(Radiographics. 2001;21:507-518.)
© RSNA, 2001
Quantification of Fluoroscopic Imaging System Contrast by Using Video Waveform Monitoring1
Jerome P. Taubel, RT(R)(QM),
Beth A. Schueler, PhD,
Thomas J. Vrieze, RT(R) and
Joel E. Gray, PhD
1 From the Department of Diagnostic Radiology, Mayo Clinic, 200 First St SW, Rochester, MN 55905. Presented as a scientific exhibit at the 1999 RSNA scientific assembly. Received April 27, 2000; revision requested July 21 and received August 21; accepted August 22. Address correspondence to J.P.T. (e-mail: taubel.jerome@mayo.edu).
 |
Abstract
|
|---|
A noninvasive method was developed for quantifying the overall contrast of fluoroscopic imaging systems within the clinical setting by using a simple phantom and common video test equipment. In this method, an acrylic phantom with four holes filled with varying amounts of air and aluminum is placed on the entrance exposure side of a patient-equivalent acrylic phantom. The air- and aluminum-filled holes provide a stepped gray-scale pattern that is displayed on the examination room viewing monitor when the phantom is fluoroscopically imaged under automatic brightness control. A video waveform monitor or oscilloscope is then used to quantify those video signal voltage levels as a contrast index value, which is defined as the maximum range of the video signal voltage levels of the gray-scale steps. The method is repeatable and allows quantification of the contrast of the imaging system. It can also be used to optimize video parameters, provide comparative data for quality control monitoring, and characterize overall contrast differences between systems. Experience with this method suggests that there is excellent correlation between the clinical perception of image contrast and the contrast index, with contrast index changes of approximately 15% being seen clinically.
Index Terms: Fluoroscopy • Images, enhancement • Phantoms • Video systems
|
Introduction
|
|---|
"The fluoroscopic image in room A looks flat compared with that in room B. Why?" This question may be a challenging problem to explain, let alone quantify. There are many variables that affect the overall contrast of fluoroscopic images, including beam energy and filtration; subject contrast and size; collimation and field size; veiling glare; image intensifier (II) size, type, and design; antiscatter grid type used, if any; off-focus radiation; patient-to-II distance; television (TV) camera type and pickup tube type; display monitor setup and phosphor type; and the vendor’s proprietary image processing methods. Although there are specifications for assessing the contrast response of individual imaging components (1), there are no tools available for assessing the contrast of the entire fluoroscopic imaging chain by using clinical technical parameters in a clinical setting.
This article describes a method of quantifying the overall contrast of a fluoroscopic imaging system by using a simple acrylic and aluminum phantom in combination with common video test equipment. Essentially, a phantom that produces a stepped gray-scale pattern is imaged fluoroscopically and a video waveform monitor (VWFM) or an oscilloscope is then used to quantify the video signal voltage levels of the gray-scale steps within the fluoroscopic image. Quantification by using image analysis software tools, such as those available on the x-ray equipment manufacturer’s digital system, can also be used.
|
Materials and Methods
|
|---|
The phantom consists of a 5-cm-thick acrylic slab with four contrast steps (Fig 1), which is placed on the entrance exposure side of a standard patient-equivalent acrylic flat-field phantom (Method for the Sensitometry of Medical X-ray Screen-Film Processing Systems, PH2.43-1982; ANSI, New York, NY). The ANSI phantom consists of 15 cm of acrylic and 3.0 mm of type 1100 aluminum. The combined ANSI and step phantoms are equivalent to approximately 22 cm of water and thus drive the fluoroscopic techniques into a clinically relevant kilovolt peak range and produce scatter radiation amounts that are similar to that associated with an actual patient. All subsequent references to the step phantom refer to this composite ANSI-step phantom.
