Genitourinar y Imaging • Original Research

Patel et al.
Single-Phase Contrast-Enhanced Dual-Energy CT for Small
Focal Renal Lesions

Genitourinary Imaging
Original Research

Characterization of Small (< 4 cm)

Focal Renal Lesions: Diagnostic
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved

Accuracy of Spectral Analysis Using

Single-Phase Contrast-Enhanced
Dual-Energy CT
Bhavik N. Patel1 OBJECTIVE. The purpose of this study is to determine whether single-phase contrast-en-
Alex Bibbey 1 hanced dual-energy quantitative spectral analysis improves the accuracy of diagnosis of small
Kingshuk R. Choudhury 2 (< 4.0 cm) renal lesions, compared with conventional single-energy attenuation measurements.
Richard A. Leder 1 MATERIALS AND METHODS. In this retrospective study, 136 consecutive patients
Rendon C. Nelson1 (95 men and 41 women; mean age, 54 years) with 144 renal lesions (111 benign and 33 ma-
lignant) underwent single-energy unenhanced and dual-energy contrast-enhanced CT of the
Daniele Marin1
abdomen. For each renal lesion, attenuation measurements were obtained, and an attenuation
Patel BN, Bibbey A, Choudhury KR, Leder RA, change of 15 HU or greater was considered evidence of enhancement. Dual-energy spectral
Nelson RC, Marin D attenuation curves were generated for each lesion. The slope of each curve was measured be-
tween 40 and 50 keV (λHU40–50), 40 and 70 keV (λHU40–70), and 40 and 140 keV (λHU40–140).
Mean lesion attenuation values and spectral attenuation curve parameters were compared be-
tween benign and malignant renal lesions by use of the two-sample t test. Diagnostic accuracy
was assessed and validated using cross-validation analysis.
RESULTS. With the use of cross-validated optimal thresholds at 100% sensitivity, speci-
ficity for differentiating between benign and malignant renal lesions improved significantly
when both λHU40–70 and λHU40–140 were used, compared with conventional enhancement
measurements (93% [103/111; 95% CI, 86–97%] vs 81% [90/111; 95% CI, 73–88%]) (p =
0.02). The sensitivity of λHU40–70 and λHU40–140 was also higher than that of conventional
enhancement measurements, although it was not statistically significant.
CONCLUSION. Single-phase contrast-enhanced dual-energy quantitative spectral
analysis significantly improves the specificity for characterization of small (< 4.0 cm) renal
lesions, compared with conventional single-energy attenuation measurements.

he widespread use of cross-sec- negligible number of renal lesions cannot be

Keywords: dual-energy CT, renal lesions, spectral analysis

DOI:10.2214/AJR.17.17824
Received December 15, 2016; accepted after revision
February 16, 2017.
R. C. Nelson is a consultant to GE Healthcare.
1
Department of Radiology, Duke University Medical
Center, Box 3808, Erwin Rd, Durham, NC 27710. Address
correspondence to B. N. Patel (bhavikp@stanford.edu).
2
Carl E. Ravin Advanced Imaging Laboratories,
Duke University Medical Center, Durham, NC.
AJR 2017; 209:815–825
0361–803X/17/2094–815
© American Roentgen Ray Society
T tional imaging has led to an in-
creasing prevalence of small
(<  4.0 cm) incidentally detected
renal lesions [1, 2]. Although most renal inci-
dentalomas are inconsequential (i.e., simple
renal cysts), not all incidental renal lesions
can be presumed to be benign. In fact, most
renal cell carcinomas are discovered inciden-
tally [3–5], thus posing a diagnostic chal-
lenge for the radiologist who is asked to ren-
der a definite diagnosis or an evidence-based
assessment of the probability of malignancy.
MDCT has been accepted as the modal-
ity of choice for the characterization of re-
nal lesions [6–8]. With the use of a dedicated
renal lesion protocol, which includes at least
unenhanced and nephrographic acquisitions
[6–8], most incidental renal lesions can be
diagnosed accurately, leading to appropriate
management. Nevertheless, a small but not
characterized with imaging alone, either be-
cause they are too small or because they are
detected incidentally on examinations that
are not optimized for evaluation of a renal
lesion [7]. For example, an incidental renal
lesion encountered on a routine single-phase
CT scan obtained for abdominal pain may
remain indeterminate on the basis of atten-
uation measurements. This common clinical
scenario may result in unnecessary follow-
up imaging, biopsy, or surgery [9].
Accumulating evidence suggests that dual-
energy CT is a promising imaging modality in
the evaluation of incidental renal lesions [10–
16]. In addition to providing material-specific
information (e.g., iodine quantification), dual-
energy CT datasets may also be used to syn-
thesize virtual monochromatic images across
a wide x-ray energy range (40–190 keV). Sev-
eral studies have shown that virtual mono-

AJR:209, October 2017 815


chromatic images have significant advantag-
es over conventional polychromatic images,
including less susceptibility to beam harden-
ing [17–19], improved image quality [20–22],
and metal artifact reduction [23, 24]. Further-
more, in interrogating the attenuation char-
acteristics of different materials at different
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved
x-ray energies, virtual monochromatic im-
ages can generate tissue-specific spectral at-
tenuation curves based on the unique K-edge
characteristics of materials with different ele-
mental composition, most notably iodine (K-
edge = 33.2 keV) [25].
Preliminary data suggest that dual-energy
quantitative spectral attenuation information
may effectively differentiate benign from
malignant disease processes, providing diag-
nostic value for patients with indeterminate
liver and soft-tissue lesions [26, 27], as well
as benign and metastatic lymph nodes [28,
29]. However, to our knowledge, no studies
to date have been specifically designed to as-
sess the diagnostic accuracy of spectral anal-
ysis for renal lesion characterization.
We postulated that rigorous quantitative
analysis of the spectral attenuation curves
may significantly improve the characteriza-
tion of renal lesions that are indeterminate on
the basis of conventional attenuation meth-
ods when encountered incidentally on non-
dedicated renal protocol single-phase single-
energy examinations. Therefore, the purpose
of this study was to determine whether sin-
gle-phase contrast-enhanced dual-energy
quantitative spectral analysis improves the
accuracy of diagnosis of small (< 4.0 cm) re-
nal lesions, compared with conventional sin-
gle-energy attenuation measurements.
Materials and Methods
This retrospective, single-center, HIPAA-com-
pliant study was approved by the institutional re-
view board at Duke University Medical Center,
and a waiver of informed consent was obtained.
Study Population
We retrospectively searched our tertiary refer-
ral academic center database (AW Server 2, re-
lease 5.5, GE Healthcare) for consecutive patients
who had undergone rapid kilovoltage-switching
single-source dual-energy CT of the abdomen be-
tween November 2011 and December 2014. Pa-
tients were eligible for enrollment if they under-
went clinically indicated dual-energy CT of the
kidneys during the nephrographic phase and had
at least one renal lesion with a craniocaudal di-
ameter between 1.0 and 4.0 cm. The 1.0-cm lower
threshold for lesion size was selected to minimize
816 AJR:209, October 2017
Patel et al.
potential confounding from partial volume aver-
aging effects, considering our reconstructed sec-
tion thickness of 0.5 cm [6].
The search yielded an initial target popula-
tion of 345 consecutive patients who were consid-
ered eligible for inclusion in the study. The clin-
ical indications for dual-energy CT examination
included characterization of an indeterminate re-
nal mass (n = 224), hematuria workup (n = 75), or
active surveillance of a small renal mass in an el-
derly patient or a patient with a high surgical risk
(n  = 46). Subjects were considered ineligible for
this study if the reference standard was inadequate
(see the Clinical Reference Standard section for
details); if the index lesion showed unequivocal
evidence of macroscopic fat (i.e., if regions had a
mean attenuation value of −20 HU or lower) or ho-
mogeneous attenuation of 70 HU or higher on the
unenhanced scan, both of which are findings in-
dicative of a benign cause (i.e., angiomyolipoma
and hemorrhagic cyst, respectively) [30]; if the pa-
tient’s total body weight was greater than 118 kg
(260 lb), because of prior evidence suggesting in-
accuracy of dual-energy material decomposition
analysis in larger patients [31]; if the image qual-
ity was deemed inadequate because of a subop-
timal injection technique or poor timing for the
acquisition of the nephrographic phase; or if the
patient had a known history of congenital or ac-
quired multifocal cystic renal disease or heredi-
tary renal cancer syndromes.
Clinical Reference Standard
The clinical reference standard was established
by a fourth-year radiology resident who had access
to patient electronic medical records, including
TABLE 1: CT Acquisition and Reconstruction Parameters
Single-Energy CT
CT Parameter (­Unenhanced Phase)
Detector configuration (mm) 64 × 0.625
Tube voltage (kVp) 120
Gantry revolution time (s) 0.6
Automatic exposure control On
Noise index 14
Tube current (mA) 200–500
Acquisition mode Single source, helical
Helical pitch 1.375
Reconstruction thickness (mm) 2.5
Reconstruction interval (mm) 2.5
Reconstruction algorithm Projection based
Reconstruction kernel Soft-tissue standard
Matrix size 512 × 512
Note—NA = not applicable.
pathology reports and all images obtained before
and after the index CT examination. For patients
with multiple renal lesions (range, two to four le-
sions), only the two largest lesions were selected
on the basis of the maximal diameter of the lesion,
to avoid data repeated-measurement bias in a sin-
gle patient (i.e., the clustering effect) [32].
MDCT Technique and Data Reconstruction
All contrast-enhanced CT examinations were
performed using a rapid kilovoltage-switching
single-source 64-MDCT scanner (Discovery
CT750 HD; GE Healthcare) (Table 1). This scan-
ner is equipped with a new garnet crystal scintilla-
tor detector with rapid optical response (Gemstone
Spectral Imaging, GE Healthcare) and a high-volt-
age generator with an ultrafast kilovoltage x-ray
tube that switches between 80 and 140 kVp every
0.5 ms. To ensure consistency and accuracy of CT
number measurements, the CT scanner was cali-
brated for attenuation values at the beginning of
each clinical day.
After acquisition of single-energy unenhanced
120-kVp images from the upper pole of both kid-
neys through the pubic symphysis, the patients
were scanned in dual-energy mode during the
nephrographic phase at a fixed time delay of 90
seconds after the beginning of IV contrast medi-
um injection. At our institution, it is standard clin-
ical practice to scan the nephrographic phase in
dual-energy mode. All patients received 150 mL
of an IV nonionic contrast medium with an iodine
concentration of 300 mg I/mL (iopamidol, Isovue
300, Bracco Diagnostics). The contrast medium
bolus was injected at a flow rate of 3 mL/s through
an 18- to 20-gauge cannula inserted into a vein in
Dual-Energy CT
­(Nephrographic Phase)
64 × 0.625
80/140
0.6
NA
NA
640
Single source, helical
1.375
2.5
2.5
Projection based
Soft-tissue standard
512 × 512
 Single-Phase Contrast-Enhanced Dual-Energy CT for Small Focal Renal Lesions
the antecubital fossa or the forearm of the patient,
with use of a dual-chamber mechanical power in-
jector (Empower; E-Z-Em).
For each patient, unenhanced single-energy
120-kVp datasets and contrast-enhanced nephro-
graphic dual-energy virtual monochromatic im-
age datasets at 40, 50, 70, and 140 keV, which
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved
were reconstructed using an advanced work-
station equipped with a commercially available
gemstone spectral imaging viewer (AW Server 2,
release 5.5), were used for analysis. Virtual mono-
chromatic images datasets were generated using
a projection-based material decomposition algo-
rithm [20]. All images were reconstructed at a
section thickness of 0.5 cm and were sent to our
departmental web-based multimodality image ar-
chive infrastructure (Candelis, Candelis) for sub-
sequent data analysis.
Quantitative Data Analysis
Quantitative image analysis was performed
independently by three abdominal imaging staff
radiologists, including a clinical fellow and two
board-certified staff and fellowship-trained ra-
diologists with 1, 4, and 5 years of experience in
genitourinary imaging, respectively. The three
readers were unaware of the final lesion diagno-
sis at the time of the data collection. Quantitative
measurements were performed using a dedicated
secondary workstation unit (Core2 Extreme Pro-
cessor X6800, Intel) equipped with a previously
validated custom Matlab-based software (Matlab,
version 2009a, Math-Works).
Attenuation measurements (expressed in Houns­
field units) were obtained by manually drawing
circular or ovoid ROIs encompassing as much of
the lesion surface area as possible for renal lesions
with homogeneous texture (mean number of pix-
els, 250; range, 100–900 pixels). In contrast, for re-
nal lesions with a complex cystic or heterogeneous
texture, ROIs had to encompass as much of the
most avidly enhancing part of a lesion as possible.
Care was also taken to avoid incorporating pixels
from outside of the mass when obtaining measure-
ments near the interface of the mass and the kidney
or the perirenal fat. For a given patient, the size,
shape, and position of the ROIs were kept constant
and were automatically propagated through all CT
datasets by applying the copy-and-paste function.
To ensure consistency and reproducibility of the
data, all measurements were repeated three times
on three contiguous images along the z-axis, and
the 10% winsorized mean value of the ROI mea-
surements was used for statistical analyses.
Lesion enhancement was measured by calcu-
lating the difference in attenuation between sin-
gle-energy polychromatic 120-kVp unenhanced
images and 70-keV contrast-enhanced virtual
AJR:209, October 2017 817
monochromatic image datasets, the latter of which
have been shown to be equivalent to 120 kVp [6,
33]. An attenuation difference of 15 HU or higher
was deemed suggestive of enhancement [13].
With the use of a commercially available gem-
stone spectral imaging viewer (AW Server 2, release
5.5), spectral attenuation curves were generated by
measuring the attenuation of each lesion across the
entire range of virtual monochromatic image ener-
gy levels (40, 50, 70, and 140 keV). Quantitative pa-
rameters were derived from the spectral attenuation
curves by measuring the slope of the spectral atten-
uation curve of each lesion between 40 and 50 keV
(λHU40–50 = [ROI40 — ROI50] / 10), 40 and 70 keV
(λHU40–70 = [ROI40 — ROI70] / 30), and 40 and 140
keV (λHU40–140 = [ROI40 — ROI140] / 100), where
ROI40, ROI50, ROI70, and ROI140 represent lesion at-
tenuation measurements obtained at 40, 50, 70, and
140 keV, respectively.
Statistical Analysis
The mean values of the lesion attenuation chang-
es and quantitative features of lesion spectral analy-
sis were compared between benign and malignant
renal lesions by use of a two-sample t test.
To account for the negative effects of misdiag-
nosing a malignant renal mass, the sensitivity of
each optimized diagnostic rule was set to 100%.
The threshold values were then optimized to maxi-
mize specificity. A leave-one-out cross-validation
analysis was performed to avoid optimistic bias
in the specificity value when choosing the opti-
mal threshold. Optimal thresholds for quantitative
measurements of lesion spectral curves were sum-
marized using mean values and standard devia-
tions. Estimated cross-validated specificity values
were compared using the exact McNemar test [34].
Linear discriminant analysis was performed
to investigate whether the combination of differ-
ent quantitative features from spectral analysis
of the lesions could improve diagnostic accuracy
for differentiating benign and malignant renal le-
sions. Linear discriminant analysis classifies ob-
servations using the optimal linear combination
of quantitative measurements that best separates
the different lesion groups. This analysis also pro-
TABLE 2: Demographic Characteristics of the Final Study Population
Finding Men (n = 95)
Patient age (y)
Mean ± SD 65 ± 11
Range 53–82
Effective diameter of patient (cm)a
Mean ± SD 31 ± 5
Range 25–43
aCalculated according to the recommendations of the American Association of Physicists in Medicine [53].
vides a quantitative form of curve fitting using
the collected data points. The accuracy of the lin-
ear discriminant analysis was assessed using the
leave-one-out cross-validation.
All statistical analyses were performed using
statistical computing and graphics software (R,
version 3.3.3, R Foundation).
Results
Study Population
Figure 1 shows the subject accrual flow-
chart, which is based on the Standards for
Reporting of Diagnostic Accuracy initiative
[35]. Of the 345 patients who were initially
deemed eligible for the study, 209 were ex-
cluded from the target population. The char-
acteristics of the final study population of
136 patients are presented in Table 2.
Clinical Reference Standard
A total of 144 renal lesions (mean max-
imal diameter, 2.5 cm; range, 1.0–3.9 cm)
were identified in the 136 patients (Table 3).
Lesions were classified as benign or malig-
nant in nature by use of a clinical reference
standard as further detailed in Appendix 1.
Among 105 patients, 111 lesions were di-
agnosed as benign (mean maximal diameter,
2.3 cm; range, 1.0–3.6 cm), including 80 sim-
ple and 23 hemorrhagic renal cysts, as well
as eight benign solid renal masses. Thirty-
three lesions in 31 patients were diagnosed
as malignant (mean maximal diameter, 3.3
cm; range, 1.5–3.9 cm).
Quantitative Data Analysis
The mean attenuation change for benign
renal lesions was significantly lower than
that for malignant renal masses (mean at-
tenuation change, 11.4  ± 12.3 [SD] HU vs
56.9 ± 30.3 HU; p < 0.0001) (Table 4). Simi-
larly, significant differences were noted be-
tween benign and malignant renal lesions for
all quantitative parameters derived from the
spectral attenuation curves (Table 4 and Fig.
2). Of note, the latter finding is also apparent
Women (n = 41) All Patients (n = 136)
58 ± 19 54 ± 16
48–75 48–82
31 ± 4 34.1 ± 4.3
25–41 25–43
 Patel et al.

TABLE 3: Characteristics of Renal Lesions and Clinical Reference Standard

Benign Renal Lesions Malignant Renal Lesions
Simple Cysts Hemorrhagic Oncocytoma Fat-Poor AMLa Clear Cell RCC Papillary RCC Chromophobe
Characteristic (n = 80) Cysts (n = 23) (n = 3) (n = 5) (n = 23) (n = 7) RCC (n = 3)
Lesion size (cm)
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved

Mean ± SD 2.5 ± 0.6 1.9 ± 0.4 3.1 ± 0.4 2.7 ± 0.3 3.1 ± 0.7 2.5 ± 0.5 3.1 ± 0.3
Range 1.0–3.5 1.3–3.1 2.3–3.6 1.8–3.2 2.1–3.9 1.9–3.1 2.8–3.6
Partial or radical nephrectomyb
No. (%) of lesions 2 (2.5) 2 (8.7) 3 (100) 3 (60) 23 (100) 3 (42.9) 3 (100)
No. of patientsc 2 2 3 3 21 3 3
Lesion size (cm), range 2.9–3.5 2.7–3.1 2.3–3.6 2.4–3.2 2.1–3.9 2.3–3.1 2.8–3.6
Time interval range (d) 23–40 16–33 19–27 9–31 8–21 15–34 23–30
Percutaneous renal biopsyb
No. (%) of lesions 0 0 0 2 (40) 0 4 (57.1) 0
No. of patientsc 0 0 0 2 0 4 0
Lesion size (cm), range 0 0 0 1.8–2.5 0 1.9–2.6 0
Time interval range (d) 0 0 0 1–14 0 17–25 0
Extended imaging follow-up
No. (%) of lesions 78 (97.5) 21 (91.3) 0 0 0 0 0
No. of patients 72 21 0 0 0 0 0
Lesion size (cm), range 1.0–3.4 1.3–2.9 0 0 0 0 0
Follow-up duration (d), range 19–36 24–32 0 0 0 0 0
Note—AML = angiomyolipoma, RCC = renal cell carcinoma.
aFat-poor AMLs had less than 25% fat cells per high-powered field on histopathologic analysis.
bThe mean time between the initial imaging study and histopathologic analysis was 21 days (range, 8–40 days) for partial or radical nephrectomy and 14 days (range, 1–25

days) for percutaneous renal biopsy.

cStability in morphologic findings and size (i.e., less than 10% variation in the transverse maximum diameter of a lesion) for at least 18 months on follow-up with CT or MRI

(mean follow-up, 22 months; range, 19–36 months).

TABLE 4: Attenuation Changes and Dual-Energy Quantitative Spectral Attenuation Parameters and Corresponding
Sensitivity and Specificity Values for Distinguishing Benign From Malignant Renal Lesions Using a Leave-
One-Out Cross-Validation Analysis
Sensitivity (%) Specificity (%)
No. of Benign No. of Malignant Optimal Threshold Values Based
Lesions, Lesions, Based on Cross- SD on Cross-
Imaging Metric Mean ± SD Mean ± SD pa Meanb 95% CIs Validation Threshold Validationb 95% CI pc
Attenuation 11.2 ± 11.4 50.5 ± 23.2 < 0.0001 94 (31/33) 80–99 NA NA 86 (96/111) 79–92 NA
change (HU)d
λHU40–50 3.0 ± 2.6 9.8 ± 4.1 < 0.0001 100 (33/33) 100–100 1.1 0.11 12 (8/111) 6–19 < 0.0001e
λHU40–70 1.7 ± 1.4 5.7 ± 1.7 < 0.0001 100 (33/33) 100–100 3.1 0.02 93 (103/111) 86–97 0.02
λHU40–140 0.9 ± 0.2 2.8 ± 0.8 < 0.0001 100 (33/33) 100–100 1.6 0.01 93 (103/111) 86–97 0.02
Note—NA = not applicable.
aThe p values for differences in mean values were calculated using the t test.
bValues in parentheses denote number of lesions with finding/total no. of lesions evaluated.
cThe p values were calculated for differences in specificity between attenuation measurements and dual-energy quantitative spectral attenuation parameters at the

highest sensitivity value of each imaging method. These values were derived using the exact McNemar test.
dLesion enhancement was defined as an attenuation difference of 15 HU or greater between the contrast-enhanced and unenhanced scans.
eThe p value for λHU
40–50 indicates significantly worse specificity compared with conventional attenuation measurements.

from visual analysis of the shape of the spec- ation parameters between benign and malig- nal masses were correctly identified, yield-
tral attenuation curves (Fig. 3). No significant nant solid renal masses (Figs. 2 and 3). ing a sensitivity of 94% (31/33; 95% CI, 80–
differences were observed in mean attenua- With the use of an enhancement threshold 99%) (Table 4). Two malignant lesions (both
tion changes or quantitative spectral attenu- of 15 HU or greater, 31 of 33 malignant re- of which were papillary RCCs) had enhance-

818 AJR:209, October 2017

 Single-Phase Contrast-Enhanced Dual-Energy CT for Small Focal Renal Lesions

ment levels that were just below the limit
of detection (differences of 13 and 14 HU)
(Fig. 3). Furthermore, of the 111 renal lesions
proved to be benign, 21 lesions in 21 patients,
including eight benign solid renal masses
(three oncocytomas and five fat-poor angio-
myolipomas), 11 simple renal cysts, and two
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved
In our patient population, the use of an op-
timized threshold (3.1 for λHU40–70 and 1.6
for λHU40–140) yielded an improvement in
specificity from 81% (95% CI, 73–88%), with
the use of conventional attenuation measure-
ments, to 93% (95% CI, 86–97%), with the
use of dual-energy quantitative spectral anal-
sistency of quantitative measurements may
yield higher diagnostic accuracy for charac-
terization of renal lesions with the use of du-
al-energy spectral analysis.
Along with improved specificity, another
advantage of dual-energy quantitative spec-
tral analysis was the higher (although not sta-

hemorrhagic renal cysts, showed evidence of ysis (p = 0.02). This improvement was related
enhancement, yielding a false-positive rate to a lower number of false-positive findings
of 19% and a specificity of 81% (90/111; 95% caused by misleading levels of enhancement
CI, 73–88%) (Figs. 4 and 5). (i.e., pseudoenhancement) in 12% of small
Compared with conventional single-en- benign renal cysts (13/113). These results may
ergy attenuation measurements, dual-en- have important clinical implications because
ergy quantitative spectral analysis yielded renal cyst pseudoenhancement is a common
significantly higher specificity for renal le- reason for repeat imaging resulting in in-
sions characterization, with a specificity of creased costs, patient anxiety, and radiation
93% (103/111; 95% CI, 86–97%) with the use exposure in clinical practice [36].
of both λHU40–70 and λHU40–140 (p  = 0.02) Many confounding factors have been
(Table 4). This improvement was related to shown to negatively affect the precision and
a decrease in false-positive findings second- accuracy of conventional attenuation mea-
ary to misleading measurements of enhance- surements of enhancement in renal lesions,
ment in 13 small benign renal cysts (mean including (but not limited to) the proper-
cyst diameter, 1.3 cm; range, 1.0–2.6 cm) ties of the x-ray spectrum and the detector
(Fig. 4). Sensitivity values were also high- energy response, the selection of the recon-
er for λHU40–70 and λHU40–140 (sensitivity, struction kernel, the amount and rate of the
100%; 95% CI, 100–100%, for both meth- contrast material injected, the imaging de-
ods) compared with conventional attenuation lay, the lesion size and location, and patient
measurements; however, this difference was body size [37, 38]. By interrogating the at-
not statistically significant (Fig. 3). tenuation changes of various tissues at differ-
Our data showed a very strong correlation ent x-ray energies, dual-energy quantitative
among the different quantitative parameters spectral analysis may partially overcome the
derived from the spectral attenuation curves confounding effect of beam-hardening ener-
of the lesions (r ≥ 0.95; Table 5). Expectedly, gy-shift phenomena, which are responsible
the results of the linear discriminant analysis for spurious attenuation changes (i.e., pseu-
showed no improvement in sensitivity, speci- doenhancement) in small renal lesions on
ficity, or both using any combination of the conventional polychromatic images [10, 14,
quantitative spectral parameters. 16]. Furthermore, by performing quantita-
tive analysis of a single contrast-enhanced
Discussion data­ set, dual-energy quantitative spectral
The results of the present study showed analysis overcomes any operator-dependent
significantly improved specificity for char- variability because of differences in ROI po-
acterization of small (<  4.0 cm) renal le- sitioning between unenhanced and contrast-
sions with the use of single-phase contrast- enhanced images, during conventional mea-
enhanced dual-energy quantitative spectral surements of enhancement [6]. We postulate
analysis, compared with conventional single- that the combination of decreased suscepti-
energy attenuation measurements (p = 0.02). bility to beam hardening and improved con-
TABLE 5: Pearson Correlation Coefficients Among Dual-Energy Quantitative
Spectral Attenuation Parameters
Spectral Attenuation
Spectral Attenuation Curve Parameter
Curve Parameter λHU40–50 λHU40–70 λHU40–140
λHU40–50 1.00 0.95 0.95
λHU40–70 0.95 1.00 0.98
λHU40–140 0.95 0.98 1.00
Note—Data are correlation coefficients (r values). The slope of each curve was measured between 40 and 50
keV (λHU 40–50), 40 and 70 keV (λHU40–70), and 40 and 140 keV (λHU40–140).
tistically significant) sensitivity for detecting
malignant renal neoplasms with subtle lev-
els of enhancement. In the present study, two
minimally enhancing papillary RCCs, which
could not be diagnosed on the basis of con-
ventional single-energy attenuation measure-
ments, were correctly classified on the basis
of their dual-energy quantitative spectral pro-
files. Our results compare favorably with sim-
ilar findings in studies of iodine quantifica-
tion for the diagnosis of minimally enhancing
renal masses [10–16]. Future studies may be
warranted to determine whether dual-energy
quantitative spectral analysis can act in syn-
ergy with iodine quantification techniques to
improve characterization of renal lesions.
Another important finding of this study is
that despite the additional information pro-
vided by quantitative spectral analysis, du-
al-energy CT may still fail to differentiate
malignant renal masses from the small (but
not negligible) fraction of benign solid renal
masses, which account for up to 25% of inci-
dental renal lesions smaller than 3.0 cm [39].
This challenging clinical task has become
critical with the increasing number of inci-
dentally discovered small renal masses. Al-
though some diagnostic features may provide
clues to diagnosis of a benign renal mass,
such as the presence of fat, homogeneous en-
hancement, or hyperattenuation relative to
the renal parenchyma on unenhanced scans
[6–9], most incidental renal masses cannot
be confidently diagnosed at imaging, leading
to additional workup or more invasive tests,
such as percutaneous biopsy or potentially
morbid surgical and ablation procedures [9].
It can be argued that one could simply
measure the difference between the attenu-
ation on virtual unenhanced (VUE) images
derived from single-phase contrast-enhanced
dual-energy images and the attenuation on
the contrast-enhanced images to characterize
a renal lesion. However, a few notable limi-
tations of VUE images exist, compared with
true unenhanced images [40]. Some investi-
gators have suggested that attenuation values
are higher on VUE images than on true un-
enhanced images in renal lesion evaluation,
although this did not ultimately affect the
categorization of the lesion as benign or ma-

AJR:209, October 2017 819


lignant [16]. Others have found that attenua-
tion values for renal cysts and renal parenchy-
ma on VUE images are significantly lower
than such values on true unenhanced imag-
es [41]. Moreover, the iodine subtraction al-
gorithms used to derive the VUE images may
be modified. Inaccurate calibration of the al-
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved
gorithm (e.g., preassigned attenuation values
for soft tissue and fat) could lead to errors in
iodine subtraction, ultimately affecting the
evaluation of an indeterminate renal lesion
[40, 41]. It has been proposed that the subtrac-
tion algorithms should be corrected for body
habitus and organ enhancement because of
their influence on differences in attenuation
values on VUE images compared with true
unenhanced images [42]. Iodine quantifica-
tion and quantitative spectral analysis could
therefore provide alternative means for renal
lesion evaluation. Other known limitations
of VUE images include image quality issues
and lack of ability to detect small (< 4 mm)
urinary calculi [43–46].
In addition to its retrospective design, some
potential limitations of this study merit con-
sideration. First, the findings reflect our expe-
rience from a relatively small retrospectively
selected patient cohort. This limitation is fur-
ther compounded by the relatively small num-
ber of those solid renal masses with a lower
prevalence in the general population, such as
fat-poor angiomyolipomas, oncocytomas, and
chromophobe RCCs. Additionally, the range of
pathology of the solid renal masses that were
included is narrow compared to other studies
evaluating the role of CT in characterizing re-
nal lesions [47, 48]. Second, we performed a
posteriori analysis to establish the optimal di-
agnostic thresholds for renal lesion character-
ization using dual-energy quantitative spectral
analysis. Although this approach may lead to a
significant risk of overinterpretation bias [49],
a leave-one-out cross-validation analysis was
used in an attempt to mitigate such bias [50].
Third, no histologic proof was available for 99
of 111 renal lesions (89%), which were regard-
ed as benign in our study. The absence of sus-
picious imaging features, however, combined
with the small size of many of these lesions,
frequently made biopsy unnecessary or im-
practical from an ethical standpoint. Further-
more, for all patients whose renal lesions were
not verified histologically, lesion stability was
confirmed at follow-up with CT and MRI for
a minimum of 18 months. Although the lack
of lesion growth does not confirm the benign
nature of a renal mass, the chosen follow-up in-
terval has been accepted as satisfactory prac-
820 AJR:209, October 2017
Patel et al.
tice for the clinical diagnosis of indeterminate
renal lesions [8].
Additionally, it remains to be determined
whether our study results can be replicat-
ed with other dual-energy CT platforms, be-
cause this study involved only patients who
underwent scanning using rapid kilovoltage-
switching single-source dual-energy CT. This
is a particularly relevant issue given the re-
cently published experimental data indicat-
ing significant variability when comparing
quantitative dual-energy attenuation mea-
surements of renal lesions on virtual mono-
chromatic images between single-source pro-
jection-based and dual-source image-based
imaging platforms [51, 52]. A recent phantom
study not only showed significant variation in
attenuation numbers derived from lesions that
were scanned using two different DECT plat-
forms but also showed that both platforms re-
sponded differently to varying lesion size and
iodine concentrations. In clinical practice, this
could potentially affect patients who have un-
dergone serial examinations performed us-
ing different platforms [51]. Similarly, varia-
tions in other quantitative lesion metrics are
also known to occur with the use of different
DECT platforms. For example, iodine concen-
tration thresholds of 2.0 and 0.5 mg/mL have
been suggested for renal lesion enhancement
using rapid kilovoltage-switching single-
source DECT and dual-source DECT, respec-
tively [12, 14, 31].
In conclusion, the present study shows that
single-phase contrast-enhanced dual-energy
quantitative spectral analysis significantly
improves specificity for the characterization
of small (< 4.0 cm) renal lesions, compared
with conventional single-energy attenuation
measurements. This improvement in spec-
ificity may decrease the diagnostic uncer-
tainty and public health burden related to
unnecessary imaging workup for small inde-
terminate renal lesions.
References
1. Hara AK, Johnson CD, MacCarty RL, Welch TJ.
Incidental extracolonic findings at CT colonogra-
phy. Radiology 2000; 215:353–357
2. Tada S, Yamagishi J, Kobayashi H, Hata Y, Kobari
T. The incidence of simple renal cyst by computed
tomography. Clin Radiol 1983; 34:437–439
3. Jayson M, Sanders H. Increased incidence of ser-
endipitously discovered renal cell carcinoma.
Urology 1998; 51:203–205
4. Konnak JW, Grossman HB. Renal cell carcinoma as
an incidental finding. J Urol 1985; 134:1094–1096
5. Luciani LG, Cestari R, Tallarigo C. Incidental re-
nal cell carcinoma: age and stage characterization
and clinical implications—study of 1092 patients
(1982–1997). Urology 2000; 56:58–62
6. Israel GM, Bosniak MA. How I do it: evaluating
renal masses. Radiology 2005; 236:441–450
7. Israel GM, Silverman SG. The incidental renal
mass. Radiol Clin North Am 2011; 49:369–383
8. Silverman SG, Israel GM, Herts BR, Richie JP. Man-
agement of the incidental renal mass. R ­ adiology
2008; 249:16–31
9. Silverman SG, Gan YU, Mortele KJ, Tuncali K,
­Cibas ES. Renal masses in the adult patient: the role
of percutaneous biopsy. Radiology 2006; 240:6–22
10. Ascenti G, Mileto A, Krauss B, et al. Distinguish-
ing enhancing from nonenhancing renal masses
with dual-source dual-energy CT: iodine quantifi-
cation versus standard enhancement measure-
ments. Eur Radiol 2013; 23:2288–2295
11. Brown CL, Hartman RP, Dzyubak OP, et al. Dual-
energy CT iodine overlay technique for character-
ization of renal masses as cyst or solid: a phantom
feasibility study. Eur Radiol 2009; 19:1289–1295
12. Chandarana H, Megibow AJ, Cohen BA, et al. Io-
dine quantification with dual-energy CT: phantom
study and preliminary experience with renal
masses. AJR 2011; 196:[web]W693–W700
13. Graser A, Becker CR, Staehler M, et al. Single-
phase dual-energy CT allows for characterization
of renal masses as benign or malignant. Invest
­Radiol 2010; 45:399–405
14. Kaza RK, Caoili EM, Cohan RH, Platt JF. Distin-
guishing enhancing from nonenhancing renal le-
sions with fast kilovoltage-switching dual-energy
CT. AJR 2011; 197:1375–1381
15. Mileto A, Marin D, Ramirez-Giraldo JC, et al.
Accuracy of contrast-enhanced dual-energy
MDCT for the assessment of iodine uptake in re-
nal lesions. AJR 2014; 202:[web]W466–W474
16. Neville AM, Gupta RT, Miller CM, Merkle EM,
Paulson EK, Boll DT. Detection of renal lesion
enhancement with dual-energy multidetector CT.
Radiology 2011; 259:173–183
17. Jung DC, Oh YT, Kim MD, Park M. Usefulness of the
virtual monochromatic image in dual-energy spectral
CT for decreasing renal cyst pseudoenhancement: a
phantom study. AJR 2012; 199:1316–1319
18. Marin D, Fananapazir G, Mileto A, Choudhury
KR, Wilson JM, Nelson RC. Dual-energy multi-
detector row CT with virtual monochromatic im-
aging for improving patient-to-patient uniformity
of aortic enhancement during CT angiography: an
in vitro and in vivo study. Radiology 2014;
272:895–902
19. Mileto A, Nelson RC, Samei E, et al. Impact of
dual-energy multi-detector row CT with virtual
monochromatic imaging on renal cyst pseudoen-
hancement: in vitro and in vivo study. Radiology
2014; 272:767–776
 Single-Phase Contrast-Enhanced Dual-Energy CT for Small Focal Renal Lesions
20. Goodsitt MM, Christodoulou EG, Larson SC. Ac-
curacies of the synthesized monochromatic CT
numbers and effective atomic numbers obtained
with a rapid kVp switching dual energy CT scan-
ner. Med Phys 2011; 38:2222–2232
21. Matsumoto K, Jinzaki M, Tanami Y, Ueno A,
­Yamada M, Kuribayashi S. Virtual monochro-
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved
matic spectral imaging with fast kilovoltage
switching: improved image quality as compared
with that obtained with conventional 120-kVp CT.
­Radiology 2011; 259:257–262
22. Yu L, Christner JA, Leng S, Wang J, Fletcher JG,
McCollough CH. Virtual monochromatic imag-
ing in dual-source dual-energy CT: radiation dose
and image quality. Med Phys 2011; 38:6371–6379
23. Bamberg F, Dierks A, Nikolaou K, Reiser MF, Beck-
er CR, Johnson TR. Metal artifact reduction by dual
energy computed tomography using monoenergetic
extrapolation. Eur Radiol 2011; 21:1424–1429
24. Lee MJ, Kim S, Lee SA, et al. Overcoming arti-
facts from metallic orthopedic implants at high-
field-strength MR imaging and multi-detector CT.
RadioGraphics 2007; 27:791–803
25. Patino M, Prochowski A, Agrawal MD, et al. Ma-
terial separation using dual-energy CT: current
and emerging applications. RadioGraphics 2016;
36:1087–1105
26. Sun X, Shao X, Chen H. The value of energy spec-
tral CT in the differential diagnosis between benign
and malignant soft tissue masses of the musculo-
skeletal system. Eur J Radiol 2015; 84:1105–1108
27. Wang Q, Shi G, Qi X, Fan X, Wang L. Quantita-
tive analysis of the dual-energy CT virtual spec-
tral curve for focal liver lesions characterization.
Eur J Radiol 2014; 83:1759–1764
28. Liu H, Yan F, Pan Z, et al. Evaluation of dual en-
ergy spectral CT in differentiating metastatic
from non-metastatic lymph nodes in rectal cancer:
initial experience. Eur J Radiol 2015; 84:228–234
29. Liu X, Ouyang D, Li H, et al. Papillary thyroid
cancer: dual-energy spectral CT quantitative pa-
rameters for preoperative diagnosis of metastasis
to the cervical lymph nodes. Radiology 2015;
275:167–176
30. Jonisch AI, Rubinowitz AN, Mutalik PG, Israel
AJR:209, October 2017 821
GM. Can high-attenuation renal cysts be differen-
tiated from renal cell carcinoma at unenhanced
CT? Radiology 2007; 243:445–450
31. Patel BN, Alexander L, Allen B, et al. Dual-ener-
gy CT workflow: multi-institutional consensus on
standardization of abdominopelvic MDCT proto-
cols. Abdom Radiol (NY) 2017; 42:676–687
32. Sica GT. Bias in research studies. Radiology
2006; 238:780–789
33. Yu L, Leng S, McCollough CH. Dual-energy CT-
based monochromatic imaging. AJR 2012; 199:S9–
S15
34. Zar JH. Biostatistical analysis, 5th ed. Upper
Saddle River, NJ: Pearson Prentice-Hall, 2010:944
35. Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD
2015: an updated list of essential items for report-
ing diagnostic accuracy studies. Radiology 2015;
277:826–832
36. O’Connor SD, Silverman SG, Ip IK, Maehara CK,
Khorasani R. Simple cyst-appearing renal masses
at unenhanced CT: can they be presumed to be
benign? Radiology 2013; 269:793–800
37. Fletcher JG, Leng S, Yu L, McCollough CH.
Dealing with uncertainty in CT images. ­Radiology
2016; 279:5–10
38. Megibow AJ, Chandarana H, Hindman NM. In-
creasing the precision of CT measurements with du-
al-energy scanning. Radiology 2014; 272:618–621
39. Ochodo EA, de Haan MC, Reitsma JB, Hooft L,
Bossuyt PM, Leeflang MM. Overinterpretation
and misreporting of diagnostic accuracy studies:
evidence of “spin.” Radiology 2013; 267:581–588
40. Heilbrun ME, Remer EM, Casalino DD, et al.
ACR Appropriateness Criteria indeterminate re-
nal mass. J Am Coll Radiol 2015; 12:333–341
41. Sahni VA, Shinagare AB, Silverman SG. Virtual
unenhanced CT images acquired from dual-ener-
gy CT urography: accuracy of attenuation values
and variation with contrast material phase. Clin
Radiol 2013; 68:264–271
42. Miller CM, Gupta RT, Paulson EK, et al. Effect of
organ enhancement and habitus on estimation of
unenhanced attenuation at contrast-enhanced du-
al-energy MDCT: concepts for individualized and
organ-specific spectral iodine subtraction strate-
gies. AJR 2011; 196:[web]W558–W564
43. Karlo CA, Gnannt R, Winklehner A, et al. Split-
bolus dual-energy CT urography: protocol opti-
mization and diagnostic performance for the de-
tection of urinary stones. Abdom Imaging 2013;
38:1136–1143
44. Moon JW, Park BK, Kim CK, Park SY. Evaluation
of virtual unenhanced CT obtained from dual-
energy CT urography for detecting urinary stones.
Br J Radiol 2012; 85:e176–e181
45. Park JJ, Park BK, Kim CK. Single-phase DECT
with VNCT compared with three-phase CTU in
patients with haematuria. Eur Radiol 2016;
26:3550–3557
46. Song KD, Kim CK, Park BK, Kim B. Utility of io-
dine overlay technique and virtual unenhanced im-
ages for the characterization of renal masses by du-
al-energy CT. AJR 2011; 197:[web]W1076–W1082
47. Millet I, Doyon FC, Hoa D, et al. Characterization
of small solid renal lesions: can benign and malig-
nant tumors be differentiated with CT? AJR 2011;
197:887–896
48. Zhang J, Lefkowitz RA, Ishill NM, et al. Solid re-
nal cortical tumors: differentiation with CT.
­Radiology 2007; 244:494–504
49. Frank I, Blute ML, Cheville JC, Lohse CM,
­Weaver AL, Zincke H. Solid renal tumors: an
analysis of pathological features related to tumor
size. J Urol 2003; 170:2217–2220
50. Venables WN, Ripley BD. Modern applied statistics
with S-Plus, 2nd ed. New York, NY: Springer, 1997
51. Mileto A, Barina A, Marin D, et al. Virtual mono-
chromatic images from dual-energy multidetector
CT: variance in CT numbers from the same lesion
between single-source projection-based and dual-
source image-based implementations. Radiology
2016; 279:269–277
52. Kaza RK, Platt JF. Renal applications of dual-en-
ergy CT. Abdom Radiol (NY) 2016; 41:1122–1132
53. American Association of Physicists in Medicine
(AAPM) Task Group 204. Size-specific dose esti-
mates (SSDE) in pediatric and adult body CT ex-
aminations. AAPM website. www.aapm.org/
pubs/reports/RPT_204.pdf. Published 2011. Ac-
cessed June 25, 2017
(Appendix and figures start on next page)
 Patel et al.

APPENDIX 1: Clinical Reference Standard

Benign Renal Lesions
Among 105 patients, 111 lesions were di-
agnosed as benign (mean maximal diameter,
2.3 cm; range, 1.0–3.6 cm); this group of pa-
tients included 99 patients with one lesion and
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved
mean time between the initial imaging study
and surgery was 26 days (range, 7–45 days).
Benign lesions were further categorized as
simple or hemorrhagic renal cysts or as be-
nign solid renal masses. Eighty lesions in 74
neous biopsy (two lesions in two patients),
including three oncocytomas and five fat-poor
angiomyolipomas (< 25% fat cells per high-
powered field at histopathologic analysis).

six patients with two lesions. In patients with
lesions presumed to be benign, the proof of
diagnosis was obtained on the basis of one of
the following conditions: stability in morpho-
logic findings or lack of lesion growth (for 99
lesions in 93 patients), which was defined by
an increase of less than 10% in the transverse
maximum diameter of the lesion for at least 18
months on follow-up with CT or MRI (mean
follow-up, 22 months; range, 19–36 months)
or confirmed by histopathologic analysis after
surgery (10 lesions in 10 patients) or percuta-
neous biopsy (two lesions in two patients). The
patients were diagnosed as simple renal cysts
on the basis of lesion attenuation of less than
20 HU (mean attenuation, 5 HU; range, −5 to
14 HU) on unenhanced CT images. Twenty-
three lesions in 23 patients were diagnosed
as hemorrhagic cysts on the basis of lesion
attenuation of 20 HU or higher (mean atten-
uation, 36 HU; range, 21–65 HU) on unen-
hanced CT images (21 lesions in 21 patients)
or after surgery (two lesions in two patients).
Eight lesions in eight patients were diag-
nosed as benign solid renal masses after sur-
gery (six lesions in six patients) or percuta-
Malignant Renal Lesions
In 31 patients, 33 lesions were diagnosed
as malignant (mean maximal diameter, 3.3
cm; range, 1.5–3.9 cm), including 29 patients
with one lesion and two patients with two le-
sions. Proof of diagnosis was obtained on
the basis of histologic confirmation at sur-
gery (29 lesions in 27 patients) or percuta-
neous biopsy (four lesions in four patients).
Final histopathologic diagnoses included 23
clear cell renal cell carcinomas (RCCs), sev-
en papillary RCCs (type 1), and three chro-
mophobe RCCs.

Target population
345 patients who underwent dual-energy CT with small (1–4 cm) renal lesion(s) Fig. 1—Flowchart
of study enrollment
population based
Excluded patients (n = 209)
on recommended
No proof of tumor burden (n = 121)
standards for
Benign findings on CT (n = 24)
reporting diagnostic
Body weight > 118 kg (n = 46)
accuracy [35]. Benign
Suboptimal image quality (n = 15)
findings on CT were
Hereditary renal cancer syndrome (n = 3)
defined as presence
of macroscopic fat
Final study population (i.e., regions with
136 patients with 144 renal lesions mean attenuation
of less than −20 HU)
or homogeneous
attenuation of
greater than 70 HU
Benign lesions Malignant lesions on unenhanced
111 lesions in 105 patients 33 lesions in 31 patients scan. Follow-up
was performed with
CT, MRI, or both
after minimum of
Surgery Percutaneous biopsy Follow-up Surgery Percutaneous biopsy 18 months (mean,
3 Fat-poor AMLs 2 Fat-poor AMLs 78 Simple cysts 23 Clear cell RCCs 4 Papillary RCCs 22 months; range,
3 Oncocytomas 21 Hemorrhagic cysts 3 Papillary RCCs 19–36 months). AML =
2 Hemorrhagic cysts 3 Chromophobe RCCs angiomyolipoma,
2 Simple cysts RCC = renal cell
carcinoma.

822 AJR:209, October 2017

 Single-Phase Contrast-Enhanced Dual-Energy CT for Small Focal Renal Lesions

10 Simple cyst 5 Simple cyst

Hemorrhagic cyst Hemorrhagic cyst
Clear cell RCC Clear cell RCC
Papillary RCC Papillary RCC
Slope of Spectral Attenuation Curves

Slope of Spectral Attenuation Curves

Chromophobe RCC Chromophobe RCC

Measured Between 40 and 140 keV

8 4
Measured Between 40 and 70 keV

Lipid-poor AML Lipid-poor AML

Oncocytoma Oncocytoma
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved

6 3

4 2

2 1

0 0

0 5 10 15 0 5 10 15
Slope of Spectral Attenuation Curves Slope of Spectral Attenuation Curves
Measured Between 40 and 50 keV Measured Between 40 and 50 keV

A B
Fig. 2—Distribution of dual-energy quantitative spectral attenuation parameters
5 Simple cyst for different subgroups of benign and malignant renal lesions, including slope of
Hemorrhagic cyst spectral attenuation curves. Vertical dotted line denotes cross-validated optimal
Clear cell RCC threshold. RCC = renal cell carcinoma, AML = angiomyolipoma.
Papillary RCC A, Scatterplot shows data for slopes of spectral attenuation curves measured
Slope of Spectral Attenuation Curves

Chromophobe RCC
Measured Between 40 and 140 keV

4 between 40 and 50 keV and 40 and 70 keV.

Lipid-poor AML B, Scatterplot shows data for slopes of spectral attenuation curves measured
Oncocytoma between 40 and 50 keV and 40 and 140 keV.
C, Scatterplot shows data for slopes of spectral attenuation curves measured
3 between 40 and 70 keV and 40 and 140 keV.

0 2 4 6 8 10
Slope of Spectral Attenuation Curves
Measured Between 40 and 70 keV

AJR:209, October 2017 823

 Patel et al.

Fig. 3—Dual-energy spectral attenuation curves for benign and solid renal
lesions. Please note that, to improve visualization, different subtypes of renal cell
500 carcinoma (RCC) were displayed together.
RCC
Simple cyst
Hemorrhagic cyst
400 Benign mass
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved

Attenuation (HU)

300

200

100

40 60 80 100 120 140

Energy (keV)

A B
325 Spectral Attenuation Curve
300
275
250
225
Attenuation (HU)

200
175
150
125
100
75
5.00 mm
50
25 keV: 70 V1: Mean 91.06. SD: 17.23

5.00 mm
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
145

Kiloelectron Volts W: 400 L: 40

C
Fig. 4—76-year-old woman with 2.0-cm biopsy-proven left papillary renal cell carcinoma.
A and B, Comparison of attenuation of lesion (arrow, A and B) on unenhanced (attenuation, 37 HU) (A) and contrast-enhanced (attenuation, 51 HU) (B) transverse CT
images shows minimal change in attenuation (14 HU), which is finding suggestive of nonenhancing renal lesion.
C, With ROI (red square enclosing yellow circle, right) placed within lesion, spectral attenuation curve (left) was derived from dual-energy dataset (right) using advanced
workstation equipped with commercially available gemstone spectral imaging viewer (AW Server 2, release 5.5, GE Healthcare). Single-phase contrast-enhanced dual-
energy spectral attenuation analysis suggests internal vascular component within lesion (slope of spectral attenuation curve of lesion measured between 40 and 70
keV, 3.3; slope of spectral attenuation curve of lesion measured between 40 and 140 keV, 1.8), contrary to findings in A and B that suggested nonenhancing renal lesion.
Patient underwent partial nephrectomy because of interval increase in lesion size during follow-up at 12 months. W = window, L = level.

824 AJR:209, October 2017

 Single-Phase Contrast-Enhanced Dual-Energy CT for Small Focal Renal Lesions
Downloaded from www.ajronline.org by 120.188.78.75 on 10/11/17 from IP address 120.188.78.75. Copyright ARRS. For personal use only; all rights reserved

A B
Spectral Attenuation Curve
650
600
550
500
Attenuation (HU)

450
400
350
300
250
200
150
5.00 mm
100
50 keV: 70 V3: Mean 206.83, SD: 13.68
V2: Mean 73.72, SD: 13.56
0
5.00 mm
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
125
130
135
140
145

Kiloelectron Volts W: 400 L: 40

C
Fig. 5—76-year-old woman with 1.2-cm benign right renal cyst.
A and B, Comparison of attenuation of lesion (arrow, A and B) on unenhanced (attenuation, 34 HU) (A) and contrast-enhanced (attenuation, 52 HU) (B) transverse CT
images shows mildly increased change in attenuation (18 HU), which is finding that may be compatible with low level of enhancement in lesion.
C, Quantitative measurements derived from dual-energy spectral attenuation analysis (slope of spectral attenuation curve of lesion measured between 40 and 70 keV, 1.1;
slope of spectral attenuation curve of lesion measured between 40 and 140 keV, 0.5) show shallower slope of spectral attenuation curve of renal lesion (yellow curve, left)
relative to that of normal renal parenchyma (red curve, left), contrary to findings in A and B that were compatible with low level of lesion enhancement. Patient underwent
long-term imaging follow-up (duration, 22 months) that showed stability of lesion. Red square (right) and yellow circle (right) are ROIs corresponding to red and yellow
spectral attenuation curves (left). W = window, L = level.

AJR:209, October 2017 825