Performance Evaluation and Quality Assurance in Digital Subtraction Angiography

Performance Evaluation and Quality Assurance in Digital Subtraction Angiography
Published by the American Institute of Physics
for the American Association of Physicists in Medicine
Maynard High, Chairman
Gerald Cohen
Pei-Jan Paul Lin
Arthur Olson
Lawrence Rothenberg
Lois Rutz
Gerald Shapiro
Chorng-Gang Shaw
Louis Wagner
May 1985
Published for the
American Association of Physicists in Medicine
by the American Institute of Physics
Further copies of this report may be obtained from
Executive Secretary
American Association of Physicists in Medicine
335 E. 45 Street
New York, NY 10017
Library of Congress Catalog Card Number: 85-72829
International Standard Book Number: 0-88318-482-6
International Standard Serial Number: 0271-7344
Copyright © 1985 by the American Association
of Physicists in Medicine
All rights reserved. No part of this publication may be
reproduced, stored in a retrieval system, or transmitted
in any form or by any means (electronic, mechanical,
photocopying, recording, or otherwise) without the
prior written permission of the publisher.
Published by the American Institute of Physics, Inc.,
335 East 45 Street, New York, New York 10017
Printed in the United States of America
Description of System and Performance Parameters
DSA Systems
Performance Parameters and Factors Affecting Them
Spatial Resolution
Low Contrast Performance
Contrast & Spatial Uniformity
Contrast Linearity
Radiation Exposure
Subtraction Artifacts
III. Performance Evaluation With A Phantom
Typical Phantom
Spatial Resolution
Low Contrast Performance
Radiation Exposure
Quality Assurance
Radiographic systems capable of digital subtraction angiography
(DSA) are now to be found in most medium-sized and large
hospitals, as well as in private offices and outpatient clinics.
At present there is no standard method of evaluating these units,
nor is there much guidance from the manufacturers regarding
The Digital Radiography/Fluorography Task
quality assurance.
Group has prepared this report with the following goals; (A)
define a set of performance parameters that relate to DSA image
quality, (B) describe the components of a digital subtraction
system and the way in which they contribute to image quality, and
(C) recommend a method of measuring the defined performance
parameters so they can be evaluated, optimized and monitored in a
quality assurance program.
The Task Group feels that the simplest and clinically most
meaningful way of measuring and monitoring the performance
parameters is by us of a properly designed non-invasive test
Although it is possible to evaluate the individual
components of the system (such as focal spot, image intensifier,
TV chain, etc.), and indeed it is advised that one should do this
in a proper acceptance test, it is actually the final image from
a chain of interrelated components that is used in the clinical
A phantom that simulates the clinical situation
allows meaningful evaluation of the DSA system without the timeconsuming disassembly needed for evaluation of the individual
components in the imaging chain. This report does not recommend
one specific
phantom design, but rather a rational method of
measuring performance parameters.
A certain phantom design will
be described for the purpose of illustrating the concepts of this
however, any phantom which can adequately measure the
performance parameters described in this report is equally
The user must always be aware that, for meaningful
comparisons of DSA systems,
for comparisons with manufacturers'
verification of
specifications, the same phantom must be used, and in the same
way. The Task Group feels strongly that the manufacturers of DSA
systems must include a properly designed phantom with each unit.
This should be specified in the purchase document.
A. Digital Subtraction Angiography Systems
Commercial DSA systems come in many forms with varying features,
but they all share a common approach: subtraction of a mask image
(background) from an image obtained at a slightly different time
that contains a contrast signal due to vascular injection of
iodine contrast material.
The visual contrast of the subtracted
image is enhanced in DSA because the background is subtracted
away and the remaining digitized signal can be enhanced through
Many modes of image
averaging, and windowing.
acquisition are available, the most common being pulse mode, in
which a high tube current (100-1000 mA) tine-type x-ray pulsed
image is collected at the rate of few frames/second during
injection of contrast material and subtracted from a stored mask
obtained just before injection.
Also common is continuous x-ray
mode, in which a lower tube current (5-50 mA) remains on
continuously during the study, high frame-rate (3Of/s) image
collection is used, and a number of frames are averaged before
Also available is a
subtraction to decrease quantum noise.
fluoroscopy mode, in which contrast images can be 'obtained at the
rate of 30 frames/second by subtraction of subsequent frames
from previous frames, sometimes with averaging or weighted
averaging. Some DSA systems can be operated in more than one, or
all, of the above modes.
In this report we will concentrate on simple mask-mode time
subtraction DSA methods and will not address the other developing
such as
subtraction, matched filtering, and recursive filtering. It is
likely that these systems can also be evaluated by the methods
described in this report, although some additional modules would
have to be developed for the phantom to allow for the dynamic
change in contrast required for these systems to operate.
Figures 1 & 2 are block diagrams of typical DSA systems.
systems employ an image intensifier (II) as the x-ray detector,
although there are systems with solid-state detection assemblies.
The x-ray and fluoroscopy system is quite standard; however, a
very-low-noise TV camera is often employed, sometimes with a
A remotely controlled
progressive-scan or slow-scan readout.
aperture is often placed before the TV camera since, in the
pulsed mode, the luminance at the II output phosphor may be 1000
times higher than that in the fluoroscopy mode, and the dynamic
range of the camera is not large enough to handle this range of
light intensities.
The video signal from the TV camera is digitized in the analogto-digital converter (ADC) and sent to one of two image memories.
If the image is a pre-injection mask, it is placed in one memory,
All subsequent
where it remains for the duration of the run.
images are placed in the other memory, where they will each
The mask image is
replace the previous image residing there.
then subtracted from each image in the other memory, as these
images are acquired, to give a series of difference images. The
difference images are then "windowed and leveled" to enhance the
contrast visibility, changed back to an analog signal by the
digital-to-analog converter (DAC), and finally sent to the
viewing monitor.
If the DSA system uses a digital storage device such as a digital
disk, the digitized images are normally stored as shown in Figure
If the system uses an analog storage device, such as a video
disk or a video tape, normally only the analog images are stored,
and redigitization is required for any post-processing of the
may be accomplished on video hard-copy
data. Long term storage
film, floppy disk, magnetic tape, or video tape.
Logarithmic processing is often used because it achieves density
uniformity of the vessels in the subtracted image as they pass
under bone and varying thicknesses of overlying tissue. This may
be done by using a logarithmic amplifier on the incoming analog
signal, or by modifying the digitized signal using a look-up
The operator's console allows initiation of the exposures and
choice of the imaging mode and frame rate; it controls the
injector and allows post-processing of the data. Post-processing
reregistration to remove motion artifacts,‘ and quantitative
analysis of the iodine concentration.
Performance parameters for DSA systems which are related to image
quality and which can be evaluated with a non-invasive phantom
are the following:
Spatial resolution
Low-contrast performance
Contrast and spatial uniformity
Contrast linearity
Radiation exposure
Subtraction artifacts
1) Spatial Resolution
Spatial resolution is a measure of the capacity of an imaging
system to resolve adjacent high contrast objects (vessels).
Spatial resolution can be expressed quantitatively in terms of
the modulation transfer function (MTF).
