Effect of radiographic techniques „kVp and mAs… on image quality
and patient doses in digital subtraction angiography
Nikolaos A. Gkanatsiosa)
Department of Radiology, University of Florida, P.O. Box 100374, Gainesville, Florida 32610
Walter Huda
Department of Radiology, SUNY Upstate Medical University, Syracuse, New York 13210
Keith R. Peters
Department of Radiology, University of Florida, P.O. Box 100374, Gainesville, Florida 32610
共Received 10 January 2002; accepted for publication 17 May 2002; published 16 July 2002兲
We investigated how varying the x-ray tube voltage and image receptor input exposure affected
image quality and patient radiation doses in interventional neuroradiologic imaging. Digital subtraction angiography 共DSA兲 images were obtained of a phantom with 1 mm diameter vessels
containing iodine at concentrations between 4.5 and 50 mg/cc. The detection threshold concentration of iodine was determined by inspecting DSA images obtained at a range of x-ray tube voltages
and input exposure levels. Surface doses were obtained from measured x-ray tube output data, and
corresponding values of energy imparted were determined using the exposure-area product incident
on the phantom. In one series of experiments, the air kerma at the image intensifier (X) was varied
between 0.44 ␮Gy per frame and 8.8 ␮Gy per frame at a constant x-ray tube voltage of 70 kVp. In
a second series of experiments, the tube voltage was varied between 50 and 100 kVp, and the mAs
adjusted to maintain a constant exposure level at the input of the image intensifier. At a constant
x-ray tube voltage, the surface dose and energy imparted were directly proportional to the input
exposure per frame used to acquire the DSA images. On our DSA system operated below 2.2 ␮Gy
per frame, the threshold iodine concentration was found to be proportional to X ⫺0.57, which is in
reasonable agreement with the theoretical prediction for a quantum noise limited imaging system.
Above 2.2 ␮Gy per frame, however, the threshold iodine concentration was proportional to X ⫺0.26,
indicating that increasing the input exposure above this value will only achieve modest improvements in image quality. At a constant image intensifier input exposure level, increasing the x-ray
tube voltage from 50 kVp to 100 kVp reduced the surface dose by a factor of 6.1, and the energy
imparted by a factor of 3.5. The detection threshold iodine concentration was found to be proportional to kV p n , where n was 2.1 at 1.1 ␮Gy per frame, and 1.6 at 3.9 ␮Gy per frame. For clinical
situations that can be modeled by a uniform phantom, reducing the x-ray tube voltage rather than
increasing the exposure level would best achieve improvements on our DSA imaging system
performance. © 2002 American Association of Physicists in Medicine. 关DOI: 10.1118/1.1493213兴
Key words: image quality, dosimetry, digital subtraction angiography, tube voltage, x-ray
techniques
INTRODUCTION
Interventional neuroradiology involves studies of the vasculature and blood kinetics of the brain by means of catherization performed with the transfemoral artery technique.1,2 In
these interventional neuroradiologic procedures, image quality is critical with neuro-vascular instruments being as small
as 90 ␮m, and even smaller vessel sizes.3 The continuous
branching of the cerebral vessels leads to a reduction of the
vessel diameter being imaged and thus the amount of iodine
present in the vessel. A higher iodine concentration may
sometimes be required to image distant vessels, especially
during the capillary phase. Iodine concentration, however, is
generally limited by direct toxicity at higher concentrations.4
An alternative to increasing the iodine concentration would
be to improve the acquired signal-to-noise ratio. Practical
ways to improve detection of small vessels during interven1643
Med. Phys. 29 „8…, August 2002
tional neuroradiologic procedures include the variation of
x-ray tube voltage5,6 and x-ray beam intensity used to produce the radiographic image.7,8
Interventional neuroradiologic procedures often require
long fluoroscopic exposure times as well as the acquisition of
a large number of radiographic images.9,10 As a result, the
amount of radiation received by the patient is a matter of
major concern. The surface 共skin兲 dose predicts the possibility of inducing deterministic injuries, and is the key dose
parameter whenever the threshold dose for the indication of
such injuries is exceeded.11,12 It is desirable to keep surface
doses below threshold doses for the induction of deterministic effects, which is often taken to be ⬃2 Gy. 13,14 For radiation doses below the deterministic threshold dose, the patient
risk consists of the stochastic risk of carcinogenesis and the
induction of genetic effects.15 For a given type of radiologic
examination such as interventional neuroradiologic proce-
0094-2405Õ2002Õ29„8…Õ1643Õ8Õ$19.00
© 2002 Am. Assoc. Phys. Med.
1643
1644
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
1644
TABLE I. Scoring scheme used to assess the visibility of vessel with varying
concentrations of iodine contrast material.