View larger version (95K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1. Diagram shows top and side views of the step phantom. The step phantom consists of a 5-cm-thick acrylic slab that is designed to fit under, and be used in conjunction with, a standard American National Standards Institute (ANSI) phantom. The slab has four holes that are filled with air, aluminum, or a combination of both.
|
|
The step phantom was designed with the aid of a VWFM to produce a stepped video signal brightness range on the fluoroscopic viewing monitor that is typical of the brightness range observed under normal clinical conditions. A VWFM, like an oscilloscope, is an instrument used to measure the voltage of a video signal within a video image. The greater the video signal strength, the brighter the image that is displayed on the viewing monitor used by the physician. The time base is dependent on the specific video line rate of the signal being analyzed. VWFMs are generally designed to work with 525-line video signals that conform to the RS-170 video standard, which prescribes that the video signal be no greater than 1.0 V peak-to-peak, with 700 mV of that range dedicated to carrying image information (2). This range limitation is necessary to ensure compatibility of the video signal with conventional video equipment (eg, videocassette recorders and TV monitors). The VWFM used for this study (model 1735HD; Tektronix, Beaverton, Ore) was equipped with a manufacturer-supplied custom E-PROM (erasable programmable read-only memory) that allowed the instrument to synchronize with, and display, nonstandard video line rates that are unique to medical imaging equipment (ie, >525 lines). A general discussion of video waveform monitoring and the quality control testing of video systems can be found in reference 3.
Clinical relevance was a prime consideration in the step phantom design. The typical clinical brightness range for patient anatomy was determined by analyzing the video signal voltage levels during actual patient examinations by using the VWFM. Several different types of fluoroscopic imaging systems from different vendors were studied by using automatic brightness control. An example of the clinical validation process is as follows (Fig 2): A clinical fluoroscopic image is obtained—in this example, a typical coronary artery view (Fig 2a). A white line was artificially superimposed to indicate the position in the image of the corresponding video signal waveform (Fig 2b). A video waveform of the step phantom was then obtained with the same x-ray imaging system (Fig 2c). Note that the vertical range (y axis) of the video signal for the clinical image (Fig 2b) closely approximates that of the step phantom waveform (Fig 2c). Approximately 0.15 V was seen for the dark areas in the coronary artery image, and 0.60 V was seen for the bright areas.
View larger version (153K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2a. Demonstration of the correspondence of video signal range between a clinical image and the step phantom. (a) Fluoroscopic image shows a typical 30° right anterior oblique, 30° craniocaudal view of a coronary artery. (b) Video waveform of the clinical image. This waveform represents the video signal information that corresponds to the white line in a. (c) Video waveform of the step phantom. The locations of the collimator blades are indicated by arrows. Note that the video signal range in b and c corresponds closely. The y-axis scale is in major units of 0.10 V in these and all subsequent video waveforms.
|
|
View larger version (91K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2b. Demonstration of the correspondence of video signal range between a clinical image and the step phantom. (a) Fluoroscopic image shows a typical 30° right anterior oblique, 30° craniocaudal view of a coronary artery. (b) Video waveform of the clinical image. This waveform represents the video signal information that corresponds to the white line in a. (c) Video waveform of the step phantom. The locations of the collimator blades are indicated by arrows. Note that the video signal range in b and c corresponds closely. The y-axis scale is in major units of 0.10 V in these and all subsequent video waveforms.
|
|
View larger version (86K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2c. Demonstration of the correspondence of video signal range between a clinical image and the step phantom. (a) Fluoroscopic image shows a typical 30° right anterior oblique, 30° craniocaudal view of a coronary artery. (b) Video waveform of the clinical image. This waveform represents the video signal information that corresponds to the white line in a. (c) Video waveform of the step phantom. The locations of the collimator blades are indicated by arrows. Note that the video signal range in b and c corresponds closely. The y-axis scale is in major units of 0.10 V in these and all subsequent video waveforms.
|
|
The standard test equipment setup within an examination room (Fig 3) should duplicate the clinical configuration of source-to–image receptor distance and source-to-phantom distance for the particular fluoroscopic system evaluated. The exit surface of the phantom is placed 10 cm from the entrance surface of the II assembly, with the step portion of the phantom placed on the x-ray entrance side of the phantom. For a system with an overtable x-ray tube, the phantom is placed on the table with the step portion on top. The step phantom is centered in the desired fluoroscopic field of view (usually 20–23 cm) so that the aluminum- and air-filled holes provide the gray-scale steps oriented horizontally across the image, parallel to the video scan lines. The close central spacing of the holes in the step phantom allows all of the holes to remain within the automatic brightness control sensing window. In addition, the spacing allows slight mispositioning of the phantom in the center of the field of view without affecting the image presentation.
Next, the video signal is routed from the TV camera or image-viewing monitor to, or through, a VWFM or oscilloscope by using RG-59, RG-11, or RG-6 video cable. The proper 75-ohm termination of the video signal must be maintained (2). The amount of collimation, which should be determined during the initial acceptance test visit, is set under fluoroscopy by using a lead graticule template or observing the collimator blade locations on the video waveform.
After all other desired fluoroscopic acquisition mode parameters being tested have been selected on the x-ray imaging system, fluoroscopy is activated and the video signal brightness level of each of the steps is quantified in millivolts with the VWFM or oscilloscope. For quantification as a single contrast value, the voltage value of the first step (peak black) on the waveform is subtracted from that of the fourth step (peak white). The resultant value represents an overall contrast index. As an alternative to use of a VWFM or oscilloscope, the step brightness values can be quantified digitally by using image analysis software tools, such as those found on the x-ray equipment manufacturer’s digital system (Fig 4) or stand-alone software commonly found on image analysis workstations.
Standardization of the imaging geometry, technique parameters, fluoroscopic acquisition modes, collimation, and testing methods is critical to the reproducibility of the test. For example, selecting the fluoroscopic dose mode in a consistent manner, first high dose and then low or vice versa, may affect test results. In some fluoroscopic systems, the dose mode is changed by varying the TV camera aperture diameter, which results in a change in kilovolt peak through the automatic brightness control circuitry. Hysteresis in the aperture sizing may result in variations in the automatic kilovolt peak selection, causing the image contrast to vary. The Table lists the imaging equipment parameters that we have determined are critical to accurately reproducing the test conditions and to interpreting future waveforms. The values and settings of these parameters should be documented if at all possible.
|
Results
|
|---|
Fluoroscopic image contrast has been measured with this method for several years at our facility as part of our quality control program. Following are examples of specific uses of this method. In each of these examples, the setup parameters that do not remain constant are described; all parameters not mentioned (eg, technique, field of view) were unchanged. In particular, the kilovolt peak remained constant within each of the examples.
Comparison of Equipment Configurations
A contrast adjustment control is commonly available on mobile radiographic/fluoroscopic units and can be the cause of occasional imaging problems if the default system setting is inadvertently changed. The contrast measurement method can be used to quantify the appearance of the image for various contrast settings. The effect of selecting adjacent steps of the preconfigured contrast adjustment of a mobile unit was evaluated with an anthropomorphic chest phantom (Fig 5). Thecontrast index as measured with the step phantom changed from 430 mV to 360 mV, producing a 16% change in contrast, which is just perceptible to the viewer. The video signal voltage range of the step phantom closely matched that of the more clinical chest phantom.
View larger version (138K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5a. Effect of contrast setting. (a, b) Fluoroscopic images of an anthropomorphic chest phantom. (c, d) Video waveforms of the video signal brightness level across the middle of the chest phantom images. (e, f) Corresponding step phantom video waveforms. a, c, and e were acquired with one contrast adjustment setting, whereas b, d, and f were acquired with the adjacent lower setting.
|
|
View larger version (130K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5b. Effect of contrast setting. (a, b) Fluoroscopic images of an anthropomorphic chest phantom. (c, d) Video waveforms of the video signal brightness level across the middle of the chest phantom images. (e, f) Corresponding step phantom video waveforms. a, c, and e were acquired with one contrast adjustment setting, whereas b, d, and f were acquired with the adjacent lower setting.
|
|
View larger version (58K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5c. Effect of contrast setting. (a, b) Fluoroscopic images of an anthropomorphic chest phantom. (c, d) Video waveforms of the video signal brightness level across the middle of the chest phantom images. (e, f) Corresponding step phantom video waveforms. a, c, and e were acquired with one contrast adjustment setting, whereas b, d, and f were acquired with the adjacent lower setting.
|
|
View larger version (55K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5d. Effect of contrast setting. (a, b) Fluoroscopic images of an anthropomorphic chest phantom. (c, d) Video waveforms of the video signal brightness level across the middle of the chest phantom images. (e, f) Corresponding step phantom video waveforms. a, c, and e were acquired with one contrast adjustment setting, whereas b, d, and f were acquired with the adjacent lower setting.
|
|
View larger version (64K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5e. Effect of contrast setting. (a, b) Fluoroscopic images of an anthropomorphic chest phantom. (c, d) Video waveforms of the video signal brightness level across the middle of the chest phantom images. (e, f) Corresponding step phantom video waveforms. a, c, and e were acquired with one contrast adjustment setting, whereas b, d, and f were acquired with the adjacent lower setting.
|
|
View larger version (61K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5f. Effect of contrast setting. (a, b) Fluoroscopic images of an anthropomorphic chest phantom. (c, d) Video waveforms of the video signal brightness level across the middle of the chest phantom images. (e, f) Corresponding step phantom video waveforms. a, c, and e were acquired with one contrast adjustment setting, whereas b, d, and f were acquired with the adjacent lower setting.
|
|
Grid Use
The step phantom and VWFM provide a mechanism for quantifying and comparing the amount of contrast loss when one is considering whether certain examinations should be performed with or without an antiscatter grid. In one system, removal of the grid decreased the contrast index by 24% (Fig 6), from 450 mV to 340 mV. This 110-mV difference represents a large difference in contrast that is easily detected in clinical practice.
View larger version (158K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6a. Effect of grid use. (a, b) Fluoroscopic images of an anthropomorphic abdominal phantom show the contrast changes seen with (a) and without (b) use of an antiscatter grid. (c, d) Corresponding video waveforms of the step phantom allow quantification of the increase in scatter as a 24% decrease in the contrast index.
|
|
View larger version (156K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6b. Effect of grid use. (a, b) Fluoroscopic images of an anthropomorphic abdominal phantom show the contrast changes seen with (a) and without (b) use of an antiscatter grid. (c, d) Corresponding video waveforms of the step phantom allow quantification of the increase in scatter as a 24% decrease in the contrast index.
|
|
View larger version (69K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6c. Effect of grid use. (a, b) Fluoroscopic images of an anthropomorphic abdominal phantom show the contrast changes seen with (a) and without (b) use of an antiscatter grid. (c, d) Corresponding video waveforms of the step phantom allow quantification of the increase in scatter as a 24% decrease in the contrast index.
|
|
View larger version (86K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 6d. Effect of grid use. (a, b) Fluoroscopic images of an anthropomorphic abdominal phantom show the contrast changes seen with (a) and without (b) use of an antiscatter grid. (c, d) Corresponding video waveforms of the step phantom allow quantification of the increase in scatter as a 24% decrease in the contrast index.
|
|
Matching Contrast between Fluoroscopic Systems
When installing new radiographic equipment, it is often desirable to adjust the image appearance of the new system to match that of an acceptable system in current use. Commonly, this type of adjustment is done subjectively by using clinical examinations on multiple subjects with much trial and error. In these circumstances, we have found the step phantom to be a useful tool to aid modifications to equipment setup in a systematic manner. One such case involved adjustment of a new conventional radiographic/fluoroscopic unit dedicated to neuroradiology applications to match a currently used neuroangiography system (Fig 7). The radiographic/fluoroscopic unit was initially installed by using the vendor’s default video configuration for gastrointestinal imaging applications. During clinical use, physicians commented that the image contrast was significantly less than that of our other neuroangiography equipment from the same vendor. Using the step phantom, we worked in collaboration with the vendor to match the contrast of the two systems.
View larger version (84K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7a. Matching contrast. (a, b) Video waveform of the step phantom (a) and fluoroscopic image of an anthropomorphic skull phantom (b) show the overall contrast of a newly installed imaging system. By benchmarking one of the vendor’s dedicated neuroangiography systems with the step phantom, adjustments to both the TV camera gamma curve and the display window width were made. (c, d) Video waveform of the step phantom (c) and fluoroscopic image of the skull phantom (d) show the effects of the adjustments, which resulted in a 25% (70-mV) increase in the contrast index.
|
|
View larger version (172K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7b. Matching contrast. (a, b) Video waveform of the step phantom (a) and fluoroscopic image of an anthropomorphic skull phantom (b) show the overall contrast of a newly installed imaging system. By benchmarking one of the vendor’s dedicated neuroangiography systems with the step phantom, adjustments to both the TV camera gamma curve and the display window width were made. (c, d) Video waveform of the step phantom (c) and fluoroscopic image of the skull phantom (d) show the effects of the adjustments, which resulted in a 25% (70-mV) increase in the contