Keywords
dual-energy computed tomography - diffuse liver disease - chronic liver disease -
hepatic steatosis - liver iron - liver fibrosis
Introduction
Dual-energy computed tomography (DECT) is a significant advancement in the field of
computed tomography (CT), where images are acquired at two different energies (kilovolts)
and “material decomposition” algorithms are used to differentiate and quantify various
materials with different effective atomic numbers such as water, fat, iron, and iodine
in a given pixel. The concept of dual-energy acquisition and its potential benefits
were first described in 1973 by Hounsfield in his original article on CT systems.[1] However, a lot of technical difficulties had to be overcome before his vision was
realized when the first clinical DECT scanner was introduced in 2006.[2] Since then, all major CT manufacturers have introduced DECT scanners with different
machine designs and image processing softwares.
Diffuse liver disease, particularly cirrhosis, is a growing public health problem
worldwide and is associated with significant morbidity and mortality.[3] Liver biopsy is currently the gold standard for diagnosis (presence of inflammation,
fibrosis, fat, iron) and staging of chronic liver parenchymal diseases.[4] However, it is invasive, associated with hemorrhagic complications, and subject
to sampling errors and inter-observer variations.[5]
[6] A repeat biopsy to monitor the course of the disease and assess response to treatment
is unjustified and has low acceptance among patients. Hence, there is a need for a
reliable, reproducible, and non-invasive alternative. Magnetic resonance imaging (MRI)
with its versatile techniques and sequences is increasingly becoming a well-established
and accurate imaging tool for the purpose of assessing fat, fibrosis, and iron deposition.[7]
[8]
[9] However, MRI is expensive, takes a long time to scan compared to CT or ultrasonography,
and requires special sequences and hardware. Although the role of DECT in diffuse
liver diseases is evolving, this article aims to review the most relevant literature
on the use of DECT in the assessment of diffuse liver diseases, mainly the deposition
of fat, fibrosis, and iron. We would also briefly discuss the basic principles of
DECT image acquisition and image reconstruction with a focus on the areas that are
relevant in the assessment of diffuse liver diseases.
Basic Principles of DECT
In CT, the photoelectric effect and Compton scattering are the dominant X-ray interactions
influencing image formation.[10] The attenuation of X-rays in each voxel is a function of the linear attenuation
coefficient of the medium and energy of incident X-rays. The linear attenuation coefficient
in turn depends on the effective atomic number of the medium and its material density.[11]
[12] On a single-energy CT (SECT), materials with different atomic numbers may have the
same CT attenuation value (Hounsfield unit [HU]) if the material densities are different.
Hence, SECT cannot differentiate them. In DECT, using a second X-ray spectrum, two
image datasets are produced, one from low energy and the other from high energy. The
degree of attenuation change (attenuation coefficient) with a change in X-ray energy
is independent of material density and is material specific. Materials such as fat,
iodine, and bone have different degrees of attenuation change and can be identified
based on this difference ([Fig. 1]). With an appropriate software algorithm, these material-specific energy-dependent
attenuation coefficients can also be used to quantify the mass density of two or three
materials in a mixture of known material composition.[13]
[14] Typically, the low-energy beam is 80 to 100 kVp (kilovoltage peak) and the high
energy beam is 120 to 150 kVp. To generate the DECT image data, different approaches
are followed, with modifications done either at the X-ray tube level or the detector
level. These vendor-specific techniques of DECT scanners are described elsewhere in
this issue.
Fig. 1 Principle of dual-energy CT material decomposition algorithm to identify and quantify
different materials. The graph shows a change in mass attenuation coefficients of
iodine, bone (calcium), and fat at a wide range of X-ray energies. At specific X-ray
energy, mass attenuation coefficient (linear attenuation coefficient/density) is only
dependent on the effective atomic number of the medium. A difference in this coefficient
at two different energy X-rays (for example 80 kVp and 140 kVp as shown in the graph)
depends on the effective atomic number of the medium and not the medium density.
