Download Evaluation of the Effective Detective Quantum Efficiency for Various

Journal of the Korean Physical Society, Vol. 58, No. 3, March 2011, pp. 532∼537
Evaluation of the Effective Detective Quantum Efficiency for Various Tube
Voltages with Digital Radiography for Chest Imaging
Hyo-Min Cho, Hee-Joung Kim,∗ Hye-Suk Park, Dae-Hong Kim, Chang-Lae Lee, Yu-Na Choi and Seung-Wan Lee
Department of Radiological Science and Research Institute of Health Science, Yonsei University, Wonju 220-710, Korea
(Received 27 September 2010, in final form 28 January 2011)
The concept of effective detective quantum efficiency (eDQE) has recently been developed and
introduced. The purpose of this study was to use the eDQE to evaluate the effect of tube voltage
and grid for chest imaging with a direct flat-panel-based digital radiographic system. The exposure
factors that we considered were the tube potential and an anti-scatter moving grid. The incident
air kerma was adjusted for each tube potential to yield an effective dose of 34 µSv. The scatter
fraction (SF), transmission fraction (TF), effective modulation transfer function (eMTF), and effective normalized noise power spectrum (eNNPS) were measured using a phantom that simulate
the attenuation and scatter properties of the human chest. Our results showed that the resolution
properties measured by using eMTFs were independent of the tube potential regardless of whether
or not a grid was used. The noise properties through the phantom were greater with the use of
an anti-scatter grid than without while the eDQE was largest at a lower tube potential, which we
tested without a grid. In conclusion, we measured a high efficiency at a low tube potential without
an anti-scatter grid for digital radiography (DR) in chest imaging. The eDQE reflected the real
exam environment better than the general DQE because it better reflected the effect of the total
X-ray system.
PACS numbers: 87.57.Ce
Keywords: eDQE, eMTF, eNNPS, Digital radiography, Chest imaging
DOI: 10.3938/jkps.58.532
timizing the imaging system’s efficiency because it considers both the sensitivity of the detector and the signal
to noise ratio (SNR) efficiency of the X-ray system, including realistic factors such as the scatter and the transmission fraction. The possibility of using effective noise
equivalent quanta (eNEQ) for system optimization was
suggested in previous studies, which also showed that low
tube potential without a static grid was the most suitable exposure condition for computed radiography (CR)
in the chest [3]. The efficiency of low tube potentials for
chest imaging is also reported based on the receiver operating characteristic method [5]. The aim of this study
was to measure the eDQEs for various tube voltages,
with and without the use of an anti-scatter moving grid,
by using a phantom simulating the human chest.
I. INTRODUCTION
The detective quantum efficiency (DQE) is widely used
as a standard to characterize all kinds of X-ray detectors
[1]. In the past few years, DQE evaluations have been
reported to measure the diagnostic utility and the clinical availability of many types of detectors. However, the
conditions used in estimating the DQE are far from real
clinical conditions because the DQE does not consider
a total X-ray system. In recent years, effective DQE
(eDQE) techniques have been developed and introduced
[2, 3]. The eDQE is a measurement of a detector’s efficiency within a total X-ray system and as such considers
the effects of focal spot size, scattered radiation, magnification and scatter reduction techniques [3].
Despite the advent of numerous new imaging techniques, chest projection radiography has remained the
most frequently investigated examination in radiology
departments. Such high tube potentials as 120 kVp are
widely used in chest radiography [4]. These exposure
conditions should be reconsidered in relation to patient
dose and detector efficiency within the systems. The
eDQE is a more useful parameter than the DQE for op∗ E-mail:
II. MATERIALS AND METHODS
1. Equipment
Images were acquired using a RAD-14 generator
(VARIAN, USA) with a FDXD-1417 (DRTECH, Korea)
selenium-based direct flat-panel detector. The matrix
size of the images was 2560 × 3072, and the pixel pitch
[email protected]; Fax: +82-33-760-2562
-532-
Evaluation of the Effective Detective Quantum Efficiency for Various Tube Voltages · · · – Hyo-Min Cho et al.
-533-
Fig. 1. (Color online) (a) Average-sized chest phantom.
(b) Geometry and configuration in a diagonal view. Al =
aluminum.
was 0.139 mm. A phantom designed by the Food and
Drug Administration for use in the Nationwide Evaluation of X-ray Trends (NEXT) program was used [6, 7].
The phantom simulates the attenuation and the scatter properties of an average human chest and consists
of 25.4 × 25.4 cm2 pieces of clear acrylic and aluminum
with a 19-cm air gap that emulates the thoracic cavity.
The configuration of the phantom is detailed in Fig. 1.
The anti-scatter grid that we tested had a density of 215
line/inch, a ratio of 8:1 and a focus distance of 180 cm
and was produced by JPI (Jungwon Precision Ind. Co.,
South Korea).
2. Dosimetry
Exposure factors of 120 kV and 3.2 mAs with a focusto-detector distance of 180 cm are the general radiographic conditions used with digital radiography (DR)
for an average-sized adult chest examination in Korea
[4]. The effective dose for an average sized adult was
calculated as 34 µSv by using PCXMC version 1.5.1
(STUK, Finland) and general radiographic conditions.
The average-sized adult phantom was set up with a
height of 171 cm and a weight of 71 kg stored in the program. The calculated effective dose, 34 µSv, was taken
to be the standard dose for various exposure conditions.
For different tube voltages, 70, 80, 90, 100, 110, and
120 kVp, the entrance dose was measured and converted
into an effective dose by using each conversion factor
which was obtained as the effective dose per mGy. The
milliampere-second values for each tube voltage were adjusted until the intended effective dose was reached.
The incident air kerma was measured using a calibrated ionization chamber (Radcal 10 × 9 - 60 ion chamber/ Radcal 9095 radiation monitor, Radcal Corporation, Monrovia, USA) for six exposure conditions. The
incident air kerma of the detector position was estimated
from the phantom’s position based on an inverse square
law. The chamber was located in front of the phantom
surface without the phantom in the beam. This set up
allowed us to measure the air kerma without backscatter
from the phantom, the so-called incident air kerma.
Fig. 2. (Color online) (a) Experimental standard set-up of
the phantom with a DR detector and anti-scatter moving grid
and (b) resolution measurement set-up using a slit camera.
Fig. 3. (a) Side-view schematic diagram for measurement
of SF, (b) a beam stop array picture, and (c) an image of
beam stop array.
3. Linearity
The linearity of the detector response is a graph of the
detector signal versus the entrance exposure [8]. The flatfield images were acquired for different exposure conditions (six different tube potentials and with/without an
anti-scatter grid). The raw images were acquired using a
standard set-up (Fig. 2). The points on the characteristic
curve are the average detector signal level in the region
of interest (ROI), which measured 125 × 125 mm2 and
was positioned over the center of the phantom, versus
the averages of three exposure measurements.
4. Scatter Fraction (SF)
The SF is defined as the ratio of the scattered radiation
to the total radiation;
SF =
S
,
S+P
(1)
where S stands for pixel value in areas without contribution from primary radiation and S+P stands for pixel
values in pixels with both primary (P) and scattered radiation (S). The scatter fractions (SFs) of the systems
-534-
Journal of the Korean Physical Society, Vol. 58, No. 3, March 2011
were measured using a beam stop array device [9, 10]
composed of a 14 × 16 array of 224 6-mm-thick and
3-mm-diameter lead cylinders. At a diagnostic X-ray
energy 120 kVp, this thickness of lead is equivalent to
about 20 half-value layers. The lead cylinders are spaced
25 mm apart in a 6-mm-thick sheet of polystyrene. The
SF measurement requires three steps. First, three raw
images of the beam’s stop array device placed in front of
the phantom were taken for each measurement condition
(Fig. 3). Then, all mean pixel values behind the 16 beam
stops at the center of the image were measured as scattered radiation. Each ROI (15 × 15 pixels) eliminated
pixels bordering the beam stop edge to prevent penumbral shadow effects [10]. Finally, the average of all mean
pixel values with the same size ROI on either side of the
16 beam stops was measured as total radiation, including
both primary and scattered radiation.
lustrated in Fig. 2. Images of the phantom were obtained
at an exposure level of K0 , where K0 is the normal level
of exposure at each measurement condition. The eNPS
was determined using the four raw images to obtain at
least 4 million independent pixels in the 125 × 125 mm
ROI.
The image processing followed established steps to deduce the eNPS. The four phantom images were averaged
and subtracted from the extracted image which was one
of the four images used for averaging to remove the background trends due to the anode-heel effect and system
noise. The center area of the processed image was divided into 16 overlapping sub-ROIs 360 × 360 in size.
The two-dimensional (2D) eNNPS was calculated using
the Fourier transform and was normalized by dividing
by the square of the mean large area signal. The onedimensional (1D) eNNPS was obtained from the horizontal axis by using the four rows above and below, excluding the axis row [12].
5. Transmission Fraction (TF)
The transmission fraction was calculated using the ratio of the air kerma with and without the phantom. The
X-ray beam was collimated to be slightly larger than the
ionization chamber size. The air kerma was measured
with and without the phantom between the X-ray tube
and the ionization chamber for each exposure condition.
6.
(eMTF)
Effective Modulation Transfer Function
The resolution properties of the system were measured
using the slit method. A schematic diagram is illustrated
in Fig. 2. The phantom was positioned just adjacent
to the detector’s cover plate. A slit camera (Nuclear
Associates, model: 07-624-1000) was placed in contact
with, and centered on, the phantom, at a 4 ∼ 5◦ angle
relative to the vertical axis of the phantom. This measurement geometry involved the effects of the focal spot
blur penumbra, magnification, scattered radiation, and
anti-scatter grid. For each of the different exposure conditions, three images of the slit were acquired at 3.2K0 ,
where K0 is the incident air kerma corrected to the detector plane and averaged. The eMTF was calculated
using the Fourier transform, from which the data were
obtained by measuring the line spread function of the
system. The details of this method have been previously
described [11].
7. Effective Normalized Noise Power Spectrum
(eNNPS)
The noise properties of the imaging system were acquired using only the phantom with the same set-up as il-
8. eDQE
The eDQE of the DR system was measured incorporating the influences of scatter, magnification, phantom
transmission fraction, anti-scatter grid, eMTF, eNNPS,
incident exposure, and number of photons per mm2 -µGy.
The q values were estimated using an X-ray spectral simulation program (SRS-78) [12]. The NEQ is defined as
the effective number of photons per unit length of the
detector [13]. The eNEQ was recalculated as the dosenormalized eNEQ (eNEQED ) by using the ratio of the
calculated effective dose for each tube potential to that of
the standard effective dose. The eDQE was determined
using the following equations:
SNR2in (u0 ) = TF · K0 · q,
SNR2out (u0 ) = eNEQ(u0 )
eMTF(u0 )2 (1 − SF)2
=
,
eNNPS(u0 )
SNR2out (u0 )
eDQE(u0 ) =
SNR2in (u0 )
eMTF(u0 )2 (1 − SF)2
,
eNNPS(u0 ) · TF · K0 · q
FOD
u0 = m · u =
u,
FDD
=
(2)
(3)
(4)
(5)
where K0 = incident air kerma corrected to the detector
plane (µGy); q = number of photons per mm2 -µGy; u
is the spatial frequency in the detector plane (mm−1 );
u’ is the spatial frequency corrected to the object plane
(mm−1 ); m = magnification; FOD = focus to object
distance (cm); FDD = focus-to-detector distance (cm)
[3]. Table 1 shows the factors used in the calculations of
the eDQE and the standard effective dose achieved using
the conversion factors for each tube potential.
Evaluation of the Effective Detective Quantum Efficiency for Various Tube Voltages · · · – Hyo-Min Cho et al.
-535-
Table 1. Factors used for the eDQE at various tube voltages.
Kilovoltage Phantom incident air Effective dose to incident air Effective dose (µSv) K0 (µGy) q (quanta µGy−1 mm−2 )
kerma (µGy)
kerma ratio (µSv/µGy)
70
320
0.103
32.9
230.4
18981
80
258
0.125
32.3
185.7
21130
90
202
0.147
29.7
145.4
22777
100
196
0.167
32.7
141
23965
110
185
0.185
34.2
133
24814
120
172
0.201
34.5
123.8
25414
Fig. 4. Scatter fraction for the phantom with and without
an anti-scatter grid.
Fig. 5. Transmission fraction for the phantom at each tube
potential.
III. RESULT
The system showed excellent linearity between the
mean pixel value and the exposure at each tube potential (R2 = 0.99). The relationships were an important
requirement of linear system analysis.
Figure 4 shows the scatter fraction with and without
an anti-scatter moving grid at each tube potential. The
scatter fraction with a grid was lower than without a grid
by approximately 10% for all tube potentials. Figure 5
shows that the transmission fractions for the phantom
increased with increasing tube potential.
Figure 6 shows the eMTF results with and without
Fig. 6. Measured eMTF (a) with and (b) without an antiscatter grid at various tube potentials.
the use of an anti-scatter moving grid at various tube
voltages. The eMTF only slightly depended on tube potential regardless whether or not a grid was used. The
variation in the resolution as a function of tube potentials was a little bigger with a grid than without a grid.
Figure 7 illustrates the eNNPS at exposure levels of
K0 for 70 and 80 kVp. The results using high exposure
conditions are not reported as they were very similar
to those using a low exposure condition. For all tube
voltages, a higher magnitude of the eNNPS was acquired
with an anti-scatter moving grid than that without a
grid.
Figure 8 shows the eNEQED with and without an antiscatter moving grid for various tube potentials. Most of
-536-
Journal of the Korean Physical Society, Vol. 58, No. 3, March 2011
Fig. 7. eNNPS with and without an anti-scatter grid for
tube voltages of (a) 70 kVp and (b) 80 kVp.
Fig. 9. eDQE measured (a) with and (b) without an antiscatter grid at different tube potentials.
the eNEQED values with an anti-scatter grid were lower
than those without a grid except for 90 and 100 kVp.
These may be affected by the gap in the NNPS values
between with and without grid being slightly smaller for
90 and 100 kVp than for other tube potentials. Low tube
potentials have many more effective photons than high
tube potentials regardless of grid use. The eNEQED data
without an anti-scatter grid indicated that 70 kVp had
the highest value below 0.25 mm−1 and that 80 kVp had
the highest value above 0.25 mm−1 .
Figure 9 shows the eDQE with and without an antiscatter moving grid at different tube potentials. Samei
et al. showed different tendency for the eDQE and the
conventional DQE in the same system [2]. The tendency
of conventional DQE will be different from the eDQE if
we evaluate the conventional DQE by considering only
the detector effect. The eDQE was higher without the
use of an anti-scatter grid for most tube potentials. Furthermore, the eDQE was larger at low tube potentials
than at high tube potentials both with and without the
use of a grid.
IV. DISCUSSION
Fig. 8. Effective dose normalized eNEQ (a) with and (b)
without a grid for various tube potentials.
The eMTF variation showed no significant differences
between with and without a moving grid. A stationary grid had an eMTF variation higher than that ob-
Evaluation of the Effective Detective Quantum Efficiency for Various Tube Voltages · · · – Hyo-Min Cho et al.
served without grid [3]. The eMTF result is influenced
by the moving grid decreasing the uncertainty caused in
the measurement by blurring [14]. The eMTF effectively
dropped close to zero above 3.5 cycles/mm. The ideal
MTF for the system is the best possible MTF as dictated by the pixel pitch and was 6.5 cycles/mm at 10%
(data not shown). Comparing the results, the eMTFs
were notably lower than ideal. These eMTF results reflect the actual resolutions that might affect chest images
obtained using this system in a clinical setting.
The eNNPS curve shape and tendency were similar
to most results of NNPS measurements obtained by using direct digital radiography, as reported in the literature [15]. The graph shows that the image with an
anti-scatter moving grid has higher noise than that obtained without grid. This is due to the lower photon
counting statics because of the grid line. The effect of
the grid structure on the resulting eNNPS was lower with
a moving grid than with a stationary grid [2]. Most digital systems use the grid to reduce scatter radiation, but
grid noise remains within the image.
Most tube voltages had higher eNEQED and eDQE
values without an anti-scatter moving grid than with
a grid. These trends were also seen in a previous report, in which they were calculated for a computed radiographic system. These results reflect the fact that the
reduction of scattered radiation in the beam does not
compensate for the loss of absorbed photons in an antiscatter grid [3]. The eDQE showed that 70 kVp had the
highest efficiency with the phantom for all frequencies,
but the eNEQED showed that 80 kVp had more effective photons than other tube potentials. These results
showed that there may be different focus of attention between the eDQE calculated for quantum efficiency and
the eNEQED calculated for effective number of photons.
V. CONCLUSIONS
This study used the eDQE to evaluate exposure conditions, including tube potentials and scatter reduction
techniques, for chest examination. The environment in
which we acquired images included scattered radiation,
magnification, and focal spot blur. Our results suggest
that the use of low tube potentials, such as 70 or
80 kVp, without an anti-scatter grid represents the
most appropriate exposure conditions for DR in chest
imaging. These results are compatible with the results
in previous reports that described the efficiency of low
tube potentials for chest imaging by using the receiver
operating characteristic (ROC) method [5]. In con-
-537-
clusion, the eDQE reflects the real exam environment
better than the general DQE because it represents by
the effects of a realistic X-ray system.
ACKNOWLEDGMENTS
This work was supported by the Basic Atomic Energy
Research Institute (BAERI) of the National Research
Foundation of Korea (KRF) funded by the Ministry of
Education, Science & Technology (MEST), Republic of
Korea (grant code: 2010-0018503).
REFERENCES
[1] U. Neitzel, S. Günther-Kohfahl, G. Borasi and E. Samei,
Med. Phys. 31, 2205 (2004).
[2] E. Samei, N. T. Ranger, A. MacKenzie, I. D. Honey,
J. T. Dobbins Iii and C. E. Ravin, Radiology 249, 926
(2008).
[3] F. Ertan, A. Mackenzie, H. J. Urbanczyk, N. T. Ranger
and E. Samei, in Proceeding Society of Photo-Optical Instrumentation Engineers (Orlando, 2009).
[4] G. H. Yu, Development of Guideline and Estimation on
the Patient Dose in Chest Radiography, Korea Food &
Drug Administration (2006).
[5] T. M. Bernhardt, U. Rapp-Bernhardt, H. Lenzen, F. W.
Roehl, S. Diederich, K. Papke, K. Ludwig and W. Heindel, Radiology 232, 693 (2004).
[6] B. J. Conway, P. F. Butler, J. E. Duff, T. R. Fewell, R. E.
Gross, R. J. Jennings, G. H. Koustenis, J. L. McCrohan,
F. G. Rueter and C. K. Showalter, Med. Phys. 11, 827
(1984).
[7] B. J. Conway, J. E. Duff, T. R. Fewell, R. J. Jennings,
L. N. Rothenberg and R. C. Fleischman, Med. Phys. 17,
448 (1990).
[8] P. R. Granfors and R. Aufrichtig, Med. Phys. 27, 1324
(2000).
[9] C. E. Floyd, Jr., J. Y. Lo, H. G. Chotas and C. E. Ravin,
Med. Phys. 18, 408 (1991).
[10] C. E. Floyd, Jr, J. A. Baker, J. Y. Lo and C. E. Ravin,
Radiology 183, 857 (1992).
[11] E. Samei, N. T. Ranger, J. T. Dobbins III and Y. Chen,
Med. Phys. 33, 1454 (2006).
[12] J. T. Dobbins III, E. Samei, N. T. Ranger and Y. Chen,
Med. Phys. 33, 1466 (2006).
[13] K. L. Boedeker, V. N. Cooper and M. F. McNitt-Gray,
Phys. Med. Biol. 52, 4027 (2007).
[14] K. A. Fetterly and B. A. Schueler, Phys. Med. Biol. 52,
4863 (2007).
[15] E. Samei and M. J. Flynn, Med. Phys. 30, 608 (2003).