670 AJR:199, September 2012

In a previous study [14], with a set of right

cylindric phantoms of various diameters, we
showed that radiation can be reduced by a
combined reduction in kilovoltage and tube
currenttime product, while preserving im-
age noise, and that technique optimization
curves could be created to guide clinical im-
aging. Furthermore, because of wide varia-
tions in pediatric body shape, radiation dos-
es are better calibrated to patient abdominal
circumference (AC) than to age or weight.
However, the cross-sectional shape of the
pediatric abdomen tends to be oval rather than
circular, especially when a child gets older.
Also, in most abdominal CT studies, iodin-
ated contrast agent is used to better visual-
ize certain abnormalities. To take advantage
of iodines K-edge (33 keV), low kilovoltage
should be used in CT [15]. However, lower ki-
lovoltage results in higher image noise; there-
fore, it may not be suitable for imaging large
patients where image noise dominates the
contrast. To better characterize both noise and
Optimization of Kilovoltage and
Tube CurrentExposure Time
Product Based on Abdominal
Circumference: An Oval Phantom
Study for Pediatric Abdominal CT
Frank Dong
1
William Davros
1
Jessica Pozzuto
2
Janet Reid
3
Dong F, Davros W, Pozzuto J, Reid J
1
Section of Medical Physics, Cleveland Clinic, 17325
Euclid Ave, CL2-MP, Cleveland, OH 44112. Address
correspondence to F. Dong (dongf3@ccf.org).
2
Department of Biomedical Engineering, Case Western
Reserve University, Cleveland, OH.
3
Section of Pediatric Radiology, Childrens Hospital
Cleveland Clinic, Cleveland, OH.
Pedi at ri c I magi ng Ori gi nal Research
AJR 2012; 199:670676
0361803X/12/1993670
American Roentgen Ray Society
D
iagnostic x-ray imaging modali-
ties, such as CT, serve an impor-
tant role in patient care [1]. In
2000, The New England Journal
of Medicine named medical imaging one of
the top 11 innovations for the past 1000 years
[2]. However, the potential risk of radiation-
induced cancers has caused concerns among
patients, physicians, and the general public
[37]. Pediatric patients are more sensitive to
radiation than adults, mainly because they
have more rapidly dividing cells and longer
life expectancy [810]. Therefore, radiation
dose reduction in diagnostic imaging proce-
dures, especially CT, has become a major fo-
cus for pediatric radiologists. Image Gently,
an alliance of multiple medical organizations
with a focus on radiation reduction, proposes
reduced tube currentexposure time product
and kilovoltage for pediatric CT [1113], but
there have yet to be standardized guidelines
offered for combined kilovoltage and tube
currenttime product reduction.
Keywords: abdominal circumference, contrast-to-noise
ratio, CT, dose reduction, pediatric imaging
DOI:10.2214/AJR.10.6153
Received November 17, 2010; accepted after revision
January 27, 2012.
OBJECTIVE. This CT study evaluates image noise and radiation dose using a modied
CT dose index phantom to approximate pediatric abdominal shape. Contrast-to-noise ratio
(CNR) and radiation dose were measured.
MATERIALS AND METHODS. The oval shape was simulated by xing 1000-mL sa-
line bags aside cylindric phantoms with variable circumferences. The doses at the center and
peripheral holes in the phantom were recorded. Measurements were obtained at 50400 mAs
and 80140 kVp. Diluted iodine contrast agent lled the center hole, and distilled water lled
the peripheral holes. CNR was dened as the difference in CT number between diluted iodine
and water divided by the standard deviation (SD) of CT number of water.
RESULTS. Dose increased linearly with increases in tube currentexposure time product
and by a power function (kVp
n
, where n = 2.643.09) for increases in kilovoltage. A range
of scanning parameters was established for each circumference from which technique opti-
mization curves were created to determine the best tube currenttime product and kilovolt-
age pairs when noise was less than 20 HU and dose was less than 2.5 cGy. CNR increased by
40% as kilovoltage was reduced from 140 to 80 kVp. A dose reduction of 70% was observed
for 140 versus 80 kVp for the same CNR.
CONCLUSION. Because pediatric patients of the same age and weight come in all shapes
and sizes, abdominal circumference is a useful clinical parameter on which to base CT scan
techniques controlling radiation outputnamely kilovoltage and tube currenttime product.
Low-kilovoltage techniques for patients with small circumference show better iodine CNR.
Dong et al.
Use of Abdominal Phantom to Determine Pediatric CT Param-
eters
Pediatric Imaging
Original Research
AJR:199, September 2012 671
Use of Abdominal Phantom to Determine Pediatric CT Parameters
iodine contrast, contrast-to-noise ratio (CNR)
is a more relevant image quality metric, which
can be used for size-dependent technique op-
timization [16].
The aim of our work is to explore the re-