Although the post-processing software of DECT is vendor-specific, in general, the
workflows are not very different from each other. After the acquisition, the DECT
images (either non-contrast or contrast-enhanced depending on the indication for which
DECT is being performed) are transferred to the DECT workstation and loaded onto the
dual-energy software provided by the vendors. Various material-specific algorithms
are inbuilt in the DECT software. The technique of quantification of a material is
usually simple ([Fig. 2]). The images are opened in the material-specific application and a region of interest
(ROI) is drawn over the liver parenchyma avoiding major vessels. The ROI displays
the concentration of the material, e.g., iodine or fat, being evaluated. Multiple
ROIs may be drawn in different lobes and segments (at least three ROIs in each lobe)
of the liver to get a mean of the values for reliable and reproducible results. Color-coded
maps can also be generated for providing an overview of the distribution of the material
in the liver parenchyma.
Fig. 2 Overview of dual-source dual-energy computed tomography (DECT) post-processing software
(Syngo.via, version VB10B, 2016; Siemens Healthineers). Based on the material being
evaluated, an appropriate application profile (in the left upper corner box) is chosen.
Liver virtual non-contrast (VNC) is selected in this image to measure liver iodine
concentration on the delayed phase DECT, for fibrosis quantification. The region of
interest placed on the axial image shows various parameters including Iodine density
and fat fraction in that region. The same application can be used for liver fat quantification.
Color-coded iodine maps are seen in three different planes.
Hepatic Steatosis
Hepatic steatosis (HS) or fatty liver, currently is most commonly associated with
non-alcoholic fatty liver disease (NAFLD).[15] The other common causes of HS are alcohol consumption,[16] hepatitis C,[17]
[18] and use of certain medications including many chemotherapeutic agents[19]
[20]
[21] and corticosteroids.[22] The estimated prevalence of HS is ∼25% worldwide.[23] NAFLD is frequently associated with metabolic syndrome, which comprises a set of
risk factors for cardiovascular disease and type II diabetes that include obesity,
hypertension, elevated triglycerides, and insulin resistance.[24]
[25]
Quantification of hepatic fat is important in patients with HS, as increasing steatosis
leads to the development of steatohepatitis and then to cirrhosis with increased risk
for the occurrence of hepato-cellular carcinoma and associated with poor liver surgical
outcomes.[26]
[27]
[28] However, with appropriate therapeutic strategies such as lifestyle modifications
(dietary regulation and physical exercise) and drugs (pioglitazone and vitamin E),
hepatic steatosis can be reversed along with the incidence of its metabolic complications.[29] Hence, we need a reliable non-invasive modality for diagnosis, prognostication,
and treatment monitoring in patients with HS. Chemical shift relaxometry and spectroscopy
on MRI and backscatter coefficients or controlled attenuation parameters on ultrasonography
(USG) are various imaging methods that can be used for liver fat quantification.[30] MR spectroscopy (MRS) has proven to be the most accurate imaging tool and has the
potential to replace biopsies.[31] However, MRS can only quantify fat in a small area at a time and multiple measurements
to cover the entire liver is a time-consuming process. Because of these reasons, MRS
has been replaced by chemical shift imaging-derived proton density fat fraction (PDFF)
as an imaging gold standard for liver fat quantification.[9]
[32] However, the higher cost of MRI scan is the major limiting factor for its routine
use. Ultrasound elastography is dependent on reference phantoms, prone to measurement
errors especially in overweight female patients, and does not measure fat accurately
as the values are influenced by co-existing fibrosis.[33] However, the controlled attenuation parameter obtained on transient elastography
(Fibroscan) has shown promising results in quantifying liver fat.[34]
[35] In SECT, semiquantitative fat measurements can be performed using the attenuation
value of the liver (HU), preferably on an unenhanced scan.[36]
[37] However, absolute liver attenuation values can be affected by various factors such
as CT acquisition parameters and the patient's cardiac and renal function. To overcome
these, several authors have compared the attenuation of the liver to that of other
abdominal organs and described various indices such as liver attenuation index (difference
between liver and spleen attenuation)[38]
[39]
[40] and L/S ratio (ratio of liver to spleen attenuation),[39]
[41] which produced better results. However, attenuation values on SECT are still influenced
by the presence of iodinated contrast material or liver iron. Further, SECT is less
sensitive in the detection of early HS (sensitivity ranges from 52 to 62% in cases
with 10 to 20% fat).[42]
In DECT, with the help of the “material decomposition” algorithm, it is possible to
not only identify but also quantify the liver fat in a defined region ([Figs. 3] and [4]).[12] A few studies have evaluated the role of DECT in the detection and quantification
of liver fat ([Table 1]).