Measurement of the MTF
is complex; therefore, in this report the spatial resolution is
expressed as the visual cut-off frequency of a standard bar
phantom and is specified in line pairs per millimeter (lp/mm).
The cut-off frequency of the subtracted image will depend on the
product of the MTF's of the individual components in the imaging
The factors which have a major influence on spatial
resolution are:
Image intensifier
Geometric magnification
Focal spot size
Display matrix size
TV chain
Items a, b, c, and e affect the amount of image blurring, whereas
d and e affect the sampling rate.
a) Image Intensifier: Current cesium-iodide image intensifiers
have cut-off frequencies of 4 to 5 lp/mm. This generally exceeds
the spatial resolution limits of present DSA systems. Therefore,
The active input
the II is usually not a limiting factor.
phosphor size is inversely related to the spatial resolution and
It is important to
is one parameter which may be selectable.
realize that the active input-phosphor size is almost always less
than the manufacturer's stated nominal size, and it should be
It is
common for II's to
magnification, which changes the active input phosphor size
For a fixed display matrix, the pixel
(i.e., dual-mode 9"/6").
size will increase with increasing active input phosphor size,
thus resulting in reduced spatial resolution.
Geometric Magnification: In all DSA systems, considerable
geometric magnification is present.
The size of the object
presented to the input-phosphor is enlarged by the magnification
factor. Hence the effective resolution of the imaging system
increases by the same magnification factor, at the expense of a
After a
decreased field of view and increased patient exposure.
certain point, increasing the magnification will decrease the
resolution due to
unsharpness resulting from the finite focal spot size.
c) Focal Spot Size: The spatial resolution limit is inversely
The choice of a focal spot size
related to the focal spot size.
in pulsed-mode DSA involves a trade-off between the increased
geometric unsharpness resulting from a larger focal spot, and the
decreased patient motion blurring resulting from the higher
and correspondingly shorter
The measured size of the
associated with larger focal spots.
focal spot may be 30-50% greater than the nominal size, and it
Although the intensity
may be even larger at high mA's.
distribution is never uniform, one can calculate the resolution
limit from a uniform square focal spot of known dimension:
R = spatial resolution limit (lp/mm)
F = measured focal spot size in mm
= geometric magnification
Figure 3 shows the effect of focal spot size and magnification on
the spatial resolution.
A magnification of at least 1.25 is
almost always present due to the geometry of the equipment.
Larger geometric magnifications may be used in practice.
Fig. 3
Measured Focal Spot Size (mm)
d) Display Matrix Size: The spatial resolution can never exceed
the limit imposed by the pixel size. The pixel size depends on
the display matrix (number of pixels) and on the diameter of the
image in the patient plane (geometric magnification and active
input-phosphor size).
Typical display matrices used in DSA are
256 x 256, 512 x 512 and 1024 x 1024. The theoretical limit for
spatial resolution due only to the display matrix is given by
= resolution limit (lp/mm)
= matrix size (ie, 256, 512, 1024)
= geometric magnification
= II active input-phosphor size or diameter (mm)
converts pixels/mm to lp/mm
4 shows the effect of active input-phosphor size,
geometric magnification, and display matrix size on spatial
Fig. 4
e) TV Chain: The spatial resolution of the TV chain is affected
by the electronic focusing of and the size of the electron beam
in the TV camera, the optical focusing of the TV camera lens to
the II output plane, and by factors which are different for the
The vertical resolution is
horizontal and vertical directions.
affected primarily by the number of active scanning lines and by
The Kell factor is an empirical factor which
the Kell factor.
expresses the fraction of the active scanning lines that are
effective in preserving detail in the vertical direction. It
has been determined experimentally to be about 0.7 for most
television systems. The nominal number of scanning lines used in
the U.S. television standard is 525, but systems with more lines
Usually only
(eg., 875, 1023) are common in medical imaging.
about 480 (or 91%) of these lines are actually used in the
formation of the image; the remainder are used for display
Thus, the vertical resolution is proportional to
(nominal number of TV lines) x (91%) x (Kell factor), up to the
limit imposed by the TV camera beam size and focusing.
There are two methods of reading out the target or faceplate of
the TV camera: the interlaced mode is the conventional method,
while progressive scan readout is sometimes used in DSA systems.
In the interlaced mode, one 525-line TV frame is read out in two
passes over the target.
The first pass scans only the oddnumbered lines and the second pass scans only the even-numbered
Each pass is called a field and consists of 262-l/2
A progressive scan system reads out the entire target
with one pass comprised of 525 lines.
As each system uses the
same number of scanning lines,
it might appear that the spatial
resolution would be equal. However, in a pulsed-mode DSA system,
the image is stored on the target of the TV camera. If the pulse
rate is 30 per second or less, and the image is read out in the
interlaced mode, most of the signal (70-80%) is read out during
the first field.
This would effectively reduce the 525-line
system to a 262-line system, hence decreasing the resolution.
For systems operating in the fluoroscopic or frame-averaging
mode, the image on the camera target is continually being
refreshed between field readouts by new information from the
image intensifier, and there is no loss of spatial resolution in
the interlaced mode (for a stationary image).
The TV spatial resolution limit in the vertical direction is
determined by the effective number of scan lines per mm in the
patient plane, which depends on the geometric magnification and
the active input-phosphor size:
where R (vertical)
= resolution in vertical direction (lp/mm)
= nominal number of TV scan lines (i.e.,
525, 1023)
= fraction of TV scan lines active in
image formation (~.91)
= Kell factor
= geometric magnification
= active input-phosphor size (mm)
converts lines/mm to lp/mm
Figure 5 demonstrates the effects of number of scan lines,
geometric magnification and input phosphor size on spatial
Fig. 5
The spatial resolution limit in the horizontal direction is
determined primarily by the number of scan lines per video frame
and by the bandwidth of the video amplifier (i.e., how many times
per scan line the video signal amplitude can be changed), up to
the limit imposed by the beam size and focusing of the TV camera.
In conventional TV systems, the bandwidth is usually set just
high enough to provide equal horizontal and vertical resolution.
However, in DSA systems, the bandwidth is often reduced by
electronic filtering so that sampling artifacts are avoided and
reduction of electronic noise,
which increases
In general, the sampling rate of the ADC should be
Commonly used ADC's
about twice the bandwidth of the TV chain.
have sampling rates of 10-15 MHz, which limits the TV bandwidth
to about 5 MHz.
This may be sufficient to give approximately
equal vertical and horizontal spatial resolution for a 525-line
The TV target in a
TV system, but not for a 1023-line system.
1023-line system would have to be read out with a slow horizontal
sweep (slow scan TV) in order to have a horizontal spatial
resolution that is even equivalent to a standard 525-line TV
because of the bandwidth limitation imposed by the ADC sampling
rate. Thus, it is important that the vertical and the horizontal
spatial resolution be evaluated separately in DSA systems, since
they may be quite different.
2) Low-Contrast Performance
spatial resolution in
angiography is generally inferior to that of conventional film
radiography, DSA is useful because its low-contrast sensitivity
is superior due to its capacity to enhance contrast through
better subtraction,
DSA is
display windowing and centering.
thus capable of imaging with smaller concentrations of iodine
in the vessels under study.
