European Journal of Radiology 101 (2018) 103–110

Contents lists available at ScienceDirect

European Journal of Radiology

journal homepage: www.elsevier.com/locate/ejrad

Attenuation values of renal parenchyma in virtual noncontrast images T

acquired from multiphase renal dual-energy CT: Comparison with standard
noncontrast CT
⁎
Yuan-Mao Lina,b, Yi-You Chioua,b, Mei-Han Wua,b, Shan-Su Huanga, Shu-Huei Shena,b,
a
Department of Radiology, Taipei Veterans General Hospital, No. 201, Section 2, Shih-Pai Road, Taipei, 11217, Taiwan
b
School of Medicine, National Yang-Ming University, No. 155, Section 2, Linong Street, Taipei, 11217, Taiwan

A R T I C L E I N F O A B S T R A C T

Keywords: Objectives: To compare the renal parenchyma attenuation of virtual noncontrast (VNC) images derived from
Dual-energy computed tomography multiphase renal dual-energy computed tomography (DECT) with standard noncontrast (SNC) images, and to
Virtual noncontrast determine the optimum phase for VNC images.
Standard noncontrast Materials and methods: Twenty-nine men and 16 women (mean age, 61 ± 13 years; range, 37–89 years) un-
Multiphase renal CT
derwent dynamic renal DECT (100/Sn140 kVp) were included in this institutional review board–approved
Attenuation value
retrospective study. There were four phases of the scan, which included noncontrast, corticomedullary (CMP),
Radiation dose
nephrographic (NP), and excretory phases (EP). The VNC images was generated from CMP, NP and EP. CT
numbers of SNC images and VNC images of each phases were measured in the renal cortex and medulla. Mean
standard deviation of subcutaneous fat was measured as image noise on SNC and VNC images. Radiation dose
was recorded and potential radiation dose reduction was estimated. Results were tested for statistical sig-
nificance using the unpaired t-test and agreement using Bland–Altman plot analysis.
Results: The difference in mean attenuation between SNC and each phase of VNC images were ≤4 HU. The mean
attenuation of renal cortex and medulla was 33.2 ± 4.4 HU, and 34.2 ± 4.8 HU in SNC, 33.6 ± 7.6 HU and
31.1 ± 8.3 HU in VNC of CMP, 34.8 ± 8.6 HU and 35.6 ± 8.5 HU in VNC of NP, 31.5 ± 7.6 HU and
32.4 ± 7.5 HU in VNC of EP. In VNC of CMP, the attenuation of the cortex was higher than the medulla
(p < 0.05), and the attenuation of medulla was significant lower than that of SNC (p < 0.01). In VNC of NP,
the attenuation of renal cortex was higher than SNC (p < 0.05). In VNC of EP, the attenuation of cortex and
medulla were lower than SNC (p < 0.05), and inadequate iodine subtraction in collecting system was noted.
Image noise was significantly greater in SNC (p < 0.001). Mean radiation dose reduction achievable by re-
moving the SNC was 12.3% ± 0.9%.
Conclusions: VNC images from multiphase renal DECT were similar to SNC images. Using the nephrographic
phase can gives more comparable VNC images to SNC images in renal parenchyma than other phases.

1. Introduction multiphase CT [3–5]. This may particularly be of use in multiphase

Noncontrast imaging remains in widespread use and offers ad-
vantages for lesion detection in abdominal CT scans. Comparing the
attenuation change before and after contrast administration may reveal
enhancement, and the attenuation value in noncontrast image is im-
portant to determine the component of fat, calcification, stones, and
blood clots [1,2]. Replacing standard noncontrast (SNC) images with
virtual noncontrast (VNC) images derived from contrast-enhanced
series could help achieve a reduction in radiation exposure in
renal CT, which is widely used for detection and staging of renal tu-
mours, but at the expense of an increased radiation dose [6–9]. The
noncontrast scan in this examination is necessary for evaluating the fat
in angiomyolipomas, renal calculi, nephrocalcinosis, hematoma, and
urinoma as well as the enhancement of masses and complicated cysts.
VNC images must satisfy several criteria, including satisfactory image
quality, low noise levels, adequate iodine subtraction, minimal sig-
nificant artefacts, and approximation of attenuation measured on VNC
images to the equivalent measurements on SNC images.

Abbreviations: DECT, dual-energy computed tomography; SNC, standard noncontrast; VNC, virtual noncontrast; CMP, corticomedullary phase; NP, nephrographic phase; EP, excretory
phase
⁎
Corresponding author at: Department of Radiology, Taipei Veterans General Hospital, No. 201, Section 2, Shih-Pai Road, Taipei, 11217, Taiwan.
E-mail address: shshen@vghtpe.gov.tw (S.-H. Shen).

https://doi.org/10.1016/j.ejrad.2018.02.001
Received 12 December 2017; Received in revised form 31 January 2018; Accepted 3 February 2018
0720-048X/ © 2018 Elsevier B.V. All rights reserved.
Y.-M. Lin et al. European Journal of Radiology 101 (2018) 103–110

Fig. 1. Left kidney of SNC image (A), images of corticomedullary (B–D), nephrographic (E–G), and delayed phases (H–J) in the same patient. The figures from left to right were contrast-
enhanced images, iodine overlay images, and VNC images. White circles demonstrated the placement of ROIs in renal cortex and medulla at level of renal hila in similar location.
Inhomogeneous attenuation of renal parenchyma in VNC images could be noted, which was more obvious in corticomedullary phase (D). And the inadequate iodine subtraction (arrow)
in urinary collecting system was demonstrated in VNC images of excretory phase (J).

Previous studies have shown that dual-energy computed tomo-
graphy (DECT) can provide high-quality VNC images and a reasonable
approximation to SNC images in multiphase contrast-enhanced imaging
[10,11]. Although some studies have attempted to identify the optimal
phase for generating VNC images in multiphase DECT of the liver
[11–14], adrenal gland [15], and kidney [3,4], it is still a controversy
with different vendors, applications, and protocol. The aim of our study
is to determine the optimum phase to compute VNC images from a
multiphase CT scan of the kidneys.

104
Y.-M. Lin et al.
2. Materials and methods
2.1. Patients
In this institutional review board-approved study, informed consent
was waived. Between October 2016 and February 2017, 45 consecutive
adult patients undergoing multiphase renal CT using DECT for renal
tumour survey or regular follow-up were identified for retrospective
analysis.
2.2. Technique
The patients were positioned supine on the table. The dual-source
DECT system (Definition Flash, Siemens Medical Solutions, Forchheim,
Germany) was used for scanning. First, a non-contrast scan of the
kidney was acquired using single energy mode, with a detector con-
figuration of 64 × 0.6 mm. For the non-contrast scans, both automated
tube current modulation and attenuation-based automated tube po-
tential selection (CARE Dose 4D and CARE kV; Siemens Healthcare)
were used, with a quality reference of 210 mAs at 120 kV. The CARE kV
software selected 100 kVp tube voltage for all patients, with an adjusted
quality reference of 352 mAs. This indicates that there was no large
patient who required a higher voltage in our patient group.
Post-contrast scans were obtained using the dual-energy mode (tube
1 = Sn140 kVp; tube 2 = 100 kVp). Automatic current modulation for
exposure control (Care Dose 4D; Siemens Healthcare, Forchheim,
Germany) with a quality reference amperage of 178 mAs for tube 1 and
230 mAs for tube 2 was used. Subsequently, a power injector (Stellant
CT Injection System, Medrad) was used for the intravenous injection of
non-ionic iodinated contrast material at a flow rate of 2–3 mL/second,
and was followed by 30-mL saline chaser. The contrast material dose
was adapted to patient weight (80 mL for patients weighting less than
50 kg, 100 mL for patients weighting between 50 and 80 kg, and 120 mL
for patients weighting over 80 kg). The postcontrast scans were in-
itiated by an automatic bolus-tracking program. Corticomedullary
phase (CMP) scans of the kidney were acquired at 15 s after the aortic
peak (100 HU); nephrographic phase (NP) scans from the dome of the
liver to the pubic symphysis were acquired at 90 s following the in-
itiation of injection; and excretory phase (EP) scans from the dome of
the liver to the pubic symphysis were acquired at 300 s. All the post-
contrast scanning were using collimation and pitch of 14 × 1.2 mm,
and 0.6, respectively.
In all examinations, iohexol (Omnipaque 350, GE Healthcare, USA),
iopamidol (Iopamiron 370, Bracco S.P.A., Italy), iobitridol (Xenetix
350, Guerbet, France), or iopromide (Ultravist 370, Bayer Schering
Pharma, Germany) was randomly delivered via an 18- or 20-gauge
catheter inserted into the antecubital vein by using the power injector.
These four non-ionic contrast mediums are routinely used in our in-
stitute, and are randomly assigned to patients.
2.3. Image reconstruction and VNC postprocessing
The noncontrast raw data were reconstructed using a thickness of
1 mm and an interval of 1 mm with kernel of I30f. The 100 kV images,
Sn140 kV images and mixed images with ratio of 0.5 were generated
from dual energy raw data, using a thickness of 1 mm and an interval of
1 mm with kernel of Q30f. These images were imported into a dedi-
cated dual-energy postprocessing software package (syngo.via, version
VB10B, Siemens Healthcare). The VNC images were generated from
each of the contrast-enhanced sequences using Liver VNC with the
manufacturer’s default setting, which is a standard three-material de-
composition algorithm for producing VNC images. The SNC and VNC
images were then relayed to a picture archiving and communication
system (PACS) for review.
105
European Journal of Radiology 101 (2018) 103–110
2.4. Quantitative analysis
Attenuation measurements were obtained by placing a circular re-
gion of interest (ROI) on axial SNC and VNC images of the kidney. The
renal cortex and medulla were analysed separately, because the en-
hancement pattern was different in each phase. Two measurements at
anterior and posterior kidney were obtained of the ipsi- and con-
tralateral renal cortex and renal medulla at the level of renal hila in
axial view. The ROIs 0.2–0.5 cm2 in size for all measurements were
maintained in similar locations on the VNC and SNC images (Fig. 1).
Measurement were performed in consensus among two abdominal
radiologists (Y.-M.L., S.-H.S.) with four years and fifteen years of ex-
perience in abdominal CT, respectively, to assure the appropriate pla-
cement of ROIs. A subset of measurements with 20 cases from the same
patient group was performed by another radiologist (M.-H.W.), with 20
years of experience in abdominal CT, to estimate the reproducibility.
The mean attenuation values and standard deviations were recorded.
The standard deviations of the anterior subcutaneous fat were also
measured to determine attenuation stability and image noise on the
basis of a standardised ROI 1 cm2 in size.
2.5. Dose evaluation
The scanner generated volume computed tomography dose index
(CTDIvol, mGy) and dose–length product (DLP, mGy·cm) were recorded
from the patient protocol in each phase. The effective radiation dose in
millisieverts was estimated for each phase by using a method proposed
by the European Working Group for Guidelines on Quality Criteria in
CT and was derived by using the DLP and abdominal conversion coef-
ficient k (kabd = 0.015 mSv/[mGy cm]) [16]. We compared the effec-
tive dose of the conventional multiphase protocol with that of the no-
SNC image protocol and calculated the percentage dose reduction va-
lues.
2.6. Statistical analysis
CT numbers measured on SNC and VNC images were tested for
normal distribution by using the Kolmogorov–Smirnov test. All the at-
tenuation values and the noise in the subcutaneous fat were tested using
the unpaired t-test. The attenuation values of VNC images from each
phase were compared with those of SNC images, and the attenuation of
the renal cortex was compared with that of the renal medulla in the
same phase. Agreement and the variation distribution between SNC and
VNC images were objectively assessed using Bland–Altman plot ana-
lysis, in which the difference between SNC and VNC images (y-axis) was
plotted against the average of SNC and VNC images (x-axis) with limits
of agreement defined as mean bias ± 1.96 standard deviations. Also,
the correlation statistics was based on Pearson’s correlation. Statistical
analysis was performed using SPSS (version 15.0; SPSS, Chicago, IL,
USA); p < 0.05 was considered significant.
3. Results
Among the 45 patients analysed (29 men, 16 women; mean age
61 ± 13 years; range 37–89 years), 160 renal cortex and medulla ROIs
were derived from each SNC and VNC image acquired in the CMP, NP,
and EP (overall, 1280 measurements; 10 patients received nephrectomy
at either side of kidney). Among these patients, there were 5 patients
with renal angiomyolipomas, 2 with complicated cysts, 9 with clear cell
type renal cell carcinomas, 1 with chromophobe renal cell carcinomas,
and 1 with urothelial cell carcinomas. The rest had normal finding in
renal parenchyma.
The average attenuation values of SNC and VNC images in different
phases were demonstrated in Table 1 and Fig. 2, and the measurements
between reader 1 and 2 gave the similar results. Statistical comparison
between SNC and VNC images was demonstrated in Tables 1 and 2, and
Y.-M. Lin et al. European Journal of Radiology 101 (2018) 103–110

Table 1
Comparison of standard noncontrast (SNC) images, virtual noncontrast (VNC) images of corticomedullary, nephrographic, and excretory phases using unpaired t-test.

Renal Cortex Renal Medulla p-value difference of cortex and

medulla
Average Hounsfield p-value difference to Average Hounsfield p-value difference to
units SNC units SNC

SNC Reader 1 33.2 ± 4.4 34.2 ± 4.8 0.131

Reader 2 33.2 ± 4.8 34.4 ± 5.2 0.23
**
Corticomedullary VNC Reader 1 33.6 ± 7.6 0.638 31.1 ± 8.3 < 0.001 0.003**
Reader 2 33.9 ± 7.5 0.657 31.3 ± 8.0 0.001** 0.029*
Nephrographic VNC Reader 1 34.8 ± 8.6 0.028* 35.6 ± 8.5 0.158 0.568
Reader 2 35.6 ± 7.5 0.015* 36.1 ± 8.8 0.184 0.645
Excretory VNC Reader 1 31.5 ± 7.6 0.025* 32.4 ± 7.5 0.009** 0.411
Reader 2 30.5 ± 7.5 0.012* 31.7 ± 7.6 0.006* 0.396

Reader 1: measurements of 45 patients, Reader 2: measurement of 20 patients from the same patient group.
SNC, Standard noncontrast images.
VNC, Virtual noncontrast images.
* =p < 0.05.
** =p < 0.01.

showed that the mean differences did not exceed approximately 4 HU;
the largest difference appeared in the medulla in the CMP, with the
VNC attenuation being lower than the SNC attenuation by 3.1 HU. The
attenuation values of VNC images acquired in the NP were slightly
higher than those of SNC images—on average 1.6 HU higher in the
cortex and 1.4 HU higher in the medulla. By contrast, the attenuation
values of VNC images acquired in the EP were slightly lower than those
of SNC images—on average 1.7 HU lower in the cortex and 1.8 HU
lower in the medulla. Generally, a larger standard deviation of VNC
attenuation values was noted, reflecting the wider variability of at-
tenuation values than that of SNC images (Fig. 2).
For the cortex, the attenuation values of VNC images showed no
significant difference from SNC images in the CMP, while they were
different in the other two phases. For the medulla, the attenuation
values of VNC images in the CMP and EP were significantly lower than
those of SNC images. On comparing the attenuation values of the cortex
and medulla in the same phase, no significant difference was found
between SNC and VNC images in both the NP and EP. However, the
attenuation values of the medulla were significantly lower than those of
the cortex in the CMP (Fig. 1). In comparisons of the three phases, NP
showed the least difference between VNC and SNC images as well as
between the cortex and medulla. In VNC images acquired in the EP,
contrast medium retained in the collecting system was noted (Fig. 1).
Bland–Altman plot analyses were performed on the attenuation
measurements to compare SNC and VNC images (Fig. 3 and Table 2).
Although most points lay within the two error bars (mean bias of 95%
confidence interval), the distance between the upper and lower error
bars was approximately 50 HU. The distribution of the points exhibited
a trend where the difference HU (y-axis) decreased as the mean HU (x-
axis) increased. There was a moderate significant correlation for dif-
ference HU and mean HU, with a Pearson’s correlation coefficient from
−0.412 to −0.583 (Table 2).
The objective analysis of noise was based on the standard deviation
of the attenuation values in each ROI drawn on subcutaneous fat. The
average standard deviation within these ROIs as the measurement of
image noise was 14.5 ± 1.4 for SNC images, 12.3 ± 1.8 for VNC
images acquired in the CMP, 12.7 ± 2.0 for VNC images acquired in
the NP, and 10.8 ± 1.7 for VNC images acquired in the EP. The noise
of SNC images was significantly higher than that of VNC images
(p < 0.001).
The mean CTDlvol, DLP and effective dose for noncontrast, CMP, NP
and EP were demonstrated in Table 3. The mean total effective dose of
standard multiphase renal DECT was 17.2 ± 3.0 (range, 10.7–30.7)
mSv. The effective doses for the CMP, NP, and EP were higher than
those for SNC images. If SNC images were omitted, the average theo-
retical dose reduction calculated as an effective dose value was
12.3% ± 0.9% in this study.

Fig. 2. Box and whisker’s plot comparing the renal cortex and medulla Hounsfield unit (HU) values for SNC and VNC images. Note the greater attenuation variability in VNC images.

106
Y.-M. Lin et al. European Journal of Radiology 101 (2018) 103–110

Fig. 3. Bland–Altman graphs show magnitude of attenuation difference between VNC of corticomedullary, nephrographic, and excretory phases and SNC images on renal cortex and
medulla. The mean attenuation differences were less than 4 HU, and the majority of points lie within the two error bars ( ± 1.96 standard deviations), which the distance was around 50
HU.

4. Discussion
VNC images from DECT can reduce radiation exposure in compar-
ison with SNC images in patients requiring both noncontrast and con-
trast-enhanced imaging, especially in multiphase CT. VNC images that
can serve as surrogates for SNC images can be produced by subtracting
the iodine from contrast-enhanced images. Previous studies have shown
that in isolated organ systems, DECT allowed for successful subtraction
of iodine-containing contrast material from single-phase contrast-en-
hanced images [10,17,18]. However, the reliability of VNC imaging

107
Y.-M. Lin et al.
Table 2
Mean attenuation difference between virtual noncontrast images and standard noncontrast images using Bland–Altman analysis and Pearson’s correlation between mean attenuation and
mean attenuation difference to SNC.
Renal Cortex
Mean Attenuation Difference to Correlation of mean attenuation
SNC (Mean Bias with 95% CI) and mean difference to SNC
Corticomedullary VNC −0.4 (22.7 to −21.7) r = −0.495, p < 0.001
Nephrographic VNC −1.6 (22.2 to −30.5) r = −0.583, p < 0.001
Excretory VNC 1.7 (25.1 to −22.8) r = −0.491, p < 0.001
SNC, Standard noncontrast images VNC, Virtual noncontrast images CI, Confidence interval.
remains uncertain in multiorgan and multiphase applications. In mul-
tiphase studies using second-generation dual-source DECT, the at-
tenuation values of VNC images of the liver, spleen, kidney, and aorta
have been reported to be overestimated in the arterial and portal phases
[12–14]. In a multiphase renal imaging study, VNC images derived
from the NP and EP exhibited overestimated attenuation values in the
liver and spleen, but underestimated values in the renal parenchyma
[3]. In a study using third-generation dual-source DECT, the attenua-
tion values of VNC images of liver were similar to those of SNC images
in the arterial and portal phases, but were underestimated in the
kidney, pancreas, adrenal gland, and aorta [11].
In this study, we used second-generation dual-source DECT, focused
on multiphase renal imaging, and evaluated the estimated attenuation
of VNC images in the cortex and medulla separately. We used small size
of ROI (0.2–0.5 cm2) to measure cortex and medulla. Although at-
tenuation measurements are known to change with different size of ROI
[19,20], the constant size and location of ROIs were selected in our
study to lessen the impact.
The principle of creating VNC images is based on three-material
deposition algorithm, assuming that each voxel was composed of fat,
soft tissue and iodine. Our results showed that the mean attenuation
difference between VNC and SNC images was less than 4 HU in all
phases for both the cortex and medulla; however, the mean bias was
greater than 20 HU in Bland–Altman spot (Table 2, Fig. 3). This indicate
that the attenuation values difference between SNC and VNC were
greater than 20 HU in some measurements, which may challenge the
routine substitution of VNC images in clinical applications that rely on
attenuation value measurements. This finding is comparable with
Miller et al. [21], who showed the substantial variations in inter-
individual VNC attenuation in different organs. They proposed a pixel-
by-pixel organ based algorithm to correct the variations caused by
different organ perfusion. Moreover, in Bland–Altman analysis, a
moderate negative linear relationship was found between mean at-
tenuation and mean attenuation difference of SNC and VNC (Table 2).
This indicates the overestimated attenuation of VNC in higher esti-
mated attenuation while underestimated attenuation of VNC in lower
estimated attenuation. This finding is also consistent with previous
research showing the difference between SNC and VNC attenuation
appeared linearly dependent on SNC attenuation [21]. These sub-
stantial interindividual variations in virtual unenhanced attenuation
reflect the fact that the algorithm for iodine subtraction is not perfect,
because the kidney is not only composed of fat and soft tissue, but also
blood, and urine. Therefore, other algorithms based on different three
materials may more accurately decompose DECT data and successfully
Table 3
CTDlvol, DLP and effective dose of noncontrast, corticomedullary, nephrographic, and excretory phases. Data are means ± standard deviations (range).
CTDlvol (mGy)
Noncontrast phase 7.2 ± 1.6 (4.2–11.7)
Corticomedullary phase 8.9 ± 1.9 (5.4–14.2)
Nephrographic phase 9.5 ± 1.8 (6.7–16.3)
Excretory phase 9.9 ± 1.9 (6.9–17.3)
108
European Journal of Radiology 101 (2018) 103–110
Renal Medulla
Mean Attenuation Difference to Correlation of mean attenuation and
SNC (Mean Bias with 95% CI) mean difference to SNC
3.1 (21.2 to −30) r = − 0.503, p < 0.001
-1.4 (19.8 to −27.1) r = − 0.514, p < 0.001
1.8 (26.6 to −22.8) r = − 0.412, p < 0.001
subtract iodine in kidney.
Second, we found that the mean attenuation of the renal cortex was
higher than that of the medulla in VNC images acquired in the CMP.
This discrepancy between the cortex and medulla has been noted in
previous studies with VNC images acquired in the arterial and portal
phases. De Cecco et al. [11,13] reported that attenuation values of the
renal cortex in VNC images acquired in the arterial phase were higher
than those from the portal venous phase, whereas the medulla showed
higher attenuation in the portal venous phase than in the arterial phase.
The factors that have been postulated to affect iodine subtraction in-
clude large patient size, areas of beam-hardening artefacts, very high
iodine concentrations, scatter and partial volume effect [10,21–23]. In
the CMP, the renal cortex shows intense enhancement due to pre-
ferential arterial flow in the cortex and glomerular filtration of the
contrast material, whereas the medulla remains relatively less enhanced
[24,25]. This discrepant attenuation may be related to the higher iodine
concentration in the cortex, creating a beam-hardening effect and
thereby producing the lower VNC attenuation in the medulla. This
phenomenon raises a concern about the reliability of attenuation
measurement on VNC images of lesions with high contrast enhance-
ment differences, and further investigation is necessary.
Third, we found that the noise of SNC images was significantly
higher than the noise of VNC images. Similar observations have been
reported previously [10–14,21,22]. The lower noise of VNC images is
likely due to smoothing induced by the postprocessing algorithm
[12,13]. Moreover, the higher tube voltage was used in postcontrast
scanning, which gives the lower noise comparing to noncontrast scan-
ning. Although the image noise was lower in VNC images, the in-
adequate iodine subtraction, mildly blurred organ margins, possibly
attributable to imperfect image fusion between the two datasets due to
motion artefacts or to the denoising algorithm applied in postproces-
sing, were also found in our study.
Although it has been reported that the contrast material phase does
not have an effect on VNC attenuation values [3,11,12,14], there is no
agreement on which phase is optimal for generating VNC images in
multiphase CT scan. In liver images, different authors have re-
commended the arterial phase [11,14] or the portovenous phase
[12,13] for generating VNC images. In the present study, we found a
trend of overestimated attenuation values of VNC images in the NP and
underestimated values in the EP. Additionally, for the renal medulla,
CMP and EP VNC attenuation values were both significantly lower than
SNC attenuation values. In the EP, inadequate iodine subtraction was
noted in the urinary collecting system (Fig. 1). Our result confirmed
that the scanning phase, the enhancement pattern of interest organ, and
DLP (mGy cm) Effective dose (mSv)
141.6 ± 35.6 (82.1–246.7) 2.1 ± 0.5 (1.2–3.7)
177.6 ± 45.5 (106.8–315) 2.6 ± 0.7 (1.6–4.7)
426.1 ± 101.6 (151.3–753.3) 6.4 ± 1.5 (2.3–11.3)
402.0 ± 87.1 (259–729.4) 6.0 ± 1.3 (3.9–11.)
Y.-M. Lin et al.
the degree of enhancement had an impact on VNC images because of
the imperfect iodine subtraction algorithm. These findings are con-
sistent with others using VNC to detect urinary stones, which showed
that oversubtraction of the stone and undersubtraction of iodine in
urinary tract affect the sensitivity of stone identification in different
phase of CT urography [3,26,27]. Overall, for multiphase renal DECT,
the attenuation of VNC in NP giving more accurate attenuation values
of cortex and medulla than other phases. Regarding the accuracy of
attenuation values, the NP is the optimal phase for generating VNC
images for kidney in multiphase renal DECT.
In regard to radiation dose, it is concerned whether using dual-
source CT investigation increase radiation dose comparing with single-
source single-energy CT. However, recent literature revealed no in-
crease in radiation dose if low tube currents are used [28–30]. Previous
studies with triple-phase examinations of the abdomen have reported
mean dose savings of 24.8%–35.1% [3,11–13]. In our study, utilising
VNC images instead of SNC images led to a mean dose reduction of
approximately 12%. This may be due to several factors. First, we used
quadruple-phase examination instead of triple-phase examination.
Second, automated tube-current modulation and attenuation-based
automated tube potential selection were used for SNC image acquisition
in our study, which allowed reduction of radiation exposure while
maintaining high image quality in contrast-enhanced CT [31]. We re-
commend that, under the ALARA (as low as reasonably achievable)
principle, dose minimisation must be attempted in any manner, espe-
cially for high-radiation-dose examinations such as quadruple-phase
study.
Our study had several limitations. First, our results can be applied
only to VNC images generated by dual-source CT technology. Other
DECT approaches, such as rapid kilovolt switching and energy-sensitive
sandwich detector technologies, may show different results in at-
tenuation values and image quality. Second, the influences of the
software settings and the voltage of the modality on attenuation were
not investigated—the default setting was used in our study; this could
have significantly affected VNC image performance. Third, the place-
ment of the ROI on thinned renal cortices was a challenge, especially in
patients who had chronic kidney disease or polycystic kidney disease.
Fourth, although Miller et al. [21] reported that habitus influences the
calculation of VNC images and Graser et al. [32] recommended not
performing DECT in patients whose body mass index is > 30 because of
increased image noise, we did not exclude large patients because we
intend our proposed technique to be applicable in the routine clinical
setting, without the need for prior patient screening and exclusion on
the basis of habitus. Fifth, there were four different brand of contrast
medium used in this study with the iodine concentration of 350 mg/mL
and 370 mg/mL. And the volume of contrast medium was not adjusted
to body mass index of patients. These may affect the iodine con-
centration of renal parenchyma; however, the iodine subtraction algo-
rithm should remove the iodine in regard of variant concentration.
Finally, only attenuation-based analysis was performed; the detection
and differentiation of the kidney lesions were not evaluated due to the
limited number of cases.
5. Conclusion
In conclusion, the VNC images were affected by the scanning phase,
the enhancement pattern of interest organ, and the degree of en-
hancement because of the imperfect iodine subtraction algorithm.
Regarding the image quality and accuracy of attenuation values, we
recommend the nephrographic phase for creating VNC images in mul-
tiphase renal imaging which are comparable to SNC images.
Conflict of interest
This research did not receive any specific grant from funding
agencies in the public, commercial, or not-for-profit sectors.
109
European Journal of Radiology 101 (2018) 103–110
Acknowledgement
This research did not receive any specific grant from funding
agencies in the public, commercial, or not-for-profit sectors.
References
[1] J. Kelly, V. Raptopoulos, A. Davidoff, R. Waite, P. Norton, The value of non-con-
trast-enhanced CT in blunt abdominal trauma, AJR Am. J. Roentgenol. 152 (1)
(1989) 41–48.
[2] R.C. Smith, A.T. Rosenfield, K.A. Choe, K.R. Essenmacher, M. Verga,
M.G. Glickman, R.C. Lange, Acute flank pain: comparison of non-contrast-enhanced
CT and intravenous urography, Radiology 194 (3) (1995) 789–794.
[3] V.A. Sahni, A.B. Shinagare, S.G. Silverman, Virtual unenhanced CT images acquired
from dual-energy CT urography: accuracy of attenuation values and variation with
contrast material phase, Clin. Radiol. 68 (3) (2013) 264–271.
[4] K.D. Song, C.K. Kim, B.K. Park, B. Kim, Utility of iodine overlay technique and
virtual unenhanced images for the characterization of renal masses by dual-energy
CT, AJR Am. J. Roentgenol. 197 (6) (2011) W1076–W1082.
[5] Y.K. Kim, B.K. Park, C.K. Kim, S.Y. Park, Adenoma characterization: adrenal pro-
tocol with dual-energy CT, Radiology 267 (1) (2013) 155–163.
[6] D.H. Szolar, F. Kammerhuber, S. Altziebler, M. Tillich, E. Breinl, R. Fotter,
H.H. Schreyer, Multiphasic helical CT of the kidney: increased conspicuity for de-
tection and characterization of small (< 3-cm) renal masses, Radiology 202 (1)
(1997) 211–217.
[7] P. Dahlman, E. Semenas, E. Brekkan, A. Bergman, A. Magnusson, Detection and
characterisation of renal lesions by multiphasic helical CT, Acta Radiol. (Stockholm,
Sweden: 1987) 41 (4) (2000) 361–366.
[8] M.A. Bosniak, N.M. Rofsky, Problems in the detection and characterization of small
renal masses, Radiology 198 (3) (1996) 638–641.
[9] D.J. Brenner, E.J. Hall, Computed tomography–an increasing source of radiation
exposure, N. Engl. J. Med. 357 (22) (2007) 2277–2284.
[10] A. Graser, T.R. Johnson, E.M. Hecht, C.R. Becker, C. Leidecker, M. Staehler,
C.G. Stief, H. Hildebrandt, M.C. Godoy, M.E. Finn, F. Stepansky, M.F. Reiser,
M. Macari, Dual-energy CT in patients suspected of having renal masses: can virtual
nonenhanced images replace true nonenhanced images? Radiology 252 (2) (2009)
433–440.
[11] C.N. De Cecco, G. Muscogiuri, U.J. Schoepf, D. Caruso, J.L. Wichmann,
P.M. Cannao, C. Canstein, S.R. Fuller, L. Snider, A. Varga-Szemes, A.D. Hardie,
Virtual unenhanced imaging of the liver with third-generation dual-source dual-
energy CT and advanced modeled iterative reconstruction, Eur. J. Radiol. 85 (7)
(2016) 1257–1264.
[12] T. Barrett, D.J. Bowden, N. Shaida, E.M. Godfrey, A. Taylor, D.J. Lomas, A.S. Shaw,
Virtual unenhanced second generation dual-source CT of the liver: is it time to
discard the conventional unenhanced phase? Eur. J. Radiol. 81 (7) (2012)
1438–1445.
[13] C.N. De Cecco, A. Darnell, N. Macias, J.R. Ayuso, S. Rodriguez, J. Rimola, M. Pages,
A. Garcia-Criado, M. Rengo, A. Laghi, C. Ayuso, Virtual unenhanced images of the
abdomen with second-generation dual-source dual-energy computed tomography:
image quality and liver lesion detection, Investig. Radiol. 48 (1) (2013) 1–9.
[14] L.J. Zhang, J. Peng, S.Y. Wu, Z.J. Wang, X.S. Wu, C.S. Zhou, X.M. Ji, G.M. Lu, Liver
virtual non-enhanced CT with dual-source, dual-energy CT: a preliminary study,
Eur. Radiol. 20 (9) (2010) 2257–2264.
[15] D. Botsikas, F. Triponez, S. Boudabbous, C. Hansen, C.D. Becker, X. Montet,
Incidental adrenal lesions detected on enhanced abdominal dual-energy CT: can the
diagnostic workup be shortened by the implementation of virtual unenhanced
images? Eur. J. Radiol. 83 (10) (2014) 1746–1751.
[16] H. Menzel, H. Schibilla, D. Teunen, European guidelines on quality criteria for
computed tomography, Luxembourg, Luxemb.: Eur. Comm. (2000) 735–738.
[17] W.H. Sommer, A. Graser, C.R. Becker, D.A. Clevert, M.F. Reiser, K. Nikolaou,
T.R. Johnson, Image quality of virtual noncontrast images derived from dual-energy
CT angiography after endovascular aneurysm repair, J. Vasc. Interv. Radiol.: JVIR
21 (3) (2010) 315–321.
[18] M. Remy-Jardin, J.B. Faivre, F. Pontana, F. Molinari, N. Tacelli, J. Remy, Thoracic
applications of dual energy, Semin. Respir. Crit. Care Med. 35 (1) (2014) 64–73.
[19] M.A. Bosniak, A.J. Megibow, D.H. Hulnick, S. Horii, B.N. Raghavendra, CT diag-
nosis of renal angiomyolipoma: the importance of detecting small amounts of fat,
AJR Am. J. Rroentgenol. 151 (3) (1988) 497–501.
[20] M.S. Davenport, A.M. Neville, J.H. Ellis, R.H. Cohan, H.S. Chaudhry, R.A. Leder,
Diagnosis of renal angiomyolipoma with hounsfield unit thresholds: effect of size of
region of interest and nephrographic phase imaging, Radiology 260 (1) (2011)
158–165.
[21] C.M. Miller, R.T. Gupta, E.K. Paulson, A.M. Neville, M.R. Bashir, E.M. Merkle,
D.T. Boll, Effect of organ enhancement and habitus on estimation of unenhanced
attenuation at contrast-enhanced dual-energy MDCT: concepts for individualized
and organ-specific spectral iodine subtraction strategies, AJR Am. J. Roentgenol.
196 (5) (2011) W558–64.
[22] C.N. De Cecco, V. Buffa, S. Fedeli, M. Luzietti, A. Vallone, R. Ruopoli, V. Miele,
M. Rengo, P. Paolantonio, M. Maurizi Enrici, A. Laghi, V. David, Dual energy CT
(DECT) of the liver: conventional versus virtual unenhanced images, Eur. Radiol. 20
(12) (2010) 2870–2875.
[23] X. Liu, L. Yu, A.N. Primak, C.H. McCollough, Quantitative imaging of element
composition and mass fraction using dual-energy CT: three-material decomposition,
Med. Phys. 36 (5) (2009) 1602–1609.
Y.-M. Lin et al.
[24] S. Sheth, E.K. Fishman, Multi-detector row CT of the kidneys and urinary tract:
techniques and applications in the diagnosis of benign diseases, Radiographics 24
(2) (2004) e20 a review publication of the Radiological Society of North America,
Inc.
[25] B.I. Yuh, R.H. Cohan, Different phases of renal enhancement: role in detecting and
characterizing renal masses during helical CT, AJR Am. J. Roentgenol. 173 (3)
(1999) 747–755.
[26] N. Takahashi, T.J. Vrtiska, A. Kawashima, R.P. Hartman, A.N. Primak, J.G. Fletcher,
C.H. McCollough, Detectability of urinary stones on virtual nonenhanced images
generated at pyelographic-Phase dual-Energy CT, Radiology 256 (1) (2010)
184–190.
[27] H. Scheffel, P. Stolzmann, T. Frauenfelder, T. Schertler, L. Desbiolles, S. Leschka,
B. Marincek, H. Alkadhi, Dual-energy contrast-enhanced computed tomography for
the detection of urinary stone disease, Investig. Radiol. 42 (12) (2007) 823–829.
[28] C.A. Coursey, R.C. Nelson, D.T. Boll, E.K. Paulson, L.M. Ho, A.M. Neville, D. Marin,
R.T. Gupta, S.T. Schindera, Dual-energy multidetector CT: how does it work, what
110
European Journal of Radiology 101 (2018) 103–110
can it tell us, and when can we use it in abdominopelvic imaging? Radiographics 30
(4) (2010) 1037–1055 a review publication of the Radiological Society of North
America, Inc.
[29] T. Henzler, C. Fink, S.O. Schoenberg, U.J. Schoepf, Dual-energy CT: radiation dose
aspects, AJR Am. J. Roentgenol. 199 (5 Suppl) (2012) S16–25.
[30] T.R. Johnson, B. Krauss, M. Sedlmair, M. Grasruck, H. Bruder, D. Morhard, C. Fink,
S. Weckbach, M. Lenhard, B. Schmidt, T. Flohr, M.F. Reiser, C.R. Becker, Material
differentiation by dual energy CT: initial experience, Eur. Radiol. 17 (6) (2007)
1510–1517.
[31] K.H. Lee, J.M. Lee, S.K. Moon, J.H. Baek, J.H. Park, T.G. Flohr, K.W. Kim, S.J. Kim,
J.K. Han, B.I. Choi, Attenuation-based automatic tube voltage selection and tube
current modulation for dose reduction at contrast-enhanced liver CT, Radiology 265
(2) (2012) 437–447.
[32] A. Graser, T.R. Johnson, H. Chandarana, M. Macari, Dual energy CT: preliminary
observations and potential clinical applications in the abdomen, Eur. Radiol. 19 (1)
(2009) 13–23.