and tin is placed along the Z-axis (out-of-plane) to split
the X-ray spectrum into a high-energy component and low-energy component that are
read out separately by corresponding detector rows.
Dual-layer DECT (dlDECT): The only available detector-based DECT system includes a horizontally aligned detector
that consists of two different scintillator materials with one being more prone to
detection of low- and high-energy photons, respectively (from a single X-ray spectrum).
Raw data can be processed aggregated or separately to obtain conventional images or
DECT results.
Photon-counting CT (PCCT): These systems use a novel detector that allows for direct conversion of photons to
energy (without the need for a scintillating layer). The detector, therefore, enables
higher resolution of images as well as multi-energy separation. However, due to the
novelty of this system, the body of literature is still comparably small.
DECT reconstructions used in the clinical routine
While each of the available DECT systems has its own advantages and disadvantages,
all systems enable material decomposition for iodine and linear blending of low and
high keV X-ray photon data.
Material decomposition exploits the fact that the attenuation of X-rays is dependent
on their energy and the composition of the absorbing material. At material decomposition,
the dependence of the photoelectric effect as well as Compton scattering, the two
main X-ray interactions in the diagnostic energy range, on the atomic number and the
physical density of a given material is modeled. This can be achieved by determining
the material-specific attenuation coefficient as a linear combination of the attenuation
of either two (i. e., two material decomposition) or three (i. e., three material
decomposition) base materials in a given voxel [7]. While material decomposition is advantageous in many instances, it is subject to
inherent limitations. First, materials with similar absorption characteristics show
a “spectral overlap”, which may result in erroneous depiction of a confounding material
in a material specific map (e. g., calcified structures in iodine maps). This spectral
overlap is influenced by different factors such as image noise and the acquisition
paradigm. Another limitation of material decomposition is that the presence of a confounding
material in a volume of interest may lead to erroneous quantification of the material
of interest, since the modeling assumptions are violated [9].
As iodine images selectively depict the iodine-associated component of X-ray attenuation,
they allow for absolute quantification of iodine in milligrams per milliliter [10]. The very same information can be subtracted from the conventional image, hence
making it possible to obtain virtual unenhanced (VUE) images that closely represent
quantitative and qualitative characteristics of a truly unenhanced acquisition [11]. A different way of visualizing iodine content is by reconstructing fused images
(similar to well-known PET/CT images): Here, the functional information (i. e., iodine
content) is color-coded and superimposed on top of morphological/anatomical reconstructions
(see the section on oncology applications). Material decomposition is also available
for uric acid as used in the imaging of gout and urinary stones (see the corresponding
sections below).
For virtual monoenergetic images (VMIs), linear bending of low- and high-energy information
is conducted or extrapolated. VMIs are considered to approximate attenuation characteristics
of images acquired with a true monoenergetic X-ray and are therefore indicated by
the corresponding, targeted keV level. VMI are available from as low as 40 keV up
to 140–200 keV (depending on the DECT technology). Generally, VMIs can be classified
according to their energy spectrum as low-energy VMIs (40–60 keV, VMI_low) that show
a greater soft-tissue and iodine contrast, and high-energy VMIs (> 90 keV, VMI_high)
that reduce beam hardening. VMIs in the mid-range (e. g., between 65 and 70 keV) have
been suggested as a working equivalent of polychromatic images at 120 kVp [12]
[13].
Other more investigational reconstructions include virtual non-calcium images, electron
density images, and Zeff images. While the latter two are thought to be useful for
radiation therapy planning, virtual non-calcium images subtract all calcium-associated
attenuation from the image (analogously to VUE for iodine) and therefore might prove
beneficial for depicting bone marrow edema or cellular infiltration in oncologic diseases.
Clinical applications
The following section will include applications of DECT-derived images for different
body regions and indications. At the end of each subsection, a short summary of feasible
clinical applications is provided.