lationship between kilovoltage, tube current
time product, noise, and dose and to identi-
fy kilovoltage and tube currenttime product
pairs in a bounded system constrained by a
maximum permissible image noise concur-
rently with maximum permissible dose. These
data were collected using a set of ovoid phan-
toms whose circumferences mimic the clinical
range from newborn to age 18 years.
Materials and Methods
In this in vitro prospective study, right cylin-
dric polymethylmethacrylate (PMMA) CT dose in-
dex (CTDI) phantoms (model 764244156, Fluke
Biomedical) with diameters of 10, 16, and 32 cm
(or 30-, 50-, and 100-cm circumference) were used
to approximate the diameters of an infant, child,
and adolescent at the waist. The diameter data rang-
es indicated a 2-year-olds range of 1416 cm, a
4-year-olds range of 1519 cm, and an 18-year-
olds range of 2233 cm [17]. To mimic the shape
of the pediatric patients abdomen, a 1000-mL sa-
line bag was xed to each side of the CTDI phan-
toms (Fig. 1). The AC of each right-circular phan-
tom plus the addition of the saline bags, one to each
side, was determined by measuring the circum-
ference of each conguration, which included the
PMMA phantoms and the saline bags, with a ex-
ible measuring tape. Two saline bags contributed an
additional 12 cm to the original AC of three CTDI
phantoms. However, for simplicity, each oval phan-
tom was still labeled as 30, 50, and 100 cm. The
combined AC was in the nal technique optimiza-
tion curve, as explained in the Results section.
The CTDI phantoms had multiple holes cut at
the center and also at the 12-, 3-, 6-, and 9-oclock
locations approximately 1 cm from the phantom
edge. These holes were mainly used for placing the
dosimetry probe.
Using a National Institute of Standards and
Technologycalibrated ion chamber and electrom-
eter (Victoreen 660, Fluke Biomedicals), radiation
doses at the center and 12-oclock locations of the
CTDI phantoms were recorded. Accuracy and re-
producibility were veried according to vendor
specications. Measurements were obtained for
varying tube currenttime products (50, 100, 200,
300, and 400 mAs) and kilovoltages (80, 100, 120,
and 140 kVp) for each oval phantom. All mea-
surements were acquired on a single 16-MDCT
scanner (Sensation-16, Siemens Healthcare). Scan
control parameters were pitch of 0.95, scan FOV
of 50 cm, image thickness of 3 mm, image-to-im-
age spacing of 3 mm, a reconstruction kernel of
B31f, and display FOVs of 40, 24, and 18 cm for
the 32-, 16-, and 10-cm-diameter phantoms, re-
spectively. There was no automated tube current
control used in either the rotational direction or in
the z-direction for any scans or any phantoms used
in this work; we used xed tube current techniques.
The clinically acceptable noise value used in
this work was determined by sampling images
deemed to be clinically acceptable for a profes-
sional dictation by experienced board-certied
pediatric subspecialty radiologists. These images
were deidentied pediatric abdominal cases ac-
quired at similar kilovoltage, tube currenttime
product, and other scan parameters used in the
phantom study portion of this work. In these im-
ages, we chose to measure noise in the liver. This
was done for several reasons: it is a clinically sig-
nicant organ prone to many diseases; it is large
enough to get a reasonable region of interest (ROI)
circle onto; and it is reasonably homogeneous.
Of interest was the magnitude of noise content
in abdominal soft organs; this was determined
to be approximately 20 HU. When our phantom
was scanned in the clinical technique range that
yielded doses of 2.02.5 cGy (CTDI), the mag-
nitude of the noise was approximately eight times
greater than that found in clinical cases. This high
noise value is the result of three causes: rst, the
saline bags create long attenuation paths; second,
the PMMA has a higher attenuation coefcient
than that of water; and third, our geometry creat-
ed structural noise from sharp angles and uid-air
interfaces. In order for us to scale how changing
exposure (coulombs per kilogram) changed the
amount of noise in an image, we chose to create a
scaling factor that, when applied to phantom data,
would yield an approximate noise value in clini-
cal images. For our setup, including materials and
techniques, this factor was about eight, meaning
that our phantom images had about eight times as
much noise as a similarly acquired clinical image.
Noise was recorded as the SD of pixel values at
each ROI within multiple images from each scan
with an ROI size of 100 mm
2
. The ROI was placed
at the same location (close to the center of the image)
using image-processing software (MATLAB ver-
sion 7.9.0, MathWorks). We used the noise values
from a random sample of deidentied clinical imag-
es with acceptable image quality from pediatric CT
scans acquired at the same kilovoltage, effective tube