FIG. 1. Schematic diagram of the acrylic phantom with the vessel and blank
inserts used to simulate small vessels for the purpose of evaluating image
quality in neuroradiology.
dures, the stochastic risk may be taken to be directly proportional to the total energy imparted to the patient.16 Reducing
the total energy imparted to the patient will minimize the
stochastic risk, and help to ensure that the patient dose is as
low as reasonably achievable 共ALARA兲.17
DSA image quality needs to be adequate for the specified
imaging task, and to ensure that there are no adverse clinical
consequences as a result of inadequate visualization of catheters or vasculature. Changing radiographic technique factors
for DSA procedures will impact both the surface dose and
energy imparted.18,19 Any attempt to improve image quality
by modification of radiographic technique factors should also
require that the corresponding patient radiation dose be minimized. A quantitative understanding of the relationship between image quality and patient dose is therefore an important step towards the development of imaging protocols that
FIG. 2. Digitally subtracted angiographic image of the DSA acrylic phantom
with the 1.0 mm vessel insert.
Medical Physics, Vol. 29, No. 8, August 2002
Score
Description
1
2
3
4
5
Vessel not seen
Vessel poorly seen (⬍50%)
50% of vessel delineated
Most of vessel visible (⬎50%)
Vessel perfectly visible
maintain adequate diagnostic imaging performance whilst
ensuring that patient radiation dose is kept ALARA. In this
study, we investigated how changing x-ray tube voltage and
input exposure to the image receptor affects radiation dose
and image quality in DSA phantom studies that simulate interventional neuroradiologic procedures.
METHOD
Image quality phantom
The phantom used to assess image quality consisted of
thirteen stacked acrylic blocks with dimensions of 30 cm
⫻30 cm⫻1.3 cm. 20 The acrylic vessel insert measured
30 cm⫻9.0 cm⫻1.3 cm and had 30 cylindrical vessels 1.0
mm in diameter and 35 mm in length drilled along its midplane at intervals of 8.0 mm apart. The vessels were filled
with iodinated contrast prepared from Ultravist 共Berlex
Laboratories, Wayne, NJ兲 300 iopromide solution diluted in
heparin solution to accurately prepare very low concentrations of iodine. Heparin was selected as the dilution medium
because this is what is used clinically to deliver iodine contrast into the blood stream. The contrast medium was mixed
thoroughly before injected into the 1.0 mm vessels of the
DSA phantom. To avoid settling of the iodine contrast at the
bottom of the vessel walls, the vessel insert was positioned
with the vessels oriented horizontally and exposures were
taken shortly after the mixing was done with the vessel insert
shaken in between exposures. The iodine concentrations of
the contrast medium used to fill each vessel ranged from 4.5
mg/cc of iodine to 50 mg/cc, with the ratio of iodine concentration in any two adjacent vessels being 0.92. The acrylic
phantom with a vessel insert appropriate for DSA is shown
in Fig. 1.
In this study, an image quality index was defined as the
lowest concentration of iodine in a 1.0 mm diameter vessel
that may be visually detected in a DSA image. An example
of DSA images obtained for this study is shown in Fig. 2.
The five-point scale used to evaluate the DSA images is summarized in Table I. An observer was trained to score images
of different iodine concentration using a representative series
of ten DSA images of the phantom obtained during preliminary experimental work. The observer was asked to maintain
a consistent criterion in using the ranking scheme listed in
Table I. After training, images were presented in a random
order. The concentration in the vessel that corresponded to a
score of three 共see Table I兲 was taken as the iodine detection
1645
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
threshold in this study, with linear interpolation used where
necessary.
TABLE II. Imaging techniques during tube voltage experiments.
1.1 ␮Gy/frame
DSA images
All DSA acquisitions were performed using a biplane
Toshiba 共Toshiba America Medical Systems, Tustin, CA兲
KXO-80 high voltage diagnostic x-ray generator and the
Toshiba DFP-2000A/A3 digital fluorography system configured for interventional neuroradiologic procedures. The TV
camera used on the Toshiba imaging system was a MTV500A, high definition (1024⫻1024), low noise CCD camera. The x-ray source-to-image receptor distance 共SID兲 was
set to the maximum value of 105 cm, with the acrylic phantom positioned so that the geometric magnification of the
vessel insert was 1.2. DSA acquisitions were obtained using
the 0.6 mm focal spot size and a 23 cm diameter image
intensifier mode. The x-ray tube filtration in the frontal and
lateral imaging planes was equivalent to 3.0 mm of aluminum 共tube inherent filtration兲. An additional 3.0 mm aluminum equivalent filtration was added in the frontal plane by
the x-ray table. The generator was set to manual techniques
allowing fine adjustments of the tube voltage 共kVp兲, tube
current 共mA兲, exposure time 共ms兲 and optical gain 共TV iris兲.