Table 1
Brief details and summaries of results of studies evaluating dual-energy computed
tomography in hepatic steatosis
|
Study (y)
|
Sample size
Cases/controls
|
Modality of comparison
|
DECT scanner type
|
DECT energies
|
Key results
|
|
Mendler et al
(1998)
|
16/11
|
DECT (non-contrast) versus USG and SECT
|
GE Pace-Plus
|
80, 140 kV
|
• Good correlation between liver –spleen attenuation difference on SECT and histology
(r = 0.74)
• No significant correlation between ∆HU on DECT and histology
• USG superior to SECT and DECT in diagnosing fatty liver
|
|
Zheng et al.
(2013)
|
32/20
|
DECT versus USG and SECT
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• Good correlation between DECT fat percentage and liver spleen attenuation difference
on SECT (r = 0.98)
• DECT superior to SECT, with its distinct mono-energetic spectral curves with different
steatosis grades and ability to delineate areas of fat infiltration on subtraction
images
|
|
Patel et al
(2013)
|
107/229
|
DECT (late arterial phase) and SECT
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• Good correlation between liver spleen attenuation difference on SECT and DECT fat
fraction (r = 0.74, p < 0.001)
• Fat quantification on contrast enhanced DECT accurate (AUROC = 0.85)
|
|
Kramer et al
(2017)
|
50/0
|
DECT (non-contrast), MR-PDFF, USG and elastography versus MRS
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• Moderate correlation between DECT fat fraction and MRS (r
2 = 0.42 for anterior liver segments, r2 = 0.26 for posterior liver segments)
• Excellent correlation between SECT attenuation and MR-PDFF with MRS (r
2 = 0.86 and 0.99, respectively)
• Ultrasound elastography showed poor correlation with MRS (r
2 = 0.004)
|
|
Hyodo et al
(2017)
|
29/4
|
DECT (late arterial, portal venous, and delayed phases) versus histology and MRS
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• DECT (unenhanced and contrast-enhanced) correlated well with histology
• No significant difference in DECT fat fraction of unenhanced, arterial, portal venous,
and delayed phases (not more than 2%)
• In differentiating steatosis grade 0 from 1–3, AUROC for DECT and MRS were 0.88
and 0.89, respectively.
|
|
Corrias et al
(2021)
|
39/0
|
DECT (late arterial phase) versus MR-PDFF
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• No significant difference in DECT fat fraction and MRI-PDFF (two readers with correlation
coefficients, 0.84 and 0.85, p < 0.05)
• Good inter-reader agreement for both techniques. (kappa coefficient - 0.88 for MRI-PDFF,
0.97 for DECT-FF, p < 0.05)
|
Abbreviations: ∆HU, difference in attenuation between low- and high-energy images;
AUROC, area under receiver operating characteristic; DECT, dual-energy computed tomography;
MR-PDFF, magnetic resonance proton density fat fraction; MRS, magnetic resonance spectroscopy;
r, correlation coefficient; r
2, correlation of determination; SECT, single energy CT; USG, ultrasonography.
Fig. 3 Measurement of hepatic fat content on dual-source dual-energy computed tomography.
(A, B) Axial virtual non-contrast (VNC) image of a patient (A) with normal liver parenchyma with two circular regions of interest (ROIs) drawn
on the liver show close to zero fat fraction and corresponding color-coded fat map
(B) shows almost no fat in the liver parenchyma. (C, D) Axial VNC image (C) of another patient with non-alcoholic fatty liver disease with two ROIs show 20%
of fat fraction and corresponding color-coded fat map (D) shows the distribution of fat in the liver parenchyma.
Fig. 4 Measurement of hepatic fat content on rapid switching dual-energy computed tomography.
(A) Axial, contrast-enhanced image of the patient with fatty liver. (B) Virtual unenhanced image at the same level with two circular regions of interest.
Mean attenuation was 32 HU and fat fraction was 15.5%. (C) The corresponding color-coded fat map shows the distribution of fat in the liver
parenchyma.
The oldest study was by Mendler et al that showed that DECT-∆HU (∆HU–difference in
attenuation between low-energy and high-energy images) generated using the sequential
technique, is inferior to both SECT liver–spleen attenuation difference and ultrasound.[43] Hyodo et al reported that the area under the receiver operating characteristic (ROC)
curve of DECT was 0.88 (95% confidence interval: 0.74–0.98) in the detection of fatty
liver with more than 5% fat in needle biopsy tissue.[44] This was comparable to MRS (area under the ROC curve = 0.89). A study by Kramer
et al showed that non-contrast DECT had a moderate correlation with MRS (r
2 = 0.423), whereas SECT had an excellent correlation with MRS (r
2 = 0.856) in the quantification of hepatic fat.[31] The authors concluded that non-contrast DECT does not improve the accuracy of fat
quantification over conventional SECT. The reason for this was that material decomposition
is less accurate in the differentiation of fat and water on non-contrast DECT compared
with iodine and water on contrast-enhanced DECT. Corrias et al compared DECT and PDFF
on MRI in quantifying fat in patients with post-chemotherapy HS.[45] They found no statistically significant difference between these two modalities
in fat quantification and showed an excellent inter-reader agreement of DECT (kappa = 0.97,
p < 0.05). Patel et al evaluated liver fat density on contrast-enhanced DECT using
spleen–liver attenuation difference < 1 HU on non-contrast SECT as the reference standard
for steatosis. The DECT fat measurement correlated well with the reference standard,
with an r-value of 0.74 (p < 0.001) and area under receiver operator curve of 0.847.[46] The role of DECT in fat quantification was also evaluated in several phantom studies
and animal studies with mixed results.[47]
[48]
[49]
[50]
[51]
[52]
[53]
Recent studies have shown that, unlike SECT, material decomposition on DECT can accurately
quantify liver fat even on contrast-enhanced images.[44]
[45]
[46] This is also supported by other studies which have shown high fidelity of HU values
measured on virtual non-contrast (VNC) images compared with unenhanced CT within the
liver and other abdominal organs.[54]
[55]
[56]
[57] This is invaluable because most fatty livers are incidental on post-contrast images
and plain scans are not acquired routinely. Hyodo et al suggested that among contrast
phases, an equilibrium phase is the best alternative to non-contrast CT in liver fat
quantification on DECT.[44]
DECT is a valid technique in quantifying liver fat and can be used reliably in all
suspected cases of hepatic steatosis. It can be seamlessly integrated into abdomen
CT scan protocols for liver fat quantification. Although a non-contrast scan is preferred,
with the use of the latest multi-material algorithms, fat can be reliably quantified
even on contrast-enhanced CT scans.
Liver Fibrosis
Liver fibrosis is the common underlying pathological process in chronic liver diseases,
regardless of the cause. This results from the excessive accumulation of the extracellular
matrix.[58] Progression of liver fibrosis parallels with the increase in portal pressure, which
worsens the prognosis due to the associated complications.[58] Hence, assessment of the severity of liver fibrosis has direct implications on the
treatment, follow-up planning, and risk stratification in patients with chronic liver
disease.[3]
[59] For instance, in chronic viral hepatitis, the presence of bridging fibrosis on histology
is a strong factor for the recommendation of antiviral therapy, whereas the presence
of cirrhosis calls for an endoscopy for esophageal varices and also screening for
hepatocellular carcinoma.[4] This clinical significance of detection and grading of liver fibrosis has led to
an active search for a non-invasive modality to quantify fibrosis in these patients.
Ultrasound elastography (transient and shear-wave elastography) techniques are simple,
inexpensive, and accurate. However, their measurements are affected by the presence
of ascites, obesity, biliary obstruction, acute inflammation, fatty liver, and narrow
intercostal spaces.[60]
[61]
[62] MR elastography is the most accurate non-invasive modality in quantifying liver
fibrosis and correlates closest to the histological staging.[60]
[63] Nonetheless, in spite of these advantages, the high cost, lack of accessibility,
the requirement of dedicated hardware, and confounding effects by coexisting iron
accumulation preclude its routine use.
DECT cannot directly detect or quantify fibrosis in the liver on non-contrast CT as
it shows a similar attenuation pattern as the rest of the parenchyma. Several studies
have attempted to quantify fibrosis by measuring extracellular volume fraction using
liver attenuation values on contrast-enhanced delayed phase SECT with good correlation
with histology.[64]
[65]
[66]
[67] The timing of the delayed phase CT in these studies was heterogeneous, ranging from
3 to 30 minutes. This is based on the hypothesis that fibrotic areas in the liver
retain more contrast material in the delayed phase due to the increased extracellular
space due to matrix deposition and reduced perfusion.[66]
[68] This contrast retention is reflected by the concentration of iodine in the liver
that can be readily measured on DECT. Therefore, the concentration of the iodine in
the liver on delayed phase DECT directly correlates to the concentration of fibrous
tissue in the liver. Lv et al[69] evaluated the role of portal venous phase (obtained 30 seconds after the arterial
phase was acquired by bolus triggering) DECT in 38 patients with cirrhotic livers
(Child-Pugh class A/B/C: n = 10/14/14) and 43 patients with healthy livers. In differentiating healthy liver
from a cirrhotic liver, a cut-off value of 0.52 of the normalized iodine concentration
(normalized iodine concentration [NIC], which is the ratio of iodine concentration
of liver to that of the aorta) in the portal venous phase showed high sensitivity
(95%), but relatively lower specificity (61%). The iodine concentration (IC) ratio
(iodine concentration of liver in the arterial phase to that in the portal vein phase)
with a cut-off at 0.3 showed high sensitivity (79%) but low specificity (49%). However,
a combination of these two parameters showed high sensitivity (77%) and specificity
(87%).
Lamb et al[12] found that NIC of liver on a 3-minute delayed phase DECT strongly correlated with
Ishak fibrosis stage (p < 0.05). Fuentes et al,[70] in their retrospective study on 12 participants, showed a strong correlation between
NIC in the delayed phase and Ishak fibrosis stage (r = 0.77, p < 0.01). Sofue et al[71] in a 47-patient cohort showed that NIC had a good correlation with the METAVIR fibrosis
stage (r = 0.65, p < 0.001). The area under the receiver operating curve (AUROC) ranged from 0.79 to
0.86 in discriminating each liver fibrosis stage (≥F1–F4). The NIC cut-off values
for the diagnosis of METAVIR stages ≥F1, ≥F2, ≥ F3 and =F4 were found to be 0.270,
0.274, 0.286, and 0.299 with accuracies of 74.5%, 78.7%, 78.7%, and 76.6%, respectively.
In a recent prospective study evaluating 107 patients of chronic liver disease by
Marri et al[68] in our institute, the NIC on the 5-minute delayed phase DECT was compared with the
METAVIR stage on histology ([Fig. 5]). The NIC showed a high positive correlation with METAVIR fibrosis stage (r = 0.81; p < 0.0001). The AUROC for the NIC in the right lobe of the liver with respect to each
METAVIR stage ranged from 0.86 to 0.96. The cut-off values (with sensitivity and specificity)
of NIC for different fibrosis stages were ≥F1 = 0.243 (85.1%, 83.3%), ≥F2 = 0.289
(83.7%, 81.4%), ≥F3 = 0.343 (86.9%, 86.8%), =F4 = 0.401 (93.3%, 84.7%), respectively.
A 5-minute delayed DECT was used in this study in contrast to the previous studies,
where a 3-minute delay was used. This longer delay had resulted in a better performance
of DECT in fibrosis quantification and fewer overlapping NIC cut-offs for the METAVIR
stages. DECT performance was also compared with elastography in this study. NIC showed
good agreement with liver stiffness as measured with transient elastography and shear-wave
elastography (Spearman r = 0.60 and 0.64, respectively; p <0.001). An overview of some of the studies is shown in [Table 2].
Table 2
Brief details and summaries of results of studies evaluating dual-energy CT in hepatic
fibrosis
|
Study (year)
|
Sample size
Cases/control
|
Modality of comparison
|
DECT Scanner type
|
DECT energies
|
Key results
|
|
Lv et al
(2012)
|
38/43
|
DECT - Portal venous phase NIC and IC ratio with Child-Pugh score
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• NIC cut-off of 0.52: high sensitivity (95%) but relatively low specificity (61%).
• IC ratio with cut-off of 0.3: high sensitivity (79%) but low specificity (49%).
• High sensitivity (77%) and specificity (87%) for combination of NIC and IC ratio
in differentiating healthy liver from the cirrhotic liver.
|
|
Lamb et al
(2015)
|
7/5
|
DECT - 3-minute delayed phase liver NIC with histology
|
GE Discovery CT 750 HD
|
80, 140 kV
|
• NIC on DECT correlated with histology and allowed statistically significant stratification
of patients by severity of fibrosis (p < 0.05 for all stages)
• On two different time point scans, NIC showed a high degree of repeatability (r
2 = 0.93)
|
|
Sofue et al
(2018)
|
47/0
|
DECT - 3-minute delayed phase liver NIC with histology
|
GSI, GE Discovery CT 750 HD
|
80, 140 kV
|
• NIC showed moderate correlation with METAVIR stage (r = 0.63, p < 0.001)
• In discriminating each fibrosis stage, AUROC for NIC ranged from 0.80 to 0.86
|
|
Marri et al
(2021)
|
107/50
|
DECT - 5-minute delayed phase liver NIC and sonoelastography with histology
|
SOMATOM DF, Siemens
|
100, 140 kV
|
• NIC showed strong correlation with METAVIR stage (r = 0.81, p < 0.001)
• In discriminating each fibrosis stage, AUROC for NIC ranged from 0.86 to 0.96
• NIC showed good agreement with liver stiffness as measured with transient elastography
and shear wave elastography (r = 0.60 and 0.64, respectively, p < 0.001)
• 5-minute delayed phase showed better sensitivity and specificity compared with 3-minute
delay, which was used in previous studies
|
Abbreviations: ∆HU, difference in attenuation between low- and high-energy images;
AUROC, area under receiver operating characteristic; DECT, dual-energy CT; GSI, gemstone
imaging; IC, iodine concentration; NIC, normalized iodine concentration; r, correlation coefficient; r
2, correlation of determination.
Fig. 5 Assessment of hepatic fibrosis on dual-source dual-energy computed tomography (DECT)
in a 40-year-old man with chronic hepatitis B cirrhosis. (A) Axial 5-minute delayed phase DECT image with circular regions of interest (ROIs)
in both lobes of the liver. Normalized iodine concentration (NIC) in the right and
left lobes were 0.571 and 0.539, respectively. (B) The iodine map of the delayed phase shows heterogeneous iodine uptake in the liver.
(C) Histology of the biopsy sample showed a METAVIR stage of F4. The NIC values were
consistent with cut-offs suggested by a recent study by Marri et al.
A significant number of patients with chronic liver disease routinely undergo multiphase
CT scans of the abdomen as part of hepatocellular carcinoma surveillance. The possibility
of grading the severity of fibrosis on the same scan would be invaluable. The addition
of dual-energy acquisition in the contrast-enhanced delayed phase, preferably 5-minute,
is suggested for this purpose ([Fig. 6]). This could reduce the number of biopsies and related complications in these patients.
Simultaneously, larger studies are needed to formulate a uniform CT acquisition protocol,
delayed phase timing, and NIC cut-off values in each fibrosis stage in these patients.
Fig. 6 Assessment of hepatic fibrosis on rapid switching dual-energy computed tomography
(DECT) in a 46-year-old man with alcoholic cirrhosis. (A) Axial 5-minute delayed-phase DECT image with circular regions of interest (ROIs)
in both lobes of the liver. Normalized iodine concentration in the right and left
lobes were 0.490 and 0.513, respectively, suggesting a stage of F4, according to Marri
et al. (B) Iodine map from the delayed phase shows heterogeneous iodine uptake in the liver.
Hepatic Iron
Liver iron deposition and chronic liver disease are interrelated. Deposition of iron
in the liver parenchyma can be due to increased total body iron as seen in hereditary
hemochromatosis or transfusion-related hemosiderosis.[72] Conversely, liver diseases such as chronic viral hepatitis, alcoholic liver disease,
and non-alcoholic steatohepatitis (NASH) also promote iron deposition in the liver.[73]
[74] Increasing iron deposition in the hepatocytes causes liver damage by oxidative stress,
eventually leading to cirrhosis and the possibility of the development of hepatocellular
carcinoma (HCC).[75] Hepatic iron deposition in chronic liver disease is also independently associated
with increased mortality.[76] Hence, liver iron detection and quantification are crucial for the management of
these patients.
MRI is widely considered the primary modality for non-invasive liver iron quantification.
Chemical shift imaging, the signal intensity of liver relative to spleen on fast-spin
echo T2-weighted sequence, liver to muscle signal intensity ratio on multi-echo gradient
recalled echo (GRE) sequence, and R2 and R2* relaxometry are the various MRI techniques
used in accurate estimation of liver iron content.[77]
[78] Among these, R2 relaxometry is the most validated and often considered as the gold
standard.[77] However, R2* relaxometry is more practical because the GRE sequence is faster, results
are reliable, and simultaneous acquisition of fat and iron is possible in a single
breath-hold.[78] Apart from the high cost and inaccessibility to MRI machine and post-processing
software, the variability of the results in the presence of high concentrations of
liver iron is a limitation.
On DECT, unenhanced CT is preferred for iron quantification. Liver iron is quantified
either by using the attenuation difference of the liver between the low- and high-energy
CT images (∆HU) or using an iron-specific material decomposition algorithm, which
gives virtual iron content based on the attenuation difference ([Table 3]; [Fig. 7]). Joe et al in their retrospective study showed that DECT-∆HU (difference of liver
attenuation between 80 and 140 kVp) correlates well with the degree of iron accumulation
(r = 0.43, p < 0.001).[79] The mean ΔH was 13.53 in patients with clinically important (≥ 10%) iron accumulation
compared with 7.39 in normal patients. The DECT-∆HU showed similar excellent performance
in diagnosing iron overload compared with MRI. The concurrent presence of liver fat
did not affect the DECT performance in this study. Luo et al found that virtual iron
concentration on DECT showed a significant correlation with R2* and MRI measured liver
iron content (r = 0.89 and 0.87, respectively; p < 0.001).[80] This study also suggested that DECT has similar diagnostic performance as MRI. A
study by Werner et al showed a strong correlation between virtual iron content on
DECT and serum ferritin in patients of transfusional hemosiderosis (r = 0.62, p < 0.001).[81] In a retrospective study by Ma et al, the authors showed a strong correlation between
liver DECT-∆HU and R2* values in MRI (r = 0.83, p < 0.001).[52]
Fig. 7 Assessment of liver iron content on dual-source dual-energy computed tomography in
a 19-year-old male patient with β thalassemia. (A, B) Axial high energy (A, 140 kVp) and low energy (B, 80 kVp) non-contrast images show
a mean difference in attenuation of liver between the two energies (ΔHU) of 14.1 HU,
suggesting iron deposition. (C) Axial multi-echo gradient MR images from a 3T machine show a gradual decrease in
liver signal with an increase in TE. The T2* value was 10.3 milliseconds and R2* was
163/s suggestive of moderate iron deposition.
Table 3
Brief details and summaries of results of studies evaluating dual-energy computed
tomography in hepatic iron deposition
|
Study (year)
|
Sample size
Cases/controls
|
Modality of comparison
|
DECT Scanner type
|
DECT energies
|
Key results
|
|
Joe et al.
(2011)
|
87/0
|
DECT-∆HU and MRI (T1 IP and OP, and HASTE) with histology
|
SOMOTOM D,
Siemens
|
80, 140 kV
|
• Significant correlation of DECT-∆HU with degree of liver iron accumulation (r = 0.43,
p < 0.01)
• AUROC for clinically significant iron accumulation with DECT and MRI were 0.88 and
0.89, respectively.
|
|
Luo et al.
(2015)
|
56/0
|
DECT- material decomposition (VIC) with R2 and R2* technique
|
SOMATOM DF, Siemens
|
80, 140 kV
|
• Significant correlation between iron content on DECT and R2* & R2 MRI (r = 0.89 and 0.87, respectively; p < 0.001)
• At liver iron content > 3.2 mg/g of dry tissue, DECT and MRI R2* similar (AUROC,
0.96 vs. 0.99, respectively; p = 0.30)
|
|
Werner et al.
(2019)
|
99/48
|
DECT - VIC with serum ferritin and estimated transfused iron
|
SOMATOM DF, Siemens
|
100, 140 kV
|
• Good correlation between DECT iron content and serum ferritin (r = 0.62, p < 0.001)
|
|
Ma et al.
(2020)
|
31/0
|
DECT-∆HU and MRI (R2* technique)
|
IQon, Philips
|
Tube voltage 120 kVp
with dual-layer spectral detector
|
• Good correlation of DECT-∆HU with MRI liver R2* (r = - 0.83, p < 0.001)
|
Abbreviations: ∆HU, difference in attenuation between low- and high-energy images;
AUROC, area under receiver operating characteristic; DECT, dual-energy computed tomography;
HASTE, half Fourier single-shot turbo spin echo; IP, in-phase; OP, opposed phase;
r = correlation coefficient; VIC, virtual iron content.
Unlike SECT, which underestimates fat in the presence of iron and vice versa, several
phantom and animal studies have shown that DECT can accurately quantify iron, even
in the presence of fat[52]
[82]
[83] and vice versa.[52]
[83]
[84] In contrast, several other studies have suggested that iron and fat affect the measurement
of each other on DECT.[85]
[86]
[87] A recent study also suggested that the phase of contrast-enhanced scan (arterial,
portal venous, and delayed) significantly affects the hepatic iron measurements on
DECT.[88] These confounding factors need to be studied further, and iron-specific algorithms,
which are immune to these confounding effects, have to be developed before DECT can
be recommended for routine use in this setting.
A summary of suggestions for practical applications of DECT in the quantification
of hepatic fat, fibrosis, and iron content is given in [Table 4].
Table 4
Summary of suggestions for practical application of DECT in diffuse liver diseases
|
Dual-energy computed tomography: practical considerations
|
|
• Most of the current published literature on the role of DECT in diffuse liver diseases
has used a single-source DECT scanner with tube energies 80 kV and 140 kV. There are
not enough data comparing single-source DECT and dual-source DECT or different tube
energies to prefer one over the other.
• When the iodinated contrast agent is used, the contrast volume, injection rate,
and CT acquisition timings for DECT are the same as for conventional SECT. For liver
fibrosis quantification, only a delayed phase CT (variably at 3 to 5 minutes) may
be acquired in the dual-energy mode to avoid unnecessary data overload.
• For measuring any material density on DECT, we recommend drawing at least three
circular regions of interest (ROIs) with a diameter measuring 2 cm in each liver lobe,
avoiding major vessels or focal lesions, to account for the heterogeneous distribution
and for reproducible results.
|
|
Fat quantification
|
Iron quantification
|
Fibrosis quantification
|
|
• Non-contrast scan preferred although fat quantification can also be done on contrast-enhanced
DECT.
• ΔHU (difference in the attenuation between low and high energy images) or fat-specific
material decomposition algorithms are useful.
|
• Non-contrast scan is preferred although iron quantification can also be done on
contrast-enhanced DECT.
• ΔHU (difference in the attenuation between low and high energy images) or iron-specific
material decomposition algorithms are useful.
|
• Contrast-enhanced delayed phase scan is preferred.
• Although data are lacking on the optimum time of delay for fibrosis quantification,
in our experience 5-minute delayed scan provides accurate results.
• Normalized iodine concentration (ratio of iodine concentration of liver to iodine
concentration of aorta) provides better results.
|
|
|
|
|
Limitations
There are a few limitations to the routine use of DECT in diffuse liver diseases.
The major limitation is the high cost of the scanners, additional software required
for post-processing, and the need for larger data storage capacity. However, with
the increasing number of installations, demands, and awareness of its considerable
clinical utilities, the cost is expected to gradually reduce. Another limitation is
the difference in the technique of DECT between different vendors. There are not enough
studies comparing intervendor and interscanner variability with respect to the application
of DECT in diffuse liver diseases. This may limit the uniformity of the study results.
When DECT was first introduced, radiation dose was the main concern. However, later
multiple studies have shown that DECT does not increase radiation dose compared with
conventional CT and it is well within the ACR standard reference upper limit of 25
mGy for abdominal CT.[89]
[90]
[91]
[92]
[93] The presence of multiple material elements in the same area could affect the measurement
of fat, iodine, or iron. It is still not clear whether the presence of fat and iron
in the same voxel results in reduced performance of DECT. Active inflammation in the
parenchyma, which is a key component of NASH and other chronic liver diseases,[94]
[95]
[96]
[97]
[98]
[99]
[100]
[101]
[102] is difficult to evaluate with an imaging modality such as DECT. Hence, a complete
assessment will have to involve imaging in conjunction with other tests for inflammation.
Conclusion
Dual-energy CT is still in its infancy and most of its applications are still in the
research phase and not yet in wide clinical use. With exponential technological advances
in scanner hardware, material decomposition algorithms, and image reconstruction software,
the role of DECT will continue to grow in body imaging, and diffuse liver diseases
is not an exception. However, MRI and its applications will remain the imaging gold
standard for quantifying liver fat, iron, and fibrosis. The results from a majority
of preliminary studies on the role of DECT in diffuse liver diseases are promising.
However, larger cohort studies are needed to further validate these results before
they come into routine clinical practice. Optimum application of DECT in routine abdominal
CT protocols would not only give additional information for better patient management
but could also potentially prevent unnecessary biopsies.