It is this low-contrast performance
that has made possible angiography via intravenous injection of
contrast material and the use of lower, less hazardous amounts of
contrast material in intra-arterial angiography.
Ideally, the
evaluation of the low contrast performance should be closely
related to the clinical imaging task of observing patency,
ulceration, aneurysms and calcification of blood
vessels. These tasks are difficult to simulate and quantify with
phantoms; however, all involve the ability to image an iodinecontrast-filled vessel well enough for detection of abnormalities
in the vessel walls.
Thus, a reasonable measure of low-contrast
performance is the smallest cylindrical vessel that can be imaged
with good visualization of the vessel edges.
The test vessel
should have an iodine concentration similar to
that normally
encountered in a patient's vessels during angiography.
The low-contrast resolution depends on the contrast signal
presented to the image intensifier (II) and the capacity of the
system to detect, preserve and display this signal. The contrast
signal from the patient depends on the iodine concentration,
vessel size, and beam quality. The signal presented to the II is
depends on
degraded by the scatter reaching the II, which
magnification, the contrast improvement factor of the grid, beam
field size, and patient thickness.
The contrast
detected by the II is generally degraded by the inability of the
components of the imaging, digitizing, subtraction, storage and
display chain to preserve this contrast. In addition, the
presence of noise in the signal will degrade the perceptibility
of low contrast.
The low-contrast performance of the DSA system
image is ultimately determined by the signal-to-noise ratio (SNR)
of the system.
Two kinds of noise are present in the system, fixed pattern noise
and random process noise.
Fixed pattern noise is usually due to
manufacturing variations in individual system components such as
II input-screen non-uniformities, granularity of the II outputphosphor, TV camera target imperfections, and ADC step size
In the subtracted difference image, these are
However, when
normally subtracted and do not degrade the image.
reregistration of the mask is carried out to correct for patient
motion, the fixed pattern noise will appear in the difference
Noise resulting from random processes fluctuates from
frame to frame and from pixel to pixel, and thus limits the lowcontrast detectability in the difference image.
These random
processes are due to statistical fluctuations in the detected xray fluence (quantum noise) and to the electronic noise in the
imaging chain.
The major sources of electronic noise are TV
camera electronic noise, TV camera time jitter, ADC quantization
noise, and electronic noise and time jitter in analog storage
The difference image SNR is defined as the ratio of the contrast
signal for an opacified object in the subtraction image to the
RMS signal fluctuation in the neighboring background region. The
contrast signal to be used in the calculation of the SNR is the
signal from an opacified vessel less the neighboring background
In DSA units with a region-ofsignal in the subtraction image.
interest (ROI) function, the contrast signal can be found if one
places the ROI over an opacified vessel, obtains the average
pixel value, and subtracts from this the average pixel value when
the ROI is placed over the neighboring background.
The noise to
be used in the calculation may be the pixel standard deviation in
the background region. This difference image SNR, along with the
vessel or object size, determines the detectability of the vessel
or object, and hence the low-contrast performance.
In any evaluation of the low-contrast performance of a DSA
system, it is important that one understand the role of the
operator-adjustable parameters, and that these parameters are set
typical clinical values after they have been properly
optimized. (See Section III) The operator-adjustable parameters
which have the greatest influence on image quality are geometric
pixel size, beam quality, and radiation exposure
at the II input.
It is most important when comparing the lowcontrast performance of two DSA systems to use exactly the same
operator-adjustable parameters in both systems.
In general,
the difference image SNR is
a) Beam
optimized for a highly filtered beam in the range of 50-60 kVp
for thin body parts (IO cm), 60-70 kVp for medium body parts (20
cm), and 65-75 kVp for thick body parts. These guidelines should
be used for
evaluation with phantoms with the understanding
that the kVp may have to be raised due to mA limitations on
radiographic equipment.
b) Geometric Magnification: Within the limitation imposed by
focal spot blurring, geometric magnification will improve
detection of low contrast objects, so long as the II input
In any evaluation of low-contrast
exposure is held constant.
performance, one should use a typical magnification encountered
in clinical studies, such as 1.25.
Pixel size depends on the matrix size, II active
c) -Pixel Size:
input phosphor size, and geometric magnification. Decreasing the
pixel size decreases the resultant SNR when the radiation
exposure to the II is low and the camera video output signal is
relatively high.
d) Entrance Exposure to
The two major contributions to
- the
- II:
the noise in DSA are quantum noise and TV camera noise. Above a
certain level of II input exposure, the quantum noise is low and
low-contrast resolution is dominated by TV noise.
In this
region, the performance is relatively independent of patient or
II exposure, and increased exposure will not result in increased
image quality.
This is the region of needlessly high patient
In the low II input-exposure range, the x-ray photon quantum
noise dominates the TV camera noise, and the low-contrast
performance can be described approximately by the following
where [µ / ρ ]I= effective iodine mass attenuation
coefficient (cm2/mg)
tI = projected iodine thickness (mg/cm 2)
= vessel diameter (cm)
at input of the II
This expression simply indicates that, for a given iodine
contrast level, the smallest detectable diameter of a vessel
varies inversely as the square root of the exposure. Conversely,
it indicates that to see the same vessel at half the iodine
This relationship
concentration requires 4 times the exposure.
can be plotted in a series of contrast vs. vessel diameter curves
There will be a
called contrast-detail-dose (CDD) curves.
separate curve for each exposure up to the point where TV camera
These CDD curves characterize the lownoise becomes dominant.
contrast performance of a DSA system and can guide the selection
of x-ray techniques.
3) Contrast and Spatial Uniformity:
Generally, DSA imaging systems use logarithmic processing of the
video signal to insure that a vessel of uniform diameter and
uniform iodine concentration appears in the subtracted image with
uniform diameter and contrast, regardless of overlying structures
such as bone, air and varying tissue thickness. The logarithmic
processing increases the dynamic range of the unsubtracted video
signals which the system can detect,
and hence is usually
necessary when a large range of x-ray attenuation is present in
The processing must be adjusted properly to insure a
the image.
uniform contrast signal in a vessel crossing areas of drastically
such as regions
structures. The contrast uniformity is degraded in a less
noticeable way by a nonuniform distribution of radiation scatter
and veiling glare in the video image.
Good spatial uniformity means a uniform magnification factor over
the entire field of the II.
This is seldom achieved due to the
curved input surface of the II and nonlinearities in the electron
optics in the II and TV chain, and in the optical lenses in the
imaging chain. It most often appears as "pin-cushion" distortion
which is normally "edited out" by the viewing radiologist, but
may become important if quantitative measurements are to be made
on the image.
4) Contrast Linearity:
A second benefit of logarithmic processing is that the difference
contrast signal is directly proportional to iodine thickness, and
independent of transmitted x-ray fluence. This allows DSA images
to be used for the quantitative measurement of physiological
quantities related to a patient's cardiovascular system.
To see this, consider the video signal levels in a subtraction of
a mask from an opacified image using logarithmic processing:
= log ( a N),
= mask image signal
= number of detected photons per pixel in
the mask
a conversion factor mapping detected x-ray
fluence to video-signal level
= log [ α N exp(l-[µ / ρ ]ItI)]
= log [ α N] - [µ / ρ ]ItI
[µ / ρ ]I
signal of image containing iodine contrast
= effective iodine attenuation coefficient
= projected iodine thickness (mg/cm 2).