Neuroradiology
Detection of intracranial hemorrhage and identification of a distortion of grey-white
matter differentiation are naturally among the key objectives of an unenhanced CT
scan of the head. However, the latter is often found to be challenging as differences
between grey and white matter are in the order of 10 HU. This in turn justifies and
requires a relatively high radiation dose to minimize image noise. In this instance,
low-energy VMIs have been reported to be instrumental, as they allow for an improved
contrast between grey and white matter ([Fig. 2]). Most previous publications investigating this topic concluded that virtual keV
levels in the range of 50–65 keV would be optimal for enhanced grey-white matter differentiation
[14]
[15]
[16]. Using such image reconstructions, a radiation dose reduction of up to 20 % appears
justifiable without image quality impairment [16]. In this context, the following thoughts require consideration: A) Some technological
DECT implementations, particularly earlier generations, demonstrate a notable increase
in image noise with photon energies decreasing towards 40 keV. A balance has to be
determined that ensures the increase in contrast is not compromised by an accompanying
increase in noise. B) When moving to a lower keV, blooming of calcific structures
is a well observed phenomenon. This possibly impairs assessment of the subcalvarial
space and hence requires careful consideration. While studies suggested both improvements
in quantitative image metrics and image quality in unremarkable cranial CT examinations,
some studies also denoted improved detection and delineation of edematous and/or hemorrhagic
areas [17]
[18]
[19]. Furthermore, low keV VMIs can improve vascular assessment, e. g., in the case of
suspected stroke. This is discussed in detail in the section “cardiovascular imaging”.
Fig. 2 Image examples showing improved gray-white matter differentiation (left), increased
contrast and conspicuity of intracerebral hemorrhage (center), and better delineation
of cerebral infarction (right) in dual-energy CT-derived virtual monoenergetic images
(VMI) at different energy levels compared to conventional images (CI).
Another common challenge in cranial CT examinations is the differentiation between
blood and iodine after endovascular treatment of ischemic stroke or aneurysm. For
this purpose, VUE images have been reported to be helpful as they eliminate iodine-associated
attenuation. Therefore, if hyperattenuating material remains in these images within
the brain parenchyma space, it is likely related to blood [20]
[21] as opposed to postinterventional contrast material accumulation. A meta-analysis
including 204 patients from nine studies on that topic concluded that DECT had a sensitivity
and specificity for differentiation between hemorrhage and contrast media or small
calcifications of 96 % and 98 %, respectively [22].
Besides the above-mentioned clinical applications of DECT for neuroradiology, a more
generic use that will be covered extensively in the “Musculoskeletal Imaging” section
is the reduction of artifacts. In the context of neuroradiology, DECT-derived metal
artifact reduction has been suggested to improve visualization in patients who underwent
clipping or coiling of intracranial aneurysms. High-energy VMIs facilitate the reduction
of artifacts. However, in the case of vascular assessment, a balance between the accompanying
decline in intravascular contrast and artifact reduction is key. Therefore, in this
specific case, the suggested optimal keV levels are not as high as those in musculoskeletal
applications and mostly range between 100 and 120 keV [23]
[24]
[25]. Other publications focusing on the assessment of the surrounding brain parenchyma
as opposed to vessels suggested higher keV levels accordingly [26].
Key clinical applications: Neuroradiology
-
Improved grey-white matter differentiation in low-keV VMIs (50–65 keV) with potential
for radiation dose reduction
-
Improved visualization of hypodense (e. g., ischemia) and hyperdense (e. g., hemorrhage)
brain lesions
-
Differentiation between intracranial hemorrhage and contrast media
-
Reduction of artifacts in patients who underwent intracranial coiling and/or clipping
(optimal keV levels need to be adjusted for assessment of the surrounding vessels
(lower) or the parenchyma (higher))
Musculoskeletal imaging
CT is one of the key modalities in acute trauma and fracture imaging. The DECT use
case in musculoskeletal imaging with the greatest clinical relevance and impact, for
which there is a large body of evidence, is the use of high keV irradiation VMIs to
reduce artifacts associated with orthopedic hardware. Moving towards higher energies
in VMIs decreases the probability for complete absorption of a photon and the risk
for an unsampled or undersampled region surrounding any implant resulting in hypodense
streaks. The value of this approach has been investigated in different use cases [27]
[28]. Research data suggests a benefit from VMIs with regards to artifacts caused by
nearly all types of implants ([Fig. 3a–c]). Most groups suggest implant-specific adjustments of keV levels in order to balance
the degree of artifact reduction versus the loss of soft-tissue contrast encountered
in high keV imaging [27]
[28].
Fig. 3 a Image example depicting artifact reduction in a patient with orthopedic hardware
in the lower lumbar spine by means of virtual monoenergetic images (VMI) at 200 keV.
Reduction of hypodense streak artifacts with improvement of the paravertebral compartments
can be clearly seen. b Artifact reduction in a patient with plate osteosynthesis following traumatic fracture
of the proximal humerus. Virtual monoenergetic images (VMI) at 200 keV effectively
reduce artifacts, allowing for an improved assessment of the remaining fracture lines
as well as the osseous tissue surrounding the fixating screws. c Different approaches to artifact reduction in a patient with right-sided, total hip
replacement. The conventional image depicts pronounced hyperattenuating artifacts
surrounding the implant and hypoattenuating artifacts superimposing the organs of
the lower pelvis (left side). Virtual monoenergetic images (VMI) at 200 keV show marked
reduction of artifacts with improved periprosthetic as well as pelvic assessment (center
image). The combination of VMI with an artifact reduction algorithm (OMAR) yields
the best possible reduction of artifacts and adequate visualization of the acetabular
bone as well as the lower pelvis (right side).
The effectiveness of VMI-enabled artifact reduction appears limited in very dense
materials (e. g., in hip protheses or endovascular coils). Here, a particular benefit
has been shown for combining high-energy VMIs and dedicated iterative reconstruction
algorithms for metal artifact reduction (i. e., MAR; [Fig. 3c]) [29]. This combination currently represents the strongest means for artifact reduction
in DECT imaging (together with optimized imaging protocols).
Another well-established application of DECT in musculoskeletal imaging employs material-specific
maps for uric acid in suspected (or clinically proven) gout arthritis [30]
[31]
[32]. These maps are typically evaluated in a fused manner that allows for visualization
of uric acid deposits. However, care has to be taken as tendon sheaths and/or cartilage
may mimic the attenuation characteristics of uric acid and, therefore, are indicated
as uric acid on such reconstructions ([Fig. 4]).
Fig. 4 A Uric acid overlay images showing gout adjacent to the tarsometatarsal joints (arrow)
and the proximal interphalangeal joint of the second toe. B Calcifications of the plantar aponeurosis, mimicking uric acid deposits.
Further suggested applications include an improvement in the imaging of vertebral
disc herniation [33] as well as the detection of bone bruise on CT imaging [34]
[35]
[36] (the latter by means of virtual non-calcium images). However, according to the opinion
of the authors of this article, these applications are not yet sufficiently validated
to allow broad clinical application outside of scientific studies and/or experienced
centers.
Key clinical applications: musculoskeletal imaging
-
Artifact reduction in patients with orthopedic implants using high-keV virtual monoenergetic
images exclusively or in combination with dedicated metal artifact-reduction algorithms
-
DECT-derived uric acid overlay images for the depiction of urate deposition in patients
with gout
Cardiovascular imaging
Following its capabilities of increasing iodine contrast by means of low keV VMI and
subtracting iodine-related attenuation components, DECT has been investigated for
various clinical use cases in the field of vascular imaging. For pulmonary embolism,
DECT-derived VMI have been shown to yield superior contrast and increase detection,
diagnostic certainty, and delineation of thromboembolism [37]
[38]
[39]. Those reconstructions are particularly helpful for salvaging the diagnostic assessability
of examinations with suboptimal contrast within the pulmonary arteries [40] ([Fig. 5]). Besides low-keV VMIs yielding superior vessel contrast, iodine maps can be used
for the assessment of pulmonary embolism by depicting areas of hypoperfusion within
the lung parenchyma. This can be of use both in the acute setting as well as in patients
with suspected chronic thromboembolic pulmonary hypertension [41]
[42].
Fig. 5 Patient with suspected pulmonary embolism. Conventional image shows severely impacted
visualization due to suboptimal contrast of the pulmonary arteries. Virtual monoenergetic
image (VMI) at 40 keV (right) salvages diagnostic quality by increasing the iodine
contrast in the pulmonary arteries, allowing confident delineation of pulmonary emboli.
One particular use case for DECT in vascular imaging that has been evaluated in the
literature is the assessment of aortic endografts. Here, iodine images and VMIs at
low energies yielded improved detection of endoleaks depicted as contrast media leakage
in the aneurysm sac [43]
[44]. Moreover, VUE images showed high sensitivity and specificity for endoleaks in aneurysm
graft assessment when used instead of true unenhanced images, enabling differentiation
between contrast media and hyper-attenuation caused by other causes e. g., calcifications
or hemorrhagic deposits. Chandarana et al. suggested that VUE and 80 kVp acquisitions
from a dual-source scanner derived from one 60-second phase might be sufficient for
endovascular aneurysm repair assessment [45]. Another important challenge for vascular imaging is the differentiation of contrast
material from hemorrhage, e. g., in the assessment of vascular hematoma, for which
VUE images may be helpful [46].
The capabilities of DECT to yield VUE and virtual high-contrast (i. e., low-energy
virtual monoenergetic) images has been suggested to be combined in a “virtual triphasic”
acquisition derived from one portal venous phase image, with significant potential
for radiation dose reduction due to the possible elimination of two acquisitions [47].
One potential application of DECT in cardiac imaging that has been investigated in
various studies is calcium scoring based on VUE images derived from CCTA, which, if
accurate, would allow for omission of an unenhanced acquisition and therefore a significantly
lower radiation dose. Whereas most studies described a good correlation between coronary
artery calcium scores, they also outlined a systematic underestimation of those metrics
in VUE [48]
[49]
[50]
[51], which can be attributed to the spectral overlap between calcium and iodine, resulting
in an erroneous subtraction of calcium-containing voxels in VUE images. As there is
no validated model taking into account this underestimation for routine clinical use,
further investigation in this regard seems necessary before routine clinical application
can be implemented.
Key clinical applications: cardiovascular imaging
-
Improved assessment of pulmonary embolism in low-keV VMIs
-
Restoring diagnostic image quality in angiographic phase examinations with suboptimal
contrast using low-keV VMIs
-
Assessment of contrast media leakage (e. g., in endograft assessment) and vascular
wall hematoma in low keV/iodine images and VUE images, respectively
Oncologic and organ-specific imaging
Liver
Accurate assessment of the liver parenchyma is paramount to determine the presence
or absence of liver metastases and safeguard optimal therapeutic concepts for cancer
patients. Due to the limited soft-tissue contrast on CT, the detection of small lesions
with subtle enhancement differences compared to the surrounding parenchyma can be
hampered. DECT-derived VMIs at low energy levels facilitate improved conspicuity of
liver lesions. For hyperattenuating liver lesions, this is achieved by increasing
the attenuation of the lesion itself, thereby improving the contrast-to-noise ratio
and its conspicuity and detectability [52]
[53]
[54]
[55], as schematically depicted in [Fig. 6]. For hypoattenuating lesions, a better depiction is attained by increasing the attenuation
of the surrounding liver parenchyma [56]
[57]. In summary, low-keV virtual monoenergetic images sent to the PACS as standard reconstructions
in oncologic patients are an easy-to-use screening tool for small, subtle, hypo- and
hyperattenuating liver lesions.
Fig. 6 Schematic diagram showing increased contrast and conspicuity of hypo- and hyperattenuating
liver lesions attained in low keV virtual monoenergetic images. Improved delineation
of hypoattenuating lesions is achieved by increased contrast of the surrounding parenchyma,
whereas delineation of hyperattenuating lesions is improved by increased attenuation
of the lesion itself.
Small, intermediate lesions of the liver, which are often called “too small to characterize”,
present another diagnostic challenge in regard to CT liver imaging. DECT-derived iodine
images may help to increase diagnostic confidence when diagnosing such lesions, for
example by ruling out any iodine uptake in iodine images. One study found that the
classification of such small, intermediate hypoattenuating liver lesions could be
improved significantly with the use of iodine images and an iodine-based threshold
as compared to conventional images [58]. However, large-scale validation of this quantitative application seems necessary
before broad clinical application can be implemented. The same applies to studies
that suggested DECT-enabled differentiation of liver lesions, e. g., between small
HCCs and metastases based on iodine quantification, which has been investigated only
in small cohorts [59]
[60].
Kidneys
Renal masses with non-cystic attenuation of equal or greater than 20 HU on contrast-enhanced
abdominal CT are a common incidental finding [61]. One study suggested that in > 50 % of adults aged 50 or older, imaging would yield
at least one incidental renal mass [62]. The differential diagnoses for such masses include hemorrhagic or proteinaceous
cysts as well as solid-enhancing masses. Differentiation between these lesions in
monophasic, portal venous phase single-energy CT is hampered by the lack of information
on the baseline unenhanced attenuation enabling assessment of lesion enhancement.
Consequently, many patients diagnosed with such incidental lesions are recommended
to undergo additional multiphasic renal protocol CT or MRI examination. By means of
material decomposition, DECT allows differentiation between the attenuation component
derived from iodinated contrast material and other non-iodine-related tissue or material
components such as protein or blood. This has resulted in a large number of studies
that aimed to assess the utility of DECT in characterizing renal lesions, which investigated
both the use of virtual unenhanced and iodine images. A meta-analysis of five studies
with a total patient number of 367 revealed a sensitivity and specificity of 96.6
and 95.1 %, respectively, for iodine images with respect to determining lesion enhancement
[63]. In another study, Meyer et al. showed that virtual unenhanced images allow for
accurate renal lesion characterization, albeit with a lower specificity compared to
true unenhanced images [64]. Accordingly, Cao et al. recently found that virtual enhancement calculated from
the portal venous phase and virtual unenhanced images correctly identified all enhancing
lesions within their study cohort, yet noted that in some cases, virtual unenhanced
images diverged from true unenhanced image attenuation with regard to the evaluation
of cystic lesions [65]. In the clinical routine, virtual unenhanced and iodine images can be used synergistically
to rule out enhancement in ambiguous renal lesions. In this instance, previously investigated,
scanner-dependent lower limits of detection and quantification of iodine should be
consulted when aiming to assess the presence of iodine enhancement [66]. We suggest that the current body of evidence supports using DECT for determining
or ruling out enhancement of hyperattenuating renal lesions. More advanced quantitative
applications such as differentiation of different subtypes of RCCs have been investigated
[67]
[68]. However, the corresponding thresholds have not been validated and are prone to
a certain variability of iodine quantification [69]
[70].
Another, and one of the earliest use cases of DECT in renal imaging is the characterization
of renal calculi. Significant evidence is available that indicates that identification
of uric acid concrements is possible by means of material-specific maps for uric acid
[71]
[72]
[73]
[74]. Furthermore, several groups investigated whether stone composition analysis is
possible beyond this. However, the body of literature is too small and furthermore
too scanner- and protocol-specific to be able to consider this a generally valid method
in this context [75]
[76].
Adrenal glands
Adrenal nodules are one of the most common incidental findings in abdominal CT with
a reported prevalence of up to 7 % [77]. In the overall population, the majority of these adrenal nodules are known to represent
benign adrenal adenomas. However, in cancer patients, there is an increased risk that
these lesions represent metastatic spread to the adrenal glands. Lipid-rich adenomas
are characterized by the presence of macroscopic fat. This can be determined by an
unenhanced CT attenuation threshold of 10 HU, which has been suggested to yield a
sensitivity/specificity of 71 %/98 % [78], or by a signal loss on chemical shift MRI. It has been described that overall,
around 70 % of adenomas can be determined as lipid-rich following an unenhanced attenuation
of < 10 HU [79]
[80]
[81]
[82]. However, in most instances, adrenal incidentalomas are detected in CT examinations
not tailored for such an assessment due to a lack of unenhanced or delayed phase image
acquisitions, based on radiation protection considerations. Consequently, definitive
characterization and exclusion of metastatic spread would require additional imaging
such as dedicated CT for adrenal assessment comprising an unenhanced and delayed phase,
or MRI with in-and-out-of-phase imaging. DECT-derived VUE and iodine images have been
extensively investigated scientifically for one-stop-shop characterization of incidental
adrenal nodules in portal venous phase abdominal CT examinations [83]
[84]
[85]
[86]
[87]
[88]
[89]. As discussed before, VUE images allow for estimation of the true unenhanced image
attenuation. Therefore, a VUE attenuation of < 10 HU calculated from portal venous
phase images is highly indicative of lipid-rich adrenal adenoma. One problem that
has been addressed in the literature is that virtual unenhanced image attenuation
shows a tendency towards overestimation of true unenhanced image attenuation in adrenal
adenomas [89]
[90]
[91]
[92], limiting the sensitivity in determining their lipid-rich nature. Different ways
of mitigating this issue have been suggested, most notably increasing the 10 HU threshold.
Another means of addressing this issue has been suggested, i. e., including iodine
measurements in a multiparametric approach, which yielded promising results. However,
these monocentric, retrospective results have not been validated yet [89]. As virtual unenhanced attenuation is known to show intra-patient variation between
different DECT platforms, state-of-the-art, device-specific data should be consulted
when using quantitative DECT measurements as stated above for the characterization
of incidental adrenal nodules [93]. [Fig. 7] illustrates an example of adrenal differentiation using VUE images.
Fig. 7 Virtual unenhanced images derived from dual-energy CT for differentiation between
adrenal adenoma (upper row) and adrenal metastasis (lower row).
Metastasis detection
With its large overall volume and high physical density, the skeletal muscle is notoriously
hard to assess on conventional CT, qualifying it as one of the “blind spots” of oncologic
staging and follow-up CT. Due to several underlying factors, the human skeletal muscle
is relatively resistant to hematogenous metastatic spread of oncologic diseases. However,
for some malignant diseases such as malignant melanoma, a higher probability of skeletal
muscle metastasis has been described [94]. In oncologic imaging, the detection of such distant metastases is important since
the new appearance of metastatic lesions may indicate the necessity for the adjustment
of treatment concepts. Several studies have described an improved sensitivity for
skeletal muscle metastases in DECT-derived reconstructions highlighting iodine contrast,
such as low keV virtual monoenergetic images and iodine overlay images, in which the
iodine content is color-coded and merged with underlying grey-scale conventional images
[95]
[96].
The underlying rationale is that images highlighting the iodine-related attenuation
allow for the differentiation of iodine-related attenuation components of enhancing
metastatic lesions vs. physical density-related attenuation components of circumjacent
muscle, which may be almost indistinguishable on conventional images ([Fig. 8]). Therefore, these images may qualify as screening tools for patients with oncologic
diseases or disease stages, respectively, that yield a higher pre-test probability
for metastatic spread to the muscle. Finally, in a prospective evaluation of patients
with occult cancer, it was described that spectral reconstructions may improve the
confidence of radiologists in characterizing lesions, thereby making it possible to
minimize correlative imaging [97].
Fig. 8 Simplified schematic illustration of attenuation components of skeletal muscle metastases
and surrounding muscle parenchyma including an exemplary case of skeletal muscle metastasis
on conventional and iodine overlay images in a patient with melanoma.
Other applications
The substantially increased attenuation of iodinated contrast material in low keV
virtual monoenergetic reconstructions can not only be used to increase conspicuity
and detectability of organ-specific lesions as described above, but also to lower
the overall amount of contrast media that has to be applied for attaining sufficient
visualization ([Fig. 9]). In a study by Nagayama et al., it was shown that VMIs from 40 to 55 keV allowed
for a reduction of contrast media dose by 50 % in multiphase liver CT [98]. In another study, VMIs at 40 keV were reported to yield organ and vessel assessment
comparable to a full-dose protocol in portal venous phase abdominal CT at a 50 % lower
iodine concentration load [99]. In regard to aortic angiography with 50 % contrast material reduction, DECT-derived
VMIs ranging from 40 to 60 keV were reported to yield similar image quality as normal
iodine contrast material doses [100]. Lastly, Agrawal et al. reported a significantly higher attenuation and CNR in low-energy
VMI CTA with a reduction of contrast media to 24 g/l compared to standard-dose CTA
(33 g/l) [101]. Based on the above-mentioned evidence, a minimum volume of 50 ml contrast media
(depending on the specific contrast media administration protocol (i. e., bolus length))
in combination with low keV VMI appears to be a valid strategy in patients at an increased
risk for contrast media-associated nephropathy or other dose-dependent adverse events.
Fig. 9 Cross-sectional image of a portal venous phase CT scan of the upper abdomen with
a contrast media reduction of 50 %. The left image represents the conventional image,
whereas the right image with only poor contrast enhancement depicts the corresponding
virtual monoenergetic image at 40 keV, restoring adequate tissue contrast.
Key clinical applications: Oncologic and organ-specific imaging
-
Liver:
-
Kidneys:
-
Adrenals: Identification of incidental adenoma with virtual unenhanced images (consult
literature for scanner-specific, adjusted thresholds)
-
Detection of skeletal muscle metastases in iodine overlay images in patients with
increased pretest probability
-
Contrast media reduction with low-keV virtual monoenergetic images to antagonize iodine
contrast deterioration