currenttime product (dened as [tube current in mil-
liamperes exposure time in seconds] / pitch fac-
tor), and other scan parameters. Because of the high
noise introduced by the saline bags, a noise calibra-
tion factor of 8.0 was determined and applied to the
phantom data to create equivalent clinical data. Ac-
ceptable image quality was determined by a pediat-
ric radiologist as the maximal tolerated image noise.
A
Fig. 1CT dose index (CTDI) phantom.
A, Saline bags are afxed to each side of cylindrical CTDI phantom (16-cm-diameter phantom shown) to mimic oval shape of pediatric abdomen.
B, CT image of 16-cm-diameter CTDI phantom with saline bags.
B
672 AJR:199, September 2012
Dong et al.
For the CNR measurement, iodinated contrast
agent (iopromide; Ultravist 300, Bayer Schering
Pharma) was diluted to match the clinically rel-
evant CT number of approximately 300 HU, and
this diluted contrast agent was injected into the
center holes of the oval phantoms. Distilled water
was also injected into one of the peripheral holes
as a reference. The image contrast was measured
on the basis of the difference of mean CT num-
ber between iodine contrast agent and water. The
CNR was computed by taking the difference be-
tween the mean CT number of diluted iodine and
water and dividing this difference by the noise
measured in water.
Reproducibility of the dose measurements was
assessed by collecting 10 sets of dose data at the
center of the 32-cm diameter phantom scanned
at 200 mAs and 120 kVp at a 12-mm collimation
width. The noise measurement error was also as-
sessed by placing the ROI near the center of the
phantom images (away from the dose probe) using
the MATLAB-based image processing tool. Im-
age noise was given as the SD of the pixel values
in the ROI.
A one-tail paired Student t test was used as the
statistical method to determine whether the dif-
ference in the measurements from two separate
sets of the scan parameters was signicant, where
p less than 0.05 was considered statistically sig-
nicant. A linear regression method was used to
curve-t the measured dose versus tube current
time product. For the measured dose at different
kilovoltages, a 10-based logarithm was taken be-
fore the linear regression was applied, because the
relationship between the dose and kilovoltage was
better described by a power function.
Results
For all phantom circumferences, doses in-
creased with increases in tube currenttime
product (Fig. 2A) and kilovoltage (Fig. 2B).
The relationship was linear for tube cur-
renttime product but increased by approxi-
mately the nth power of kilovoltage, with n
of 2.643.09 (Table 1). As dose increased,
noise decreased for all circumferences with
a sharp decrease at lower doses and a pla-
teau effect at higher doses (Fig. 3). For ex-
ample, if the dose increased from 2 to 4 cGy,
noise decreased by 5.0 HU, whereas when
the dose increased from 6 to 8 cGy, noise de-
creased by only 1.4 HU. While holding ki-
lovoltage constant at 120 kVp and increas-
ing tube currenttime product, the absorbed
dose increased as circumference decreased
(Fig. 2A). The increase in dose at two small-
er circumference phantoms (30- and 50-cm
AC) was signicant compared with the larg-
er 100-cm AC phantom, and it appeared not
to be linearly proportional to the decrease in
AC. As kilovoltage increased, signicant in-
creases in the absorbed dose for the same AC
resulted (p < 0.03, based on one-tail paired
Student t test) (Fig. 4). The combination of
dose reduction from tube currenttime prod-
uct and kilovoltage is tabulated in Table 2.
Using a noise threshold of 20 HU and a dose
threshold of 2.5 cGy (dose measured with
16-cm diameter CTDI phantom with 50-cm
equivalent AC), a range of imaging param-
eters was established for each AC (Table 2).
From these data, a technique optimization
curve was established to determine optimal
tube currenttime product and kilovoltage
pairs for equivalent AC (Fig. 5). Here, be-
cause of the saline bag on each side, 12 cm
has been added to the equivalent AC. Esti-
mated optimal parameters were as follows:
for the small patients (equivalent AC, 4060
cm), 80 kVp and less than 120 mAs; for the
medium pediatric patients (equivalent AC,
6080 cm), 100 kVp and less than 65 mAs;
and for larger size pediatric patients (equiva-
lent AC, 80100 cm), 120 kVp and less than
65 mAs. For large pediatric patients (equiv-
alent AC, > 110 cm), either 140 kVp can be
used to increase the x-ray penetration with
less than 50 mAs, or 120 kVp with tube cur-
renttime product increased to 70 mAs.
Figure 6 shows how CNR increases with
lower kilovoltage for a constant dose lev-
el (0.65 cGy, measured with a 16-cm diam-
eter phantom). At 140 kVp, the CNR was
26; when kilovoltage decreased to 80 kVp,
the CNR increased to 46 (> 70% improve-
100-cm AC
50-cm AC
30-cm AC
0
0
5
10
15
20
30
25
100 200 300 400 500
Tube CurrentTime Product (mAs)
D
o
s
e

(
c
G
y
)
100-cm AC linear fit
50-cm AC linear fit
30-cm AC linear fit
Measured 100 cm
Measured 50 cm
Measured 30 cm
60
0
5
10
15
20
30
40
35
25
70 80 90 100 110 120 130 140 150
Tube Voltage (kVp)
D
o
s
e

(
c
G
y
)
Power-fit 100 cm
Power-fit 50 cm
Power-fit 30 cm
A
Fig. 2Absorbed dose in center of CT dose index phantom.
A, Graph shows center absorbed dose versus tube currenttime product for three abdominal circumferences (ACs) at 10 kVp.
B, Graph shows center absorbed dose versus kilovoltage for 50-cm AC at 400 mAs.
B
TABLE 1: Linear Regression Coefficients (a, b, and R
2
) for Dose Versus
Effective Tube CurrentTime Product (mAs) and Linear Regression
Coefficients (a, n, and R
2
) for Dose Versus Kilovoltage (kVp)
Abdominal
Circumference (cm)
Dose = (a mAs) + b Dose = a (kVp)
n
a b R
2
a n R
2
30 0.0627 0.269 0.996 8.19 10
5
2.64 0.986
50 0.0558 0.331 0.997 3.17 10
5
2.81 0.984
100 0.0131 0.0527 0.997 1.92 10
6
3.09 0.980
AJR:199, September 2012 673
Use of Abdominal Phantom to Determine Pediatric CT Parameters
ment). Figure 7 shows that the dose can be re-
duced while maintaining the same level of the
CNR (i.e., 40) if lower kilovoltage can be used.
For example, at 80 kVp, 0.5 cGy was needed
to achieve a CNR of 40. However, at 140 kVp,
1.75 cGy was needed to achieve the same level
of CNRa 250% dose increase.
Repeatability measures for exposure data
gathered yielded approximately 1.5% repro-
ducibility at all phantom sizes, kilovoltages,
and tube currenttime products. This high
level of reproducibility resulted in statistical-
ly signicant differences (p < 0.05) in doses
for data shown in Figures 2, 4, and 5.
The noise measurement error was as-
sessed by placing the ROI near the center
of the phantom images (away from the dose
probe). The statistical calculations resulted
in mean ( SD) noise of 35.23 2.25 HU.
The SD was within 10% of the mean.
The center dose versus tube currenttime
product and kilovoltage was curve-tted to
the following functions:
dose = a mAs + b (1),
dose = a (kVp)
n
(2).
Equation (1) shows that the dose has a lin-
ear relationship with tube currenttime prod-
uct. Both a and b are the tting parameters.
Equation (2) indicates that the dose varies with
kilovoltage by an nth power relationship. The
coefcients were generated by a regression
routine (LINEST in Excel, Microsoft) (Table
1). The correlation coefcient (R
2
) is great-
er than 0.98 for all regressions, indicating it
was reasonable to use the proposed functions
in Equations (1) and (2). The exponent (n) in
the power-tting curve for the dose versus ki-
lovoltage was 2.64 for the smallest phantom
and 3.09 for the largest phantom, indicating
the beam-hardening effect was stronger with
the large phantom (Table 1).
Discussion
Radiation dose reduction is a key focus in
the medical and lay press [3, 4, 16]. The vari-
able standards across the country prompted
formation of Image Gently, a consortium of
experts in diagnostic imaging; the groups
have been very successful in disseminating
information and solutions for dose reduc-
tion to medical professionals and laypeople
worldwide [1113].
Pediatric radiology has led the way in cre-
ating weight-based charts for tube current
time product reduction for CT [18]. Some use
age as their dening parameter [19]; however,
Kleinman et al. [20] caution against using age,
because predicted patient size does not corre-
late well with age and could lead to under- or
overdosing at CT. In addition, because x-ray
attenuation is related to tissue thickness, two
children with the same weight or age could
differ signicantly in height, in which case
their abdominal thicknesses could be vastly
different. According to the current weight-
based standards, children identical in weight
and age would undergo CT of the abdomen
using the same dosing parameters; without
considering the impact height may have on
abdominal thickness, the radiation dose could
be too high for one patient and too low for an-
other. Regardless of the parameter used, most
centers performing pediatric CT now use safe
practices for radiation dose in CT [21].
The data collected that depict how dose
changes as tube currenttime product is in-
creased exhibited the expected linear behav-
ior for a xed phantom size and ion chamber
position. The data taken using the 30-cm cir-
cumference phantom at 400 mAs were twice
that of data taken at 200 mAs. This trend
also held for the 50- and 100-cm circumfer-
ence phantoms. Regarding how dose changes
at a given tube currenttime product while
phantom diameter varies, it can be seen that
the highest dose is recorded with the small-
est size phantom. This is so because there is
a high proportion of primary photons and
scattered photons reaching the ion chamber
for this size phantom. As the phantom diam-
eter increases, the amount of primary pho-
tons reaching the central axis of the phantom
starts to decrease. This decrease in photon
ux is not completely offset by increased
scatter because of increased phantom vol-
ume irradiated. The number of primary pho-
tons reaching the central axis is smallest in
the largest phantom. The large numbers of
scattered photons created in the rst few cen-
timeters of the largest phantom do not fully
replace or completely penetrate down to the
central axis, given that they are scattered in
4 steradians and may, in fact, suffer mul-
tiple Compton events; this results in a lower
dose at the central axis compared with the
peripheral location. This pattern of dose dis-
tribution as a function of size may be impor-
tant clinically because it implies that small-
er patients suffer proportionately higher core
doses per tube currenttime product used
than do larger patients. This pattern may
also have long-term consequences related to
radiogenic cancers in subdiaphragmatic or-
gans such as the liver pancreas, and kidneys.
There have been many reports touting the
benets of lowering kilovoltage, as well as
tube currenttime product, for CT using either
0 5 10 20 15
5
10
15
20
25
30
35
Dose (cGy)
N
o
i
s
e

(
H
U
)
0
0
5
10
15
20
30
45
35
40
25
100 200 300 400 500
Tube CurrentTime Product (mAs)
D
o
s
e

(
c
G
y
)
80 kVp 30-cm AC
100 kVp 30-cm AC
120 kVp 30-cm AC
140 kVp 30-cm AC
Fig. 3Graph shows that center noise versus center dose for 50-cm abdominal
circumference phantom followed power function tting of noise, calculated as
25.51 dose
0.518
(R
2
= 0.984).
Fig. 4Graph shows absorbed dose versus tube currenttime product for 30-cm
abdominal circumference (AC) phantom at variable kilovoltage.
674 AJR:199, September 2012
Dong et al.
weight or age as a guide [16, 2227]. The ben-
ets include further dose reduction, improved
image contrast, and better iodinated contrast
agent visualization, the last related to imaging
closer to the inherent K-edge of iodine (33 keV)
[15, 27]. There is sufcient evidence suggest-
ing that combining kilovoltage and tube cur-
renttime product reduction is both wise and
achievable in pediatric CT [16], but there are
currently no clinical guidelines for this. The
impetus for the preliminary study [14] was to
dene the best clinical parameters to achieve
radiation dose reduction in CT (using both ki-
lovoltage and tube currenttime product) that
would render diagnostic quality images with
acceptable noise. Using right cylindric phan-
toms simulating the infant, child, and adoles-
cent abdomen, image noise and dose were most
closely matched to AC, and from this, dosing
parameters were created to achieve the balance
between image noise and patient dose. One lim-
itation of the study was the use of right cylin-
dric phantoms, where most pediatric abdomens
are not true cylinders but more often ovoid in
shape, especially when a child gets older. In ad-
dition, the study did not assess the CNR effects
of changes in kilovoltage.
Boone et al. [16] conducted a similar phan-
tom study on an MDCT system, looking at the
effects on dose and noise in varying both kilo-
voltage (80140 kVp) and tube currenttime
product (30300 mAs) according to object di-
ameter. The effects of parameter manipulation
on CNR, as it applied to abdominal and head
CT, were reported, but unfortunately, the ta-
bles included dose reduction factors that ap-
plied only to tube currenttime product and
were not easily transferable to the clinical are-
na. In addition, that studys results were con-
ned to only one vendor platform.
The current study was designed to better
simulate the pediatric abdominal shape be-
yond infancy, to reproduce the essence of ki-
lovoltage and tube currenttime product tech-
nique charts from the prior study [14], and to
evaluate the CNR as an image quality met-
ric, in addition to noise. As in the prior study,
a noise threshold of 20 HU and dose thresh-
old of 2.5 cGy were used to establish a range
of parameters to generate technique optimiza-
tion curves. Adaptation of the phantom to an
oval shape led to only minor adjustments in
recommended kilovoltage and tube current
time product pairs for each patient size for the
same dose and noise range. The most strik-
ing ndings were related to the impact on the
CNR when a low-kilovoltage technique was
used. For example, lowering the kilovoltage
T
A
B
L
E

2
:

A
b
s
o
r
b
e
d

D
o
s
e

a
n
d

N
o
i
s
e

V
a
l
u
e
s

a
t

V
a
r
y
i
n
g

T
u
b
e

C
u
r
r
e
n
t

T
i
m
e

P
r
o
d
u
c
t
,

K
i
l
o
v
o
l
t
a
g
e
,

a
n
d

A
b
d
o
m
i
n
a
l

C
i
r
c
u
m
f
e
r
e
n
c
e

(
A
C
)

S
e
t
t
i
n
g
s
A
C
,

T
u
b
e

C
u
r
r
e
n
t

T
i
m
e

P
r
o
d
u
c
t
8
0

k
V
p
1
0
0

k
V
p
1
2
0

k
V
p
1
4
0

k
V
p
C
T
D
I

(
c
G
y
)
O
v
e
r
a
l
l

N
o
i
s
e

(
H
U
)
L
i
v
e
r

N
o
i
s
e

(
H
U
)
C
T
D
I

(
c
G
y
)
O
v
e
r
a
l
l

N
o
i
s
e

(
H
U
)
L
i
v
e
r

N
o
i
s
e

(
H
U
)
C
T
D
I

(
c
G
y
)
O
v
e
r
a
l
l

N
o
i
s
e

(
H
U
)
L
i
v
e
r

N
o
i
s
e

(
H
U
)
C
T
D
I

(
c
G
y
)
O
v
e
r
a
l
l

N
o
i
s
e

(
H
U
)
L
i
v
e
r

N
o
i
s
e

(
H
U
)
1
0
0

c
m
5
0

m
A
s
0
.
4
3
1
4
.
0
3
9
.
3
1
.
0
1
6
6
.
3
2
0
.
8
1
.
4
9
9
.
9
1
2
.
5
2
.
1
7
4
.
7
9
.
3
1
0
0

m
A
s
0
.
8
1
8
1
.
6
2
2
.
7
1
.
8
1
1
1
.
0
1
3
.
9
2
.
6
6
1
.
3
7
.
7
3
.
7
4
5
.
3
5
.
7
2
0
0

m
A
s
1
.
7
1
4
1
.
2
1
7
.
7
3
.
6
6
1
.
0
7
.
6
5
.
1
3
8
.
0
4
.
8
7
.
6
3
1
.
8
4
.
0
3
0
0

m
A
s
2
.
5
1
2
0
.
2
1
5
.
0
5
.
2
4
6
.
5
5
.
8
7
.
7
3
1
.
4
3
.
9
1
1
.
7
2
3
.
7
3
.
0
4
0
0

m
A
s
3
.
6
9
8
.
1
1
2
.
3
7
.
6
3
5
.
9
4
.
5
1
1
.
3
2
7
.
0
3
.
4
1
5
.
5
2
0
.
2
2
.
5
5
0

c
m
5
0

m
A
s
1
.
0
2
9
.
7
3
.
7
2
.
1
1
8
.
4
2
.
3
3
.
0
1
3
.
8
1
.
7
4
.
6
1
2
.
0
1
.
5
1
0
0

m
A
s
2
.
1
1
8
.
7
2
.
3
4
.
2
1
3
.
3
1
.
7
5
.
7
1
0
.
4
1
.
3
8
.
3
8
.
7
1
.
1
2
0
0

m
A
s
3
.
7
1
4
.
4
1
.
8
7
.
7
8
.
8
1
.
1
1
1
.
5
7
.
9
1
.
0
1
6
.
0
6
.
1
0
.
8
3
0
0

m
A
s
5
.
7
1
2
.
1
1
.
5
1
2
.
3
7
.
3
0
.
9
1
8
.
4
6
.
1
0
.
8
2
5
.
4
5
.
3
0
.
7
4
0
0

m
A
s
8
.
0
9
.
0
1
.
1
1
6
.
4
6
.
9
0
.
9
2
4
.
4
6
.
0
0
.
8
3
5
.
5
4
.
7
0
.
6
3
0

c
m
5
0

m
A
s
1
.
1
1
6
.
3
2
.
0
2
.
3
1
0
.
3
1
.
3
3
.
5
8
.
9
1
.
1
4
.
8
8
.
3
1
.
0
1
0
0

m
A
s
2
.
2
1
0
.
8
1
.
4
4
.
4
8
.
3
1
.
0
6
.
5
7
.
0
0
.
9
8
.
9
7
.
0
0
.
9
2
0
0

m
A
s
4
.
3
8
.
4
1
.
1
8
.
9
6
.
8
0
.
9
1
2
.
4
6
.
9
0
.
9
1
7
.
5
5
.
7
0
.
7
3
0
0

m
A
s
6
.
5
8
.
1
1
.
0
1
3
.
2
6
.
1
0
.
8
1
8
.
8
5
.
3
0
.
7
2
8
.
3
4
.
4
0
.
6
4
0
0

m
A
s
8
.
8
8
.
1
1
.
0
1
7
.
1
5
.
4
0
.
7
2
6
.
5
5
.
2
0
.
7
3
9
.
1
4
.
4
0
.
6
N
o
t
e

O
p
t
i
m
a
l

v
a
l
u
e
s

w
e
r
e

c
h
o
s
e
n

o
n

t
h
e

b
a
s
i
s

o
f

o
p
t
i
m
a
l

n
o
i
s
e

(

2
0

H
U
)

a
n
d

d
o
s
e

(

2

c
G
y
)
,

s
h
o
w
n

i
n

b
o
l
d

t
y
p
e
.

B
o
r
d
e
r
l
i
n
e

n
o
i
s
e

a
n
d

d
o
s
e

l
e
v
e
l
s

(
2

2
.
5

c
G
y
)

a
r
e

s
h
o
w
n

i
n

i
t
a
l
i
c

t
y
p
e
.

C
T
D
I

=

C
T

d
o
s
e

i
n
d
e
x
.
AJR:199, September 2012 675
Use of Abdominal Phantom to Determine Pediatric CT Parameters
from 140 kVp (used in the largest patients) to
80 kVp (used in infants and small children)
resulted in a 40% increase in CNR. On the
other hand, a dose reduction of 70% can be
achieved from 140 to 80 kVp while still main-
taining a desired CNR. The technique optimi-
zation tables include tube currenttime prod-
uct and kilovoltage recommendations based
on AC for all children.
Efforts to decrease kilovoltage have im-
portant implications in pediatric abdominal
CT in that intrinsic tissue contrast rises with
a decrease in kilovoltage, leading to superi-
or ability to distinguish between two tissues
with similar attenuation properties. In addi-
tion, because the K-edge of iodine is 33 keV,
lower-kilovoltage imaging lies closer to the
iodine K-edge, prompting discussion of po-
tentially lowering the volume of iodinated
contrast agent needed for an individual pe-
diatric CT scan [15]. In a study of 40 adult
patients randomized to one of two low-kilo-
voltage abdominal CT protocols, Nakayama
et al. [27] showed that one could reduce the
iodinated IV contrast bolus by at least 20%
at 90 kVp and achieve solid organ and vas-
cular enhancement that was superior to that
achieved at 120 kVp. In a similar study de-
sign, Marin et al. [28] showed improved sol-
id organ and vascular enhancement in CT for
pancreatic carcinoma using a low kilovoltage
and higher tube currenttime product tech-
nique, while lowering the radiation dose by
71% and improving the CNR by 37%. The
double dose reduction achieved in reducing
kilovoltage and IV contrast agent is an ap-
pealing concept for pediatric CT.
Improvements in CNR are related to both
increases in contrast and decreases in noise.
Improved contrast provides the greatest ben-
et in detecting small low-contrast structures,
such as liver lesions in fungal septicemia. One
must bear in mind that, by lowering the ki-
lovoltage, improved lesion detection through
improved contrast might be outweighed by
the negative effects of decreased dose (and
increased noise) such that kilovoltage reduc-
tion would have to be paired with a modest
increase in tube currenttime product. In the
case of small-vessel detection with CT angi-
ography, a reduction in kilovoltage provides
its greatest benet in dose reduction, not low-
contrast lesion detection, because contrast is
already maximized through IV iodinated con-
trast agent administration. In these cases, tube
currenttime product may be further reduced
without signicant effect on image quality.
A limitation of the current work was the
inability to adequately address how body
shape changes with age and how this change
can affect girth and, therefore, CT techniques
based on girth. We recognize that, as chil-
dren grow in height and weight from infancy
to adulthood, the patterns of growth can be
nonuniform [17]. Some children will add sig-
nicantly to their weight while increasing in
height marginally for a time. This pattern of
growth will likely add to girth. Other children
may grow rapidly in height and marginally in
weight for periods in their growththis pat-
tern, while not affecting girth much, may af-
fect the oval ratio of the childs cross-section.
In each case, girth and cross-sectional shape
can change dramatically as age increases. In
the tall lean child, signicant oval shape may
be present, greater than that studied in this ar-
ticle. In the child who adds weight without a
signicant increase in height, a more circu-
lar cross-section may be present; this was ad-
dressed in our rst article [14]. More work is
needed concerning how different oval ratios
at xed girth and also xed oval ratio at vari-
ous girths affect CT techniques to maintain
adequate noise at reasonable radiation doses.
Another limitation was that one vendors CT
platform was used to collect data. A more com-
plete sampling of various manufacturers and
detector geometries with a vendor product list
0
0.0
0.5
1.0
1.5
2.0
3.0
3.5
5.0
4.5
4.0
2.5
60 50 40 30 20 10 80 70 90 100 110 120
AC (cm)
D
o
s
e

(
c
G
y
)
80 kVp (120 mAs)
100 kVp (65 mAs)
120 kVp (65 mAs)
140 kVp (50 mAs)
Tube Voltage (kVp)
80
10
20
30
40
50
100 120 140
C
N
R
Tube Voltage (kVp)
80
0.0
0.4
0.8
1.2
1.6
2.0
100 120 140
D
o
s
e

(
c
G
y
)
Fig. 5Technique optimization curve shows optimal kilovoltage and tube
currenttime product dose pairs (shaded area) for absorbed dose and abdominal
circumference (AC).
Fig. 6Contrast-to-noise ratio (CNR) was reduced with higher kilovoltage while
keeping dose at 0.65 cGy (measured with 16-cm-diameter CT dose index phantom).
Fig. 7Radiation dose
was measured at different
kilovoltage to maintain
contrast-to-noise ratio at
40 (dose measured with
16-cm-diameter CT dose
index phantom).
676 AJR:199, September 2012
Dong et al.
is warranted. This is especially true regarding
row number; as detector array width increases,
the effects of scattered radiation into adjacent
rows may have an effect on signal-to-noise ra-
tio, thereby changing our results. No attempt
was made in this article to maximize or even
characterize CNR. In large measure, this was
the result of limitations in our phantom and lack
of funding for this work. Future CNR work in
conjunction with other limitations, as already
discussed, is planned. We did not explore in
this work how our technique recommendations
would affect spatial resolution. The phantom
we used did not contain tools adequate to in-
vestigate this question. The issue of selecting
an upper level of tolerable noise in a clinical
setting is not so much a limitation as it is an
unavoidable reality; it is, however, somewhat
subjective. In a more exhaustive setting, other
radiologists would be queried as to what upper
level of noise they would accept. The question
of what upper level of noise is acceptable in CT
images where noise is a limiting factor is under
way and will be reported separately. Finally, a
limitation of this work was to use the liver as
a proxy for all soft abdominal or retroperitone-
al organs. We chose the liver for the following
reasons: it is host to many diseases; it is large
enough (even in small patients) to get an ade-
quate ROI circle from which noise can be as-
certained; and it is reasonably homogeneous. It
is planned that in clinical studies more organs
of interest will be included.
Regarding automated dose reduction tech-
niques available on commercial CT scanners,
we acknowledge their presence, use, and po-
tential for dose reduction in certain situations.
The goal of our work was to study the basic
effect that kilovoltage has on dose in the con-
text of adequate image quality, as determined
by a maximum permissible noise value. Our
intent was to use kilovoltage as the primary
driver for dose reduction and to use tube cur-
renttime product as the ne-tuning mecha-
nism within a kilovoltage selection (and girth
range). In later work, we plan to look at in-
plane and z-axis modulation once a kilovolt-
age stop has been selected according to girth.
Many centers performing pediatric CT are
now using safe practices for radiation dose re-
duction using technique charts for tube cur-
renttime product reduction based on patient
weight or age. We have shown that, because
pediatric patients of the same age and weight
come in all shapes and sizes, AC is a useful
clinical parameter on which to base CT scan
techniques controlling radiation output, name-
ly kilovoltage and tube currenttime product.
References
1. Amis ES, Butler PF, Applegate KE, et al. American
College of Radiology white paper on radiation dose
in medicine. J Am Coll Radiol 2007; 4:272284
2. [No authors listed]. Looking back on the millen-
nium in medicine. N Engl J Med 2000; 342:4249
3. Brenner DJ. Should we be concerned about the
rapid increase in CT usage? Rev Environ Health
2010; 25:6368
4. Fayngersh V, Passero M. Estimating radiation risk
from computed tomography. Lung 2009; 187:143148
5. Schauer DA, Linton OW. National Council on Ra-
diation Protection and Measurements report
shows substantial medical exposure increase. Ra-
diology 2009; 253:293296
6. Rabin RC. With rise in radiation exposure, ex-
perts urge caution on tests. The New York Times,
June 19, 2007
7. Mozes A. Certain tests in ERs raise cancer risk for
some. The Washington Post, May 29, 2008
8. Brenner DJ, Elliston C, Hall E, Berdon W. Esti-
mated risks of radiation-induced fetal cancer from
pediatric CT. AJR 2001; 176:289296
9. National Research Council, Committee on the Bio-
logical Effects of Ionizing Radiations. Health effects
of exposure to low levels of ionizing radiation: BEIR
V. Washington, DC; National Academy Press, 1990
10. National Research Council, Committee on the
Biological Effects of Ionizing Radiations. Health
effects of exposure to low levels of ionizing radia-
tion: BEIR VII. Washington, DC; National Acad-
emy Press, 2007
11. Goske MJ, Applegate KE, Boylan J, et al. The Im-
age Gently campaign: increasing CT radiation dose
awareness through a national education and aware-
ness program. Pediatr Radiol 2008; 38:265269
12. Strauss KJ, Goske MJ, Frush DP, Butler PF, Mor-
rison G. Image Gently Vendor Summit: working
together for better estimates of pediatric radiation
dose from CT. AJR 2009; 192:11691175
13. Strauss KJ, Goske MJ, Kaste SC, et al. Image
Gently: ten steps you can take to optimize image
quality and lower CT dose for pediatric patients.
AJR 2010; 194:868873
14. Reid J, Gamberoni J, Dong F, Davros W. Optimi-
zation of kVp and mAs for pediatric low-dose
simulated abdominal CT: is it best to base param-
eter selection on object circumference? AJR 2010;
195:10151020
15. Kalva SP, Sahani DV, Hahn PF, Saini S. Using the
K-edge to improve contrast conspicuity and to
lower radiation dose with a 16-MDCT: a phantom
and human study. J Comput Assist Tomogr 2006;
30:391397
16. Boone JM, Geraghty EM, Seibert JA, Wootton-
Gorges SL. Dose reduction in pediatric CT: a ra-
tional approach. Radiology 2003; 228:352360
17. Fernndez JR, Redden DT, Pietrobelli A, Allison
DB. Waist circumference percentiles in nationally
representative samples of African-American, Eu-
ropean-American and Mexican-American children
and adolescents. J Pediatr 2004; 145:439444
18. Frush DP, Soden B, Frush KS, Lowry C. Improved
pediatric multidetector body CT using a size-based
color-coded format. AJR 2002; 178:721726
19. Thomas KE, Wang B. Age-specic effective dos-
es for pediatric MSCT examinations at a large
childrens hospital using DLP conversion coef-
cients: a simple estimation method. Pediatr Ra-
diol 2008; 38:645656
20. Kleinman PL, Strauss KJ, Zurakowski D, et al.
Patient size measured on CT images as a function
of age at a tertiary care childrens hospital. AJR
2010; 194:16111619
21. Arch ME, Frush DP. Pediatric body MDCT: a 5-year
follow-up survey of scanning parameters used by
pediatric radiologists. AJR 2008; 191:611617
22. Kim J-E, Newman B. Evaluation of a radiation
dose reduction strategy for pediatric chest CT.
AJR 2010; 194:11881193
23. Huda W, Scalzetti EM, Levin G. Technique factors
and image quality as functions of patient weight at
abdominal CT. Radiology 2000; 217:430435
24. Szucs-Farkas Z, Kurmann L, Strautz T, et al. Pa-
tient exposure and image quality of low-dose pul-
monary computed tomography angiography:
comparison of 100- and 80-kVp protocols. Invest
Radiol 2008; 43:871876
25. Wintersperger B, Jakobs T, Herzog P, et al. Aorto-
iliac multidetector-row CT angiography with low
kV settings: improved vessel enhancement and
simultaneous reduction of radiation dose. Eur Ra-
diol 2005; 15:334341
26. Nakayama Y, Awai K, Funama Y, et al. Abdominal
CT with low tube voltage: preliminary observations
about radiation dose, contrast enhancement, image
quality, and noise. Radiology 2005; 237:945951
27. Nakayama Y, Awai K, Funama Y, et al. Lower
tube voltage reduces contrast material and radia-
tion doses on 16-MDCT aortography. AJR 2006;
187:1266 [web];W490W497
28. Marin D, Nelson RC, Barnhart H, et al. Detection of
pancreatic tumors, image quality, and radiation dose
during the pancreatic parenchymal phase: effect of a
low-tube-voltage, high-tube-current CT technique
preliminary results. Radiology 2010; 256:450459