Thus, the difference signal is
= M-C
= [µ / r ]ItI ,
which is proportional to iodine thickness and independent of the
transmitted x-ray fluence N.
The contrast linearity is dependent on the proper adjustment of
the logarithmic processor, the luminance linearity of the image
intensifier and TV camera, and the linearity of the ADC.
It is
thus important that the II input exposure not be so high that its
luminence becomes nonlinear with input. The effective f-stop of
the TV camera must also be carefully chosen to maintain video
signal linearity.
When these conditions are satisfied, the
contrast linearity should be good for a localized small iodine
signal. For large opacified structures, such as the cardiac
chambers or the aortic arch, nonlinearity often occurs as the
result of varying radiation scatter,
veiling glare, and large
variations in iodine thickness.
In the presence of such
nonlinearities, the difference signal can be approximated by:
scatter radiation ratio
veiling glare ratio.
In order to compare DSA systems and to verify specifications, one
This is
must understand what is meant by "iodine contrast."
variously referred to as percent contrast, projected iodine
thickness (mg/cm2), or iodine concentration (mg/cm3). Each of
these has a different meaning and use:
Iodine Concentration (mg/cm 3) This is the concentration of
iodine in a vessel and is the quantity of interest in
It is also the easiest way to
quantitative flow studies.
specify the amount of iodine to be placed in a vessel
phantom since this quantity is independent of beam quality,
phantom thickness, and vessel shape and size.
concentration forms the link between simulated vessels and
patient studies, since the iodine concentrations in patient
vessels from intravenous injections are at the same levels
as the iodine concentrations in the phantom vessel insert.
The iodine concentration depends on the type and dose of the
contrast agent, the injection technique, and the dilution
curve for the patient and the vessel site under study.
Projected Iodine Thickness (mg/cm ): This is the total
thickness of iodine seen by the x-ray beam in any given
It is dependent not only on the concentration of
iodine in a vessel, but also on the thickness and
of the vessel.
It is independent of x-ray beam quality and
phantom thickness. For example, a round vessel with a given
value at
iodine concentration will have a smaller mg/cm 2
This is the
the edges of the vessel than at its center.
quantity proportional to the numerical pixel values in a
subtracted image when logarithmic processing is used.
This is a traditional term from
Percent Object Contrast:
radiography which refers to the percentage attenuation of
The percent object
primary x-rays by the iodine present.
contrast is dependent on the iodine concentration, the
vessel thickness, and the x-ray beam quality within the
pixel of interest. Percent object contrast is a useful term
for comparison of the low contrast performance of DSA
systems to other radiographic
systems, but
thickness is
more useful quantity
specification of low-contrast performance because it is
directly related to the clinical iodine contrast
techniques and is independent of beam quality.
This is the iodine contrast signal
Percent Image Contrast:
in the subtracted image and is defined as the mask image
signal for a certain pixel minus the corresponding signal
In a perfectly linear
from the image containing iodine.
imaging system with no scatter, the percent image contrast
In reality
would be equal to the percent object contrast.
image contrast is reduced by the presence of
scatter and veiling glare.
Like object contrast, image
contrast is dependent on iodine concentration,
thickness, and x-ray beam quality.
Radiation Exposure:
The radiation exposure produced in a DSA examination will affect
the quality of the images produced, as well as the magnitude of
the potential radiation risk to the patient and to the personnel
examination. Several
parameters are associated with the DSA procedure:
Patient Entrance Exposure: A measurement of the exposure
produced in the plane at
which the beam enters the patient
will be necessary for comparison with other procedures, or for
determination of the dose to specific organs. Since the exposure
is normally different for different settings of the II field
size, it is important to make these measurements at all the II
field sizes used.
b) Exposure at Image Intensifier Input Phosphor: A major factor
determining the quality of the final image is the x-ray exposure
(or exposure rate) reaching the input surface of the II. If the
exposure (or exposure rate) is low, the system noise will be
dominated by the x-ray quantum noise rather than by the system
electronic noise.
The exposure at the II input phosphor will
therefore play an important role in determining the low-contrast
performance and hence the CDD curve for the system.
In order to
compare the image performance of one system to that of another,
the CDD curves must be labelled with the exposure (or exposure
rate) at the II input-phosphor.
Continuous Mode: When the system is run continuously,
as in fluoroscopy for localization or in fluoroscopy-mode
subtraction, a measurement of the exposure rate at the II
input surface will provide the fluoroscopic input exposure
rate sensitivity (FIERS).
The FIERS depends on the f-stop
of the TV camera lens, the TV camera target voltage, and the
control settings for the
Typical values are 30-600 uR/s for the 6" mode
and 15-300 UR/S for the 9" mode.
ii) Pulsed or Cine Mode: The image intensifier input
exposure sensitivity (IIIES), typically expressed in units
of uR/frame, is the parameter which determines system
information content when the system is run in a pulsed x-ray
Typical IIIES values vary from 100 to 1000 uR/frame
and normally are inversely proportional to the active inputphosphor area for II's of different size.
c) Personnel Exposure: Exposure to personnel will normally come
from radiation scattered by the patient to the side of the x-ray
table, and from the leakage radiation fron the x-ray tube
The magnitude of this hazard will depend upon the
technique factors, beam quality, field size, intensifier size and
location of drapes and shields, and the position of
personnel in the room. The location of the x-ray tube strongly
influences personnel exposure.
Exposure to the operator is
generally much less when the x-ray tube is under the table,
Iso-exposure curves for various
rather than over the table.
conditions should be measured.
Typical examples are given in
AAPM Report No. 12, "Evaluation of Radiation Exposure Levels".
6) Subtraction Artifacts
If the pixel location of a high contrast background object such
as bone, air, bowel gas, or metal should move during mask-mode
subtraction, the result of this misregistration between the mask
and the opacified image will be either the introduction of an
artifact or incomplete removal of overlying structures. This, in
turn, may obscure the visibility of an opacified vessel.
Misregistration can be caused by imaging chain deficiencies or by
motion of the patient,
x-ray tube, or II.
artifacts can be due to TV camera
instabilities, II power supply fluctuation, and time base jitter.
Artifacts from these sources reflect improper functioning of the
equipment and should not be tolerated.
Patient motion is the primary cause of misregistration artifacts
and of unsatisfactory studies.
Motion artifacts can often be
decreased by remasking, a process in which one of the frames
taken after the motion occurred is used as the new mask and
subtracted from subsequent images.
A second technique is
reregistration, in which the two images to be subtracted are
shifted with respect to each other and sometimes rotated. It
appears that shifting by a fraction of a pixel is necessary, and
that the whole image cannot be perfectly subtracted,
but that
one can optimize the subtraction in a vessel area of interest.
Incomplete subtraction can occur due to instability of the x-ray
tube output, which might fluctuate by several percent during the
However, because of the logarithmic
imaging sequence mode.
processing of
the video image before
fluctuation is turned into a uniform brightness change from one
image to another.