The x-ray tube voltage was monitored using a Machlett Dynalyser III 共Greenwich Instrument CO., Inc., Greenwich,
CT兲. A 10⫻5-60 共the 10⫻5-60 ionization chamber has a
gain of ⫻10 for better resolution兲 ionization chamber of a
MDH 1015C exposure meter was positioned behind the grid
of the image intensifier to record the input exposure to the
image receptor. An exposure of 2.58⫻10⫺4 C/kg 共1 Roentgen兲 was taken to correspond to an air kerma of 8.73 mGy.21
During digital subtraction acquisitions, a mask frame was
acquired of the blank phantom insert. The blank insert was
replaced by the vessel insert and twenty additional frames
were acquired 共3 frames/s兲, with the tenth frame of each
DSA image acquisition used for the observer study. The window and level of the displayed images were adjusted to optimize signal detection during each DSA image acquisition.
The optical gain was controlled by electronic adjustments of
the iris diameter to produce a constant pixel value between
2000 and 2100, which corresponds to the target values used
by the biplane imaging system for routine patient examinations.
In the first experiment, a series of measurements were
performed where the air kerma at the input of the image
intensifier was maintained at either 1.1 ␮Gy/frame or 3.9
␮Gy/frame, and with the x-ray tube voltage varied between
50 and 100 kVp. As the x-ray tube voltage was increased, the
corresponding mAs value was reduced to maintain a constant
air kerma level at the image intensifier input, which was
monitored with an ionization chamber. Table II summarizes
the technique factors used in these experiments 共i.e., kVp
and mAs兲 together with corresponding measured pixel values. The data in Table II demonstrate the pixel value was
kept relatively constant in these experiments, with average
values (⫾standard deviation) of 2068⫾34 at 1.1 ␮Gy/frame
and 2039⫾56 at 3.9 ␮Gy/frame. In the second experiment,
Medical Physics, Vol. 29, No. 8, August 2002
1645
a
3.9 ␮Gy/frame
kVp
mAs
Video level
mAs
Video level
50
56
60
66
70a
80
90
100
28.0
14.4
9.28
5.40
4.00
2.24
1.50
0.960
2075
2018
2104
2028
2068
2054
2117
2079
80.0
47.5
33.0
20.5
17.0
7.50
4.60
3.20
1914
2052
2031
2085
2018
2056
2080
2075
Baseline techniques.
another series of measurements were obtained at a constant
tube voltage of 70 kVp, and with the input air kerma to the
image intensifier varied from 0.44 ␮Gy/frame to 8.8 ␮Gy/
frame in thirteen steps. The optical gain was adjusted to
maintain constant pixel value as the input air kerma was
modified, with an average pixel value for the thirteen measurements of 2051⫾28.
Radiation dosimetry
The entrance exposure to the acrylic phantom was recorded by a 10⫻5-6 ionization chamber of an MDH 1015C
exposure meter attached to the beam entrance surface of the
phantom during each DSA acquisition. The ionization chamber was positioned to avoid any overlap with the iodinated
vessels of the vessel insert, and the measured exposure included backscatter radiation coming from the acrylic phantom. Frame entrance exposures were obtained from the integral exposure measured for each image acquisition sequence
divided by the number of acquired frames. A conversion factor of 2.58⫻10⫺4 C/kg 共1R兲 corresponding to an absorbed
dose of 9.3 mGy for muscle tissue was used to convert the
beam entrance exposure to surface dose.21
The energy imparted was computed using the method described in detail elsewhere.22,23 The total phantom thickness
was 16.9 cm of acrylic, which was taken to be equivalent to
20 cm of water for computing values of energy imparted,
given that the density of acrylic is 1.19 g/cm3 . 24 Since the
exposure was obtained from direct exposure measurements
during each image acquisition sequence, it included backscatter. The backscatter radiation fraction was measured for
the corresponding phantom at each applied tube voltage and
was subtracted from the measured exposures to obtain the
free-in-air exposure.20 The exposure area at the beam entrance plane of the phantom was computed from geometry
assuming a circular beam area of 23 cm diameter at the image intensifier plane. Values of energy imparted for head
x-ray examinations were converted into the corresponding
adult effective doses using E/␧ 共mSv/J兲 conversion factors
averaged for the posteroanterior and lateral views and ranged
between 4.1 and 5.9 mSv/J depending on the selected x-ray
tube voltage.25
1646
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
1646
FIG. 3. Threshold iodine concentration as a function of tube voltage. The
asterisks correspond to the 3.9 ␮Gy/frame, and the triangles correspond the
1.1 ␮Gy/frame.
RESULTS
Measurement precision
Five digital subtraction acquisition sequences, obtained at
70 kVp, were generated at different times over a one-day
period. These five sequences were used to determine the radiation dosimetry precision and reader performance precision. The average entrance exposure measured was 0.852
⫾0.006 mGy/frame. The average of five observer readings
for the threshold contrast concentration at 70 kVp and 2.6
␮Gy/frame was 8.0⫾0.96 mg/cc. As expected, the precision
of radiation exposure measurements (⬍1%) is much lower
than the reproducibility of the detection threshold value
共12%兲, given that the latter is based on the subjective assessment of a human observer.26
X-ray tube voltage
Figure 3 shows the threshold iodine concentration as a
function of tube voltage. The solid and dashed lines shown in
Fig. 3 are least squares fits to the experimental data. At 1.1
␮Gy/frame, the threshold iodine concentration was proportional to the kVp 2.1, where the coefficient of determination
of the least squares fit (r 2 ) was 0.92. At this input exposure,
a 10 kVp increase of the tube voltage from the baseline 共70
kVp兲 increased the threshold iodine concentration by 32%.
Reducing the x-ray tube voltage from 70 kVp to 60 kVp
reduced the detection threshold concentration by 27%. At 3.9
␮Gy/frame, the threshold concentration was proportional to
kVp 1.6, where the coefficient of determination of the least
squares fit (r 2 ) was 0.97. At an exposure level of 3.9 ␮Gy/
frame, increasing the tube voltage from 70 kVp to 80 kVp
increased the threshold concentration by 23%, whereas reducing the x-ray tube voltage from 70 kVp to 60 kVp reduced the detection threshold concentration by 21%. Over
the x-ray tube voltage range investigated in this study 共50–
100 kVp兲, the average increase in detection threshold as a
result of reducing the exposure level from 3.9 to 1.1 ␮Gy/
frame was 1.7⫾0.3.
Figure 4 shows the relative behavior of surface dose and
energy imparted as a function of tube voltage. The data of
Medical Physics, Vol. 29, No. 8, August 2002
FIG. 4. Relative behavior of surface dose and energy imparted as a function
of tube voltage. Solid diamonds and lines correspond to 1.1 ␮Gy/frame;
triangles and broken lines correspond to 3.9 ␮Gy/frame. Data has been
normalized so that values at 70 kVp were equal to 1.
Fig. 4 have been fitted to power functions with resulting
coefficient of determination (r 2 ) between 0.95 and 0.98.
Table III provides a summary of the key dosimetry parameters 共i.e., surface dose, energy imparted and effective dose兲
obtained at the two exposure levels investigated. As expected, relative changes of both surface dose and energy imparted as a function of x-ray tube voltage were similar at
TABLE III. DSA dosimetry summary.
a
kVp
Surface dose
共mGy/frame兲
Energy imparted
共mJ/frame兲
Effective dose
共␮Sv/frame兲
50
60
70
80
90
100
1.2a关3.9兴b
0.62关2.4兴
0.38关1.3兴
0.28关0.96兴
0.24关0.65兴
0.20关0.66兴
1.6关5.0兴
0.95关3.7兴
0.66关2.3兴
0.54关1.9兴
0.50关1.4兴
0.44关1.5兴
6.5 关20兴
4.4 关17兴
3.3 关11兴
2.9 关10兴
2.8关7.9兴
2.6关8.8兴
1.1 ␮Gy/frame input air kerma.
3.9 ␮Gy/frame input air kerma.
b
1647
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
1647
FIG. 5. Threshold iodine concentration as a function of image intensifier
input air kerma at 70 kVp and a constant pixel intensity.
exposure levels of 1.1 ␮Gy/frame and 3.9 ␮Gy/frame. Increasing the x-ray tube voltage from 50 kVp to 100 kVp
reduced the surface dose by a factor of 6.1, and reduced the
energy imparted by a factor of 3.5.
Input air kerma
Figure 5 shows the threshold iodine concentration as a
function of image intensifier input air kerma at a constant
x-ray tube voltage of 70 kVp where both the ordinate and
abscissa are plotted using a logarithmic scale. Data below
and above 2.2 ␮Gy/frame were fitted separately to a least
squares fit to a line. Below 2.2 ␮Gy/frame, the detection
threshold concentration was found to be proportional to
X ⫺0.57 where the coefficient of determination (r 2 ) was 0.98.
Above 2.2 ␮Gy/frame, the detection threshold concentration
was found to be proportional to X ⫺0.26 where the coefficient
of determination (r 2 ) was 0.96.
For our DSA imaging system, increasing the image intensifier input air kerma from 1.1 ␮Gy/frame to 2.2 ␮Gy/frame
reduced the detection threshold concentration by 34%,
whereas increasing the input air kerma from 3.9 ␮Gy/frame
to 7.8 ␮Gy/frame reduced the detection threshold concentration by only 16%. The empirical data presented in Fig. 5
indicate that above an input air kerma of about 2.2 ␮Gy/
frame, there is only modest improvement in the detection
threshold concentration of 1.0 mm diameter vessels with increasing input air kerma. These results are only applicable to
the imaging system investigated in this study, and could be
different for other types of DSA imaging system.
air kerma by 63%. At 3.9 ␮Gy/frame, a 25% reduction in
detection threshold concentration may be achieved by either
lowering the tube voltage to 59 kVp, or by increasing the
input air kerma by a factor of 200%. The increased surface
dose and energy imparted required to achieve this 25% reduction on the iodine threshold concentration are summarized in Table IV. Improvements in image quality by reducing the x-ray tube voltage always resulted in lower doses
than could be achieved by increasing the x-ray intensity level
to the image receptor at a constant x-ray tube voltage.
Radiation dose versus image quality
DISCUSSION
Figures 6 and 7 show the relative increases in surface
dose and energy imparted as the iodine detection thresholdconcentration is reduced by changing either the image intensifier input air kerma or the x-ray tube voltage. The data
generated in Figs. 6 and 7 are based on the least squares
curves fitted to the empirical data in Figs. 3–5.
At 70 kVp and 1.1 ␮Gy/frame, a 25% reduction in detection threshold concentration may be achieved by either lowering the tube voltage to 61 kVp, or by increasing the input
For a film-screen imaging system, the exposure level is
effectively fixed by the need to produce an optimum optical
density. Any variations in the exposure level in screen-film
radiography will produce over 共or under兲 exposed films, with
a concomitant drop in image contrast. In digital imaging systems, however, the range of receptor input exposure can vary
by over an order of magnitude 共e.g., Fig. 5兲, yet still produce
diagnostic quality images. The advent of relatively sophisticated digital imaging equipment thus offers considerable
Medical Physics, Vol. 29, No. 8, August 2002
FIG. 6. Increased surface dose resulting from reducing the x-ray tube potential or increasing the exposure level to achieve the specified improvement in
threshold detection concentration of iodine.
1648
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
1648
TABLE V. Computed mean x-ray energies of x-ray spectra, half-value layers
共mm Al兲, and iodine attenuation coefficients as a function of x-ray tube
voltage.
kVp
Incident mean
energy and
HVL
Exit mean
energy and
HVL
Mass
attenuation
coefficient of I
(cm2 /g) c
50
60
70
80
90
100
32.6a关1.8兴b
36.3关2.2兴
39.7关2.5兴
43.1关2.8兴
46.2关3.2兴
49.0关3.6兴
39.1关2.2兴
44.6关3.9兴
49.4关5.2兴
54.0关6.26兴
58.1关7.2兴
61.7关7.9兴
23.4
16.6
12.7
10.0
8.26
7.03
a
Incident mean energy in keV.
Half-value layer in mm Al.
c
Mass attenuation coefficient is determined at the mean x-ray energy.
b
FIG. 7. Increased energy imparted resulting from reducing the x-ray tube
potential or increasing the exposure level to achieve the specified improvement in threshold detection concentration of iodine.
flexibility in acquiring x-ray image data.27 The wide range of
input radiation exposures, as well as variations in the selected x-ray tube voltage, could play a significant role in
helping to optimize patient dose and image quality. A quantitative understanding of the relationship between dose and
image quality is therefore important to help operators make
the best use of digital imaging equipment. Improvements in
imaging performance should require the modification of
those radiographic technique factors that minimize the patient radiation risk. Conversely, if it is possible to reduce
TABLE IV. Changes in surface dose and energy imparted required to reduce
the iodine threshold concentration by 25% by reducing the x-ray tube potential, or by increasing the input air kerma to the image intensifier.
1.1 ␮Gy/frame
3.9 ␮Gy/frame
Parameter
X-ray tube
potential
Exposure
level
X-ray tube
potential
Exposure
level
Surface dose
Energy imparted
44%
29%
63%
63%
70%
46%
200%
200%
Medical Physics, Vol. 29, No. 8, August 2002
image quality without adversely affecting diagnostic performance, technique factors should be modified to maximize
patient dose savings.
In screen-film radiography, it is well established that increasing x-ray tube voltage or x-ray beam quality degrades
image quality.18,28 –30 The data presented in Fig. 3 shows that
tube voltage has a significant effect on the iodine detection
threshold concentration in DSA, as well. Changing the x-ray
tube voltage will predominantly affect the contrast of the
signal detected by the imaging system if the absolute x-ray
intensity level remains constant. Reducing the x-ray tube
voltage lowers the average photon energy and increases
x-ray absorption because of the rapid rise in photoelectric
absorption as the average photon energy approaches the iodine 33 keV K edge. An additional factor affecting image
quality at different x-ray tube voltages is the presence of
scatter radiation. Scatter radiation generally increases with
increased x-ray tube voltage, and will thereby reduce image
contrast.31
The data in Fig. 3 may be compared with theoretical
changes in x-ray spectral energies and x-ray absorption coefficients for iodine. For the theoretical calculation, average
incident photon energies and half-value layers 共HVL兲 were
computed for a 3 mm Al filtered x-ray tube running at constant voltage.32 The exit mean energy and HVL were obtained by filtering the x-ray spectra with 20 cm of water. The
absorption coefficient of iodine was determined using the
exit mean photon energies.21,33 Table V shows the computed
values of incident photon energy and exit photon energy,
together with the corresponding iodine mass attenuation coefficient. Reducing the x-ray tube voltage from 100 kVp to
50 kVp would be expected to improve the image contrast by
about a factor of 3. The relationship between selected kVp
and mass attenuation coefficient in Table V was fitted to a
power law 共i.e., ␮ inversely proportional to kVp n 兲, with the
theoretical value of n determined to be 1.7. This theoretical
prediction may be compared to the experimental data obtained at doses where quantum mottle is dominant—this corresponds to a slope of 2.1 obtained at an exposure rate of 1.1
␮Gy/frame 共see Fig. 3兲. The differences between the measured and theoretical values may be due to uncertainties in
1649
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
the x-ray spectra as well as changes in quantum noise as a
function of kVp which were neglected in our computations.
The primary source of noise in 共digital兲 x-ray imaging is
usually quantum mottle, which corresponds to random spatial fluctuations of the distribution of x-ray quanta absorbed
by the detector.8,34 Since the production and attenuation of x
rays are governed by Poisson statistics, quantum mottle is
inversely proportional to the square root of the radiation exposure level used to generate an image. The results shown in
Fig. 5 demonstrate that for input air kerma smaller than 2.2
␮Gy/frame, this specific DSA imaging system is quantum
noise limited, since the observed performance shows proportionality close to the value of to X ⫺0.5 expected for a quantum noise limited system. A significant performance gain can
be achieved by increasing the input air kerma at these lower
exposure levels.
At exposure levels greater than 2.2 ␮Gy/frame, the measured iodine detection threshold concentration is proportional to X ⫺0.26, demonstrating that imaging performance is
no longer limited by quantum mottle alone. At these higher
exposure levels, additional noise factors become significant
and would need to be included in any analysis of system
imaging performance. These noise sources include factors
such as electronic noise, time jitter and structure noise.8,34 –37
The relative importance of these additional noise sources is
also the reason for the differences in the slopes observed in
Fig. 3. For a quantum noise limited system, the slope of the
curve plotting threshold concentration and kVp would be
independent of the radiation exposure level. System noise
will set an upper limit to the x-ray flux, which may be taken
to occur when the quantum and electrical noise components
are equal.38
One limitation of our study is that the phantom used had a
constant thickness which is rarely achieved in clinical practice. For a non-uniform patient, the highest exposure level
occurs in the thinest body region and this should be mapped
to the maximum signal level of the video camera.38 The resultant signal to noise ratio of the vessel containing the iodine will depend on the transmitted x-ray intensity in the
region of interest. This limitation is of much greater importance in nonuniform areas such as the lung or extremities,
than for abdominal or head imaging. Bright spots in DSA
imaging should be minimized by appropriate bolusing to attempt to ensure that all regions of the image are obtained at
maximum signal levels.31
The data presented in Figs. 6 and 7 indicate how modification of radiographic technique factors affect imaging performance and patient dose, given a detection task of identifying a circular vessel 共1.0 mm diameter兲 containing iodine
contrast media. It is evident that achieving improved imaging
performance by reduction in x-ray tube voltage is the preferred option, since the patient doses are significantly lower
than when the same level of imaging performance is
achieved by increasing the input radiation exposure level to
the image receptor. Our results also show that increasing the
image intensifier input air kerma beyond 2.2 ␮Gy/frame on
our specific DSA system provided relatively little improvement in measured imaging performance. Use of such higher
Medical Physics, Vol. 29, No. 8, August 2002
1649
radiation exposure levels would require explicit justification
in terms of the anticipated benefit to the patients to offset risk
of deterministic effects and the higher stochastic risks.
At our institution, the typical interventional neuroradiologic DSA x-ray tube voltages are currently 76 to 80 kVp
for the frontal plane and 66 to 70 kVp for the lateral plane,
and with corresponding air kerma values at the input of the
image intensifier set at 4.4 and 6.1 ␮Gy/frame. A typical
diagnostic examination can result in 400 DSA images in each
of the two imaging projection planes. As a result, patient
surface doses from DSA imaging could easily be of the order
of 0.65 Gy, which would need to be added to any dose associated with fluoroscopic imaging, typically 0.25 Gy. Such
data clearly demonstrate the potential for inducing deterministic effects in DSA studies, particularly in studies that require an increased number of radiographic images. Typical
value of energy imparted to a patient undergoing an interventional neuroradiologic procedure is 6 J, and the corresponding patient effective dose would therefore be of the order of
30 mSv.20 These effective doses are much higher than is
normally encountered in conventional radiography.39 DSA
patient effective doses are at the upper end of the range of
patient doses in radiology, and emphasize the need to pay
careful attention to minimizing patient doses.
CONCLUSION
The results obtained in this study demonstrate that for
improving low contrast detection on our DSA imaging system, reducing x-ray tube voltage 共i.e., kVp兲 was a superior
strategy to that of increasing the x-ray beam intensity level
共i.e., mAs兲. For a given x-ray DSA imaging system, there is
an exposure level at which quantum mottle will cease to be
the dominant source of image noise. For the DSA system
investigated in this work, this exposure level was found to be
about 2.2 ␮Gy per frame at the input to the image intensifier.
For this specific DSA system, increasing the image intensifier input exposure level beyond 2.2 ␮Gy per frame would be
of limited benefit given that the improvement in low contrast
detection performance is relatively modest.
ACKNOWLEDGMENTS
We would like to thank Lynn Rill for reading the images,
and Dennis Pinner 共Toshiba兲 for technical support in using
the biplane imaging system.
a兲
Current address: LORAD, A Hologic Company, 36 Apple Ridge Road,
Danbury, CT 06810.
1
F. Vinuela, Interventional Neuroradiology 共Raven, New York, 1992兲.
2
A. G. Osborn, Diagnostic Cerebral Angiography, 2nd ed. 共Lippincott,
Philadelphia, PA, 1999兲.
3
R. I. Grossman and D. M. Yousem, Neuroradiology: The Requisites
共Mosby-Year Book, St. Louis, MO, 1994兲.
4
J. J. Caro, E. Trindade, and M. McGregor, ‘‘The risks of death and of
severe nonfatal reactions with high- vs low-osmolality contrast media: a
metaanalysis,’’ J. Radiol. 156, 825– 832 共1991兲.
5
W. M. Thompson, R. A. Halvorsen, R. K. Gedgaudas, F. M. Kelvin, R. P.
Rice, S. Woodfield, G. A. Johnson, L. W. Hedlung, and D. B. Jorgensen,
‘‘High kVp vs. low kVp for T-tube operative cholangiography,’’ Radiology 146, 635– 642 共1983兲.
1650
Gkanatsios, Huda, and Peters: Effect of radiographic techniques
6
M. J. Tapiovaara and M. Sandborg, ‘‘Evaluation of image quality in fluoroscopy by measurements and Monte Carlo calculations,’’ Phys. Med.
Biol. 40, 589– 607 共1995兲.
7
H. Fujita, K. Doi, M. L. Giger, and H.-P. Chang, ‘‘Investigation of basic
imaging properties in digital radiography. 5. Characteristic curves of
II-TV digital systems,’’ Med. Phys. 13, 13–18 共1986兲.
8
M. L. Giger, K. Doi, and H. Fujita, ‘‘Investigation of basic imaging
properties in digital radiography. 7. Noise Wiener spectra of II-TV digital
imaging systems,’’ Med. Phys. 13, 131–138 共1986兲.
9
B. Berthelsen and Å. Cederblad, ‘‘Radiation doses to patients and personnel involved in embolization of intracerebral arteriovenous malformations,’’ Acta Radiol. 32, 492– 497 共1991兲.
10
N. A. Gkanatsios, W. Huda, K. R. Peters, and J. A. Freeman, ‘‘Evaluation
of an on-line patient exposure meter in neuroradiology,’’ Radiology 203,
837– 842 共1997兲.
11
W. Huda and K. R. Peters, ‘‘Radiation-induced temporary epilation after
a neuroradiologically guided catheterization procedure,’’ Radiology 193,
642– 644 共1994兲.
12
E. Vano, L. Arranz, J. M. Sastre, C. Moro, A. Ledo, M. T. Garate, and I.
Minguez, ‘‘Dosimetric and radiation protection considerations based on
some cases of patient skin injuries in interventional cardiology,’’ Br. J.
Radiol. 71, 510–516 共1998兲.
13
L. K. Wagner, P. J. Eifel, and R. A. Geise, ‘‘Potential biological effects
following high x-ray dose interventional procedures,’’ J. Vasc. Interv Radiol. 5, 71– 84 共1994兲.
14
International Commission on Radiological Protection 共ICRP兲, Publication 47: Annals of the ICRP 14(3). Non-Stochastic Effects of Ionizing
Radiation 共Pergamon, Oxford, 1984兲.
15
International Commission on Radiological Protection 共ICRP兲, Publication 60: Annals of the ICRP 21(1–3). 1990 Recommendations of the
International Commission on Radiological Protection 共Pergamon, Oxford, 1991兲.
16
B. Wall, R. Harrison, and F. Spiers, ISPM Report #53: Patient Dosimetry
Techniques in Diagnostic Radiology 共The Institute of Physical Science in
Medicine, York, England, 1988兲.
17
International Commission on Radiological Protection 共ICRP兲, Publication 34: Protection of the patient in diagnostic radiology 共Pergamon,
Oxford, 1982兲.
18
P. C. Shrimpton, D. J. Jones, and B. F. Wall, ‘‘The influence of tube
filtration and potential on patient dose during x-ray examinations,’’ Phys.
Med. Biol. 33, 1205–1212 共1988兲.
19
A. M. Norbash, D. Busick, and M. P. Marks, ‘‘Techniques for reducing
interventional neuroradiologic skin dose: tube position rotation and
supplemental beam filtration,’’ Am. J. Neuroradiol. 17, 41– 49 共1996兲.
20
N. A. Gkanatsios, Doctoral thesis: Patient doses and image quality in
interventional neuroradiology 共University of Florida, Gainesville, FL,
1998兲.
Medical Physics, Vol. 29, No. 8, August 2002
21
1650
H. E. Johns and J. R. Cunningham, The Physics of Radiology, 4th ed.
共Charles C. Thomas, Springfield, IL, 1983兲.
22
N. A. Gkanatsios, Master thesis: Computation of energy imparted in diagnostic radiology 共University of Florida, Gainesville, FL, 1995兲.
23
N. A. Gkanatsios and W. Huda, ‘‘Computation of energy imparted in
diagnostic radiology,’’ Med. Phys. 24, 571–579 共1997兲.
24
The Health Physics and Radiological Health Handbook, edited by B.
Shleien 共Scinta, Inc., Silver Springs, MD, 1992兲.
25
W. Huda and N. A. Gkanatsios, ‘‘Effective doses and energy imparted in
diagnostic radiology,’’ Med. Phys. 24, 1311–1316 共1997兲.
26
W. R. Hendee and P. N. T. Wells, The Perception of Visual Information,
2nd ed. 共1997兲.
27
W. R. Hendee and J. H. Trueblood, Medical Physics Monograph No. 22:
Digital Imaging 共Medical Physics Publishing, Madison, WI, 1993兲.
28
E. D. Trout, J. P. Kelley, and G. A. Cathey, ‘‘The use of filters to control
radiation exposure to the patient in diagnostic roentgenology,’’ Am. J.
Roentgenol., Radium Ther. Nucl. Med. 57, 946 –963 共1952兲.
29
K. Koedooder and H. W. Venema, ‘‘Filter materials for dose reduction in
screen-film radiography,’’ Phys. Med. Biol. 31, 585– 600 共1986兲.
30
H. D. Nagel, ‘‘Comparison of performance characteristics of conventional and K-edge filters in general diagnostic radiology,’’ Phys. Med.
Biol. 34, 1269–1287 共1989兲.
31
B. H. Hasegawa, The Physics of Medical X-Ray Imaging, 2nd ed. 共Medical Physics Publishing, Madison, WI, 1991兲.
32
M. Tucker, G. Barnes, and D. Chakraborty, ‘‘Semiempirical model for
generating tungsten target x-ray spectra,’’ Med. Phys. 18, 211–218
共1991兲.
33
J. Hubbell, ‘‘Photon mass attenuation and energy-absorption coefficients
from 1 keV to 20 MeV,’’ Int. J. Appl. Radiat. Isot. 33, 1269–1290 共1982兲.
34
H. Fujita, K. Ueda, J. Morishita, T. Fujikawa, A. Ohtsuka, and T. Sai,
‘‘Basic imaging properties of a computed radiographic system with photostimulable phosphors,’’ Med. Phys. 16, 52–59 共1989兲.
35
B. A. Arnold and P. O. Scheibe, ‘‘Noise analysis of a digital radiography
system,’’ Am. J. Roentgenol. 142, 609– 613 共1984兲.
36
M. K. Giger, Ph.D. dissertation: Investigation of basic imaging properties
in digital radiography 共University of Chicago, Chicago, IL, 1985兲.
37
J. Esthappan, H. Harauchi, and K. R. Hoffmann, ‘‘Evaluation of imaging
geometries calculated from biplane images,’’ Med. Phys. 25, 965–975
共1998兲.
38
R. A. Kruger, C. A. Mistretta, and S. J. Riederer, ‘‘Physical and technical
considerations of computerized fluoroscopy difference imaging,’’ IEEE
Trans. Nucl. Sci. NS 28, 205–212 共1981兲.
39
United Nations Scientific Committee on the Effects of Atomic Radiation
共UNSCEAR兲, 1993 Report to the General Assembly: Medical Radiation
Exposures 共United Nations, New York, 1993兲.