Although this brightness change obviously
needs correction for quantitative use of the image data, such as
in blood flow analysis, it does not affect the visualization of
On the other hand,
the opacified vessels in subtraction images.
excessive brightness change in an image sequence indicates
possible malfunction of the synchronization of x-ray pulsing to
image acquisition, the television camera, or the x-ray tube
The performance evaluation method to be outlined in this section
is non-invasive and involves the use of a patient-simulating
Also needed will be standard fluoroscopic resolution
test devices and a dosimeter, whose characteristics are described
in the exposure measurement section.
A. Typical Phantom
Figure 6 illustrates a possible phantom design which would be
acceptable for performance evaluation.
The basic attenuation
phantom could be made of lucite or tissue-equivalent plastic and
should allow assembly or stacking in such a way that the
thickness of any part of the body could be simulated.
should be a step-wedge section which has an x-ray attenuation
dynamic range greater than 15:l to simulate the large range of
signals encountered in DSA examinations. There should be a bone
section which further simulates extremes in dynamic range, the
high contrast overlying structures in actual patients, and beam
hardening effects. The bone could be actual bone or bone
equivalent plastic, but it should be very similar to bone in
attenuation and scatter characteristics for use in systems that
are capable of dual energy subtraction.
The phantom should have a slot somewhere near mid-plane where the
resolution inserts can be placed with a magnification simulating
clinical conditions. Figure 7 shows the blank insert which is to
be used for mask images (pre-contrast). It is important that this
blank insert be made of the same material and be the same
thickness as the resolution inserts that will be replacing it in
this slot.
Figure 8 illustrates a possible vessel insert with simulated
vessels of 0.5, 1, 2, and 4 mm diameter. These vessels should be
filled with iodinated epoxy, with a different concentration for
each set of vessels. Concentrations of 10 mg/cm 3, 5 mg/cm3, and
2.5 mg/cm3, of iodine simulate typical clinical conditions for
intravenous injection. If intraarterial injection techniques are
used, an insert of higher concentration, such as 60 mg/cm 3, would
be useful to assure proper operation of the unit under such
The vessel insert would be used for the evaluation
of low-contrast performance and uniformity. It is desirable that
the simulated vessels be circular (or semi-circular) in cross
section to provide the same difficulty of clearly imaging vessel
edges that occurs with actual blood vessels.
Although the vessel insert described is necessary to simulate the
clinical situation, particularly for quantitative analysis and
uniformity, a line-pair type of insert as seen in Figure 9 may be
A line-pair target test
more useful for optimizing the system.
observer-dependent than determining the
perceptible single vessel with indistinct edges. The line-pair
insert has channels of rectangular cross section to give high
contrast edges.
All channels (1, 0.7, 0.5, 0.35, 0.25, 0.175,
0.125 lp/mm) have the same 1 mm depth; therefore, the contrast
(mg/cm ) is constant for all sizes, as opposed to the vessel
insert, whose contrast decreases as the vessel size decreases.
with iodinated epoxy at
concentration of 5 mg/cm3 results in an iodine thickness of 0.5
Similarly, a set with 10 mg/cm iodine concentration
mg/cm .
would give an iodine thickness of 1.0 mg/cm2.
Figure 10 shows a misregistration insert. It is a standard l/16"
(1.6 mm) thick perforated aluminum plate with l/8u (3.2 mm)
diameter holes which provide many high contrast edges.
It contains six
Figure 11 illustrates a linearity insert.
0.5, 1, 2, 4, 10, and 20
regions of different iodine thickness;
It should be made of the same material as the blank
manufacture of iodinated
The typical mixture of
skill and practice.
iodobenzene and casting resin requires very careful measurement
of small amounts of iodine to insure accuracy, careful mixing to
insure uniformity, and vacuum curing to eliminate air bubbles.
There is also a question about the possible diffusion of iodine
out of the channels over time.
Figure 6. Attenuation Phantom. Stepwedge Sections I and II
Can Be Combined To Form A Uniform Thickness Of 3 Inches.
Figure 7.
Figure a.
Blank Insert.
Vessel Insert.
Groove Width 4.0
0.7 0.5(mm)
One insert is filled with 10 mg/cm 3 of Iodine.
A second insert is filled with 5 mg/cm 3 of Iodine.
Fig. 9.
Low Contrast Line Pair Insert
(Use of l/8" perforated aluminum plate is acceptable)
Figure 10.
Misregistration Insert.
Iodine Concentration of the Discs:
0.5 mg/cm
1.0 mg/cm
2.0 mg/cm
4.0 mg/cm
(E) 10.0 mg/cm
(F) 20.0 mg/cm
Figure 11.
Linearity Insert.
B. Spatial Resolution
1) Method: The spatial resolution should be determined at the
location where the radiologist will view the image.
This provides a measurement of
this is at the display console.
the overall performance of the system rather than of the
It may also be necessary to
operation of individual components.
check the performance of these components to diagnose problems or
to optimize the system performance.
The spatial resolution may be different in the horizontal and
vertical directions due to the bandwith and the number of TV
lines, therefore it is recommended that both horizontal and
vertical spatial resolution be determined, as well as
This can be done by placement of a standard
resolution at 45°.
lead bar line-pair pattern parallel, perpendicular, and at 45° to
the TV raster lines.
The pattern should go up to 5 lp/mm, and
If this
the lead thickness should be no greater than 0.1 mm.
cause3 dynamic range problems, a thinner lead pattern (0.01 mm)
or a wire mesh pattern might be used.
The three patterns should
be placed in a 15 cm (6 inch) thick phantom to simulate a
It is recommended that a magnification of 1.25 be used
if this is consistent with the system geometry.
The spatial
resolution should be determined by viewing of both unsubtracted
and subtracted images.
It will be necessary to remove the
pattern3 for the mask image.
2) Evaluation:
The spatial resolution cut-off frequency is a
complex function of the II active input-phosphor size, geometric
magnification, focal spot size, number of TV scan lines and
Table I summarizes the theoretical cut-off
display matrix size.
frequencies for a number of combinations of the variables.
should be understood that the cut-off frequency listed in the
table is that of the component in the imaging chain which ha3 the
poorest cut-off frequency ("weakest link") and ignore3 the
degrading influence of the other components.
Thus, in practice,
the measured cut-off frequency may
be slightly below these
Interpolation for combinations not shown can be made
with the formula3 given in Section II-B1.
Gross deviation3 from the value3 in the table, or degradation
over time, should be investigated by isolation of the influence
of the various component3 and variables.
Some are easily
isolated, but others require additional test equipment.
size and geometric magnification influence3
can be
eliminated if the test pattern3 are placed against the II input
surface, in which case the magnification would be very close to
To eliminate the loss of contrast due to scatter and to
increase the contrast of the test pattern, one can remove the
phantom and use a very low kVp.
To isolate the II, one can view
II =
Nominal image intensifier size (inches); calculation
assumes active II size = 0.90 nominal size
Display matrix size
Geometric magnification
Actual focal spot size (mm)
Nominal TV scan lines; calculation assumes a=0.91, K=0.7
Theoretical spatial resolution cut-off frequency
the output phosphor of the II directly by using a telescope
focused-at infinity. It may be necessary to remove the TV camera
for this.
The adjustment of the II electronic focusing is best
made in this manner. AAPM Report #4 provides more details on the
TV electronic
For optimization of the
evaluation of an II.
test pattern3 should be placed against
and optical focusing,
the II input surface with no phantom in place, a low kVp used,
AAPM Report
and adjustment3 should be made during fluoroscopy.
#4 provide3 additional details.
The subtracted image should be recorded on the video hard copy
imager and the spatial resolution compared with that obtained on
the display monitor.
C. Low Contrast Performance:
1) Method: The attenuation phantom should be assembled with the
blank insert to give a uniform (no step wedge) thickness of 15 cm
(6 inches) and should be set up in a geometry similar to that
used for patient studies.
The x-ray tube potential is to be set
at 70 kVp.
Scout pulse3 are taken and the mAs and/or camera
aperture adjusted until they are within the manufacturer's
recommended operating range.
The mask image is obtained. Next
the vessel insert (or low-contrast iodine line pair insert) is
put into the phantom in place of the blank, and subtracted images
are obtained without adjustment of any parameters. The kVp, mAs,
exposure time, focal spot size, focal spot-to-object distance,
focal spot-to-image receptor distance, II active input-phosphor
aperture setting,
phantom thickness, and all other
pertinent variable3 are recorded. The test may be repeated for
other phantom thicknesses for simulation of the type of studies
being done at a particular facility. It is useful to measure the
patient and/or II input-exposure at the same time (See section
III.F, Radiation Exposure).
NOTE: Units that operate only in the fluoroscopic mode cannot be
evaluated with the above method; the iodine thickness must be
changing to be seen.
Motorized vessel insert3 would be needed
accurate quantitative evaluation of
performance of these DSA systems. However, it may be possible to
image vessels by manually moving the vessel insert at an
appropriate speed perpendicular to the direction of the vessels.
The smallest-diameter vessel (or finest linepair) seen for each iodine concentration is determined by
optimizing of the display window and level controls.
information can then be compared with manufacturer or purchase
document specifications, can be used for comparison of two DSA
systems, used for optimization of component3 and operating
variables, or recorded periodically for quality
CDD curve3 may be drawn which correlate the data.
The optimization of the imaging variables, such a3 kVp, mAs, beam
filtration, aperture setting, frame averaging, and magnification,
is best accomplished with the use of a phantom.
important is the optimization of the input exposure to the image
In general, raising the radiation exposure will
allow smaller vessel3 with lower concentrations of iodine to be
seen, because the quantum noise will be reduced (the TV camera
aperture may have to be reduced for increased exposure).
point at which increasing the exposure no longer reduces the
noise (and no longer increase3 vessel visibility) is the point at
which the TV camera noise dominates.
Radiation exposures above
this point should never be used.
The optimum exposure level may not be the same for different
clinical procedures which involve different vessel sizes and
Each clinical situation should
different iodine concentrations.
be simulated with the phantom (proper phantom thickness) and the
input exposure (IIIES) determined for the desired
Evaluation should also be made with and without
a grid to determine if a grid is necessary for all procedures.
Without a grid both patient and personnel exposures are greatly
D.1 Contrast Uniformity:
1) Method: The phantom is used with the step wedge in place and
a subtracted image of the vessel insert is obtained following the
procedure outlined in Section III-C.
The logarithmic amplifier
must be on. The direction of the vessel3 should be perpendicular
to the steps. The test should also be done with the bone section
added to the step wedge.
2) Evaluation: The vessels should appear uniform in density and
There will be
width a3 they pass under the steps and bones.
increased noise for those portions of the vessel3 that pa33 under
the thickest portion3 of the phantom, due to added photon
attenuation. Some steps may disappear due to the limited dynamic
If the uniformity of density and width is not adequate,
particularly for the highest contrast vessels, some adjustment of
the logarthmic processor may be needed. The service engineer
should be consulted.
This test also measure3 the dynamic range of the system.
number of steps which can be seen (by the presence of noise
changes) is related directly to the dynamic range.
A system
which saturates and cannot see all six steps will most likely
require bolus material around the patient, or a shaped filter at
the collimator.
Spatial Uniformity:
A subtracted image of the vessel insert is obtained
1) Method:
with a thin uniform phantom (no step wedge).
The difference between any two vessels is
measured at the center and edges of the image with the electronic
calipers, or with a ruler on the hard copy. This is a measure of
the amount of distortion, which should be taken into account when
size measurements are performed.
E. Contrast Linearity:
A subtracted image is obtained with the linearity
1) Method:
insert and the uniform phantom, following the methods of Section
The average pixel value is obtained for each of the
For systems with quantitative analysis
linearity sections.
For systems not having this,
packages the ROI function is used.
a narrow window can be set and the level setting scrolled until
the mean pixel value is found for each section.
The pixel values are plotted as a function of
the iodine thickness (mg/cm2) in each section.
They should fall
on a reasonably straight line. If not, the logarithmic processor
may not be properly adjusted. The user should be aware that the
logarithmic processor is often adjusted to compensate for the
effects of scatter and veiling glare for a certain phantom
thickness and kVp, and it may not be possible to achieve precise
linearity for all clinical situations.
The service engineer
should be consulted.
This method also calibrates the contrast
scale of the unit; this is useful for quantitative flow studies
and for comparison of different DSA systems.
Radiation Exposure:
1) Method:
Instrumentation: Three different dosimeters may be required
for these measurements.
exposure - The ionization
associated electrometer should be capable of measuring
exposure rates up to 200 R/min or an accumulated exposure of
50 R per run. The energy response should be calibrated over
a range of beam qualities from 1.5 mm Al to 7.0 mm Al HVL.
IIIES or FIERS - The ion chamber system should be
capable of measuring exposure rates as low as 0.1 mR/min or
an accumulated exposure as low as 0.1 mR per run.
chamber should be small enough to be contained within the
field of view of the intensifier and relatively thin so that
it can be located near the input surface of the II.
Personnel exposure - An ionization chamber survey
meter capable of reading scatter exposure rates up to 1
R/hr should normally be used.
Phantom Thickness:
At least two phantom thicknesses should
be employed for these measurements;
a 10 cm (4 inch) thick
section to simulate the neck or cervical spine and an 18 cm (7
inch) thick section to simulate the abdomen. This thicker section
require technique factors which
appropriate for cerebral examinations.
c) Experimental Arrangement: The phantom is placed on the table
top and centered in the x-ray beam.
The II input surface is
positioned 30 cm above the table top.
For C-arm units, the
target-intensifier distance is adjusted to 76 cm (30 inches), if
possible. If the phantom does not take up the full field of view
of the II, the x-ray field is collimated to within the area of
the phantom. The field size is recorded.
i) FIERS or IIIES measurements are made with the phantom in
The ion chamber is to be supported at the input
surface of the II between the anti-scatter grid and the II
input surface.
For positioning the ion chamber, the antiscatter grid will have to be removed temporarily.
For image intensifier systems in which the anti-scatter grid
can not be easily removed, removal of the grid MUST be
performed by a qualified person.
This is especially
important when the grid is a built-in type design or is an
integral part of the II housing.
The II tube is both an
implosion hazard and very expensive to replace if it is
damaged. The removal of this type of grid housing may cause
room light to affect the II and the operation of an
automatic exposure system.
For patient entrance exposure measurements, the ion
chamber is placed on the table top (or at 30 cm from the II
input-surface for C-arm units) centered in the beam.
anti-scatter grid should be in place on the II, if the unit
is used this way clinically.
The unit is run at typical clinical technique factors for these
measurements. If typical factors are not available, the unit is
set for 70 kVp and the mAs is adjusted according to the DSA
manufacturer's instructions for the phantom thickness used; or
the automatic exposure technique factor control circuitry is used
for determination of the technique factors.
It is generally desirable to perform measurements with the
Exposure times
dosimetry system operated in the rate mode.
should be long enough to permit automatic brightness control (if
any) of the x-ray system and the ionization chamber measurement
system to reach stable values. The exposure/frame is obtained by
division of the measured exposure rate (R/min) by the known frame
The exposure measurement mode may be
rate ((f/s) X 60).
appropriate for those systems which produce clearly separated
individual exposures at rates less than 1 frame/sec.
exposure per frame is then just the total exposure measured,
divided by the total number of frames. The particular model of
dosimeter used should be carefully evaluated as to whether it
gives correct readings for pulsed beams in the readout mode being
If the grid cannot be removed, the exposure rate into the grid
must be divided by the grid transmission factor. distance
corrections may also be necessary if the ion chamber is not
located close to the II input surface.
2) Evaluation:
The patient entrance exposure is determined either by
multiplication of the measured entrance exposure rate in R/min by
the total time of the exposure expressed in minutes, or by
multiplication of the measured exposure in mR/f by the total
number of frames used. To obtain the total entrance exposure for
the examination, one must remember to add the exposures required
to obtain scout images, subtraction mask(s), and/or fluoroscopy
to those required for the
subtraction images.
b) The IIIES will normally be expressed as the II input exposure
in uR/f.
If several frames are averaged to yield an image, the
total exposure for the several frames should be stated.
FIERS should be given as an exposure rate in UR/S or divided by
the television frame rate of 30 f/s to provide an effective value
of uR/f.
Typical values of the IIIES and FIERS are given in
Section II-B.5.
Scatter measurements within the examination room under
typical clinical conditions will provide information necessary
for estimation of risk to personnel performing the examination,
or for determination of the need for additional shields or drapes
within the room.
G. Artifacts
1) Method:
The misregistration insert is placed in the uniform
This image is then subtracted from itself, i.e., the
same phantom is used for the mask and the live image.
method can be used in exactly the same manner for subtraction
This test should be carried out over a time equal
to or larger than a typical clinical run, as instabilities
usually increase with time.
The auto-subtraction should result in a
Any evidence
featureless image with only differences in noise.
of hole edges, whether over the entire image or in bands, usually
indicates problems such as TV camera sweep instability, time base
The amount of
or II power supply instabilities.
misregistration can often be quantized by pixel shifting until
the image disappears.
The number of pixels shifted quantifies
the extent of the problem.
A complete program of quality assurance involves (a) specifying
performance in the purchase document, (b) acceptance testing of
the performance at installation, and (c) regular monitoring of
performance after installation.
and Frequency
-Items-To Be
- Monitored This should be routinely
1) X-ray and Fluoroscopy System:
and calibrated exactly like
radiographic equipment. Results must be recorded in a log book.
(See AAPM Report #4. Ref. 53).
The performance parameters
Digital Subtraction System:
outlined in Section II should be monitored with a phantom such as
in Section III. All parameters should be evaluated on at least a
Each day a vessel insert subtraction with
six month schedule.
the step wedge phantom should be run fixed operating variables
Results are recorded in a log
and examined for consistency.
Video Image Hard Copier
The hard copy
- Film
- Processor:
film should reproduce as closely as possible the image on the
viewing monitor.
The film processor must be monitored daily as
In addition, it is advisable to
outlined in AAPM Report #4.
photograph and process a digitally stored gray wedge such as the
SMPTE test pattern (Ref. 54) each day and compare it visually to
a master in order to detect deviations in the hard copier.
density and contrast of the processed gray wedge should be
plotted at least monthly and entered in the log book.
General Review:
1. R.A. Kruger and S.J. Riederer, Basic Concepts of Digital
Subtraction Angiography, (Hall, Boston, 1984).
C.A. Mistretta, A.B. Crummy, C.M. Strother, "Digital
angiography: a perspective", Radiology 139,273-276 (1981).
3. C.A. Mistretta, "The use of a general description of the
enchancement procedures," Optical Eng. 13,134-138 (1974).
4. R. Brennecke, T.K. Brown, J. Bursch, et al., "Computerized
video-image processing with application to cardioangiographic
roentgen image series," edited by H.H. Nagel. Digital image
processing," (New York, Springer-Verlay, 1977) p.244.
Ovitt, M.P. Capp, H.D. Fischer, et al., "The
development of a digital video subtraction system for intravenous
angiography," Proc SPIE. 167,61-66 (1978).
Mistretta, T.L. Houk, et al.,
Kruger, C.A.
real time
fluoroscopy in
visualization of the cardiovascular system", Radiology 130,49-57
Basic Physics:
7. R.A. Kruger, C.A. Mistretta, S.J. Riederer, "Physical and
technical considerations of computerized fluoroscopy difference
imaging," IEEE Transactions on Nuclear Science, NS-28,206-212.
8. S.J. Rieder, A.L. Hall, G.S. Keyes, N.J. Pelc, "Contrast
sensitivity of digital fluorographic systems," Proc IEEE 1982,
Int'l. Workshop on Physics and Engineering in Medical Imaging,
IEEE Computer Society, 66-73.
Intravenous Applications:
Crummy, C.M. Strother, J.F. Sackett, et al.,
"Computerized fluoroscopy: digital subtraction for intravenous
angiocardiography and arteriography, "AJR l35,1131-1140 (1980).
10. R.C.
Christenson, T.W. Ovitt, H.D. Fisher, et al.,
angiography using digital
intravenous cervicocerebrobascular angiography," AJR 135,11451152 (1980).
11. T.F. Meaney, M.A. Weinstein, E. Buonocore, et al.,
subtraction angiography of the human cardiovascular system," AJR
l35,1153-1160 (1980).
C.A. Strother, J.F. Sackett, A.B. Crummy, et al., "Clinical
applications of computerized fluoroscopy:
the extracranial
carotid arteries, " Radiology 136,781-783 (1981).
13. W.A. Chilcote, M.T. Modic, W.A. Pavlicek, et al., "Digital
a comparative
subtraction angiography of the carotid arteries:
study of 100 patients, " Radiology 139,287-295 (1981).
E. Buonocore, T.F. Meaney, G.P. Borkowski, et al., "Digital
angiography of the abdominal aorta and
comparison with conventional aortography," Radiology
139,281-286 (1981).
15. N.J. Hillman, T.W. Ovitt, S. Nudelman, et al., "Digital
video subtraction angiography of renal vascular abnormalities,"
Radiology 139,277-280 (1981).
16. G.D. Pond, R.W. Osborne, M.P. Capp, et al.,
angiography of
procedures," AJR 138,279-281 (1982).
Enzmann, W.R.
Brody, S.J. Riederer, et al.,
17. D.A.
intravenous digital subtraction
Neuroradiology 23,241-251 (1982).
18. F.H. Burbank, W.R. Brody, A. Hall, G. Keyes. "A quantitative
in vivo comparison of six contrast agents by digital subtraction
angiography," Invest Radio1 17,610-616 (1982).
G.B.J. Mancini, D.R. Ostrander, R.A. Slutsky, et al.,
"Intravenous vs. left ventricular injection of ionic contrast
hemodynamic implications for digital
angiography," AJR 140,425-430 (1983).
Weinstein, W. Pavlicek, et al.,
20. M.T.
Modic, M.A.
peripheral versus
"Intravenous digital subtraction angiography:
central injection of contrast material," Radiology 147,711-715
21. W.D. Foley, E.T. Stewart, J.R. Milbrath, et al., "Digital
angiography of the portal venous system," AJR
140,497-499 (1983).
"Determinants of contrast enhancement for
22. F.H. Burbank,
intravenous digital subtraction angiography (IV-DSA)," presented
at the meeting of the Association of University Radiologists,
Mobile, AL, March 22-24, 1983.
23. M.L. Wagner, E.F. Singleton, M.E. Egan. "Digital subtraction
angiography in children," AJR 140,127-133 (1983).
Weinsten, D.L. Starnes, et al.,
24. M.T.
Modic, M.A.
"Intravenous digital subtraction angiography of the intracranial
veins and dural sinuses, " Radiology 146,383-389 (1983).
25. E. Buonocore, W. Pavlicek, M.T. Modic, et al., "Anatomic and
functional imaging of congenital heart disease. with digital
subtraction angiography, " Radiology 147,647-654 (1983).
26. J.W. Ludwig, L.A.J. Verhoeven, J.J. Kersbergen, T.T.C.
Overtoom. "Digital subtraction angiography of the pulmonary
arteries for the diagnosis of pulmonary embolism,"
147,639-645 (1983).
27. B.M. Brown, D.R. Enzmann, M.L. Hopp, R.A. Castellino.
Radiology 147,655-657
subtraction laryngography,"
IA Applications:
28. C.M. Strother, M.F. Stieghorst, P.A. Tursky, et al.,
digital subtraction angiography," Proc SPIE
3l4,235-238 (1981).
A.B. Crummy, M.F. Stieghorst, P.A. Tursky, et al., "Digital
subtraction angiography: current status and use of intraarterial
injection," Radiology 145,303-307 (1982).
30. F.J. Miller, D.E. Mineau, P.R. Koehler, et al., "Clinical
using recursive
techniques injecting only small amounts of contrast material or
carbon dioxide," Submitted to Radiology.
31. M. Brant-Zawadski, R. Gould, D. Norman, et al., "Digital
subtraction cerebral angiography by intraarterial injection:
comparisonwith conventional angiography," AJNR 3,593-599 (1982).
32. M.A. Weinstein, et al., "Intraarterial digital subtraction
angiography," Radiology 147,717-724 (1983).
Energy Subtraction:
33. R.A. Kruger, C.A. Mistretta, A.B. Crummy, et al., "Digital
K-edge subtraction radiology, " Radiology 125,243-245 (1977).
34. T.L. Houk, R.A. Kruger, C.A. Mistretta, et al., "Real time
digital K-edge subtraction fluoroscopy," Invest Radio1 14,270-278
S.J. Riederer, R.A. Kruger, C.A. Mistretta, "Limitations to
iodine isolation using a dual-beam non-K-edge approach," Med Phys
8,54-61 (1981).
W.R. Brody, D.M. Cassel, F.G. Sommer, et al., "Dual energy
initial clinical experience,"
projection radiography:
137,201-205 (1981).
37. S.J. Rieder, R.A. Kruger, C.A. Mistretta, "Three beam K-edge
imaging of iodine using difference between video fluoroscopic
images: theoretical principles," Submitted to Med Phys.
38. S.J. Riederer, R.A. Kruger, C.A. Mistretta, D.L. Ergun, C.G.
Shaw, "Three beam K-edge imaging of iodine using differences
between video fluoroscopic images:
Submitted to Med Phys.
Hybrid Subtraction:
39. W.R. Brody, "Hybrid subtraction for improved arteriography,"
Radiology 141,828-831 (1981).
40. G.S. Keyes, S.J. Riederer, B.F Belanger, W.R. Brody, "Hybrid
subtraction in digital fluorography, " Proc SPIE 347,34-41 (1982).
A.B. Crummy, M.F. Stieghorst, P.A. Tursky, et al., "Digital
(DSA) with time-energy
subtraction: clinical trials," presented at the Sixty-eighth
Scientific Assembly and Annual Meeting of the Radiological
Society of North America, Chicago, Nov 27-Dec 3, 1982.
M.S. Van Lysel, J.T. Dobbins, W.W. Peppler, et al., "Hybrid
temporal-energy subtraction in digital fluoroscopy: preliminary
results, " Radiology 147,869-874 (1983).
Temporal Filtering Techniques:
43. R.A. Kruger, "A method for time domain filtering using
computerized fluoroscopy, " Med Phys 8,466-469 (1981).
Lipton, P. Mengers, R. Dahlberg,
R.G Gould, M.J.
"Investigations of a video frame averaging digital subtraction
fluoroscopic system, " Proc SPIE 314,184-191 (1981).
45. R.A. Kruger, P. Liu, W. Bateman, J. Nelson, "Time domain
angiography applications," Proc SPIE 314,319-324 (1981).
J.A. Nelson, F.J. Miller, R.A. Kruger, P-Y Liu, W. Bateman,
"Digital subtraction angiography using a temporal
filter: initial clinical results, " Radiology 145,309-313 (1982).
47. R.A. Kruger, P-Y Liu, "Digital angiography using a matched
filter." IEEE Trans Med Imag 1,16-21 (1982).
48. R.A. Kruger, F.J. Miller, J.A. Nelson, et al., "Digital
angiography using a temporal bandpass
Radiology 145,315-320
associated patient motion properties,"
49. D.R. Enzmann, W.T. Djang, S.J. Riederer, et al., "Low-dose
rate versus regular-dose low-frame-rate
subtraction angiography," Radiology 146,669-676 (1983).
50. S.J. Riederer, D.R. Enzmann, W.R. Brody, A.L. Hall, "The
application of matched filtering to contrast dose reduction in
digital subtraction angiography, " Radiology 147,853-858 (1983).
51. S.J. Riederer, W.R. Brody, D.R. Enzmann, et al., "The
filtering techniques to
application of
subtraction in DSA, " Radiology 147,859-862 (9183).
52. S.J. Riederer, D.R. Enzmann, A.L. Hall, et al.,
application of matched filtering to x-ray exposure reduction in
digital subtraction angiography: clinical results," Radiology
146,349-354 (1983).
Quality Assurance:
Basic quality
53. M. Siedband, et al., AAPM REPORT No. 4:
in diagnostic radiology,
(American Associates of
Physicist in Medicine, New York, 1978).
54. J.E. Gray, K.G. Lisk, D.H. Haddick, et al., "Test pattern
for video displays and hard-copy cameras," Radiology 154, 519-527
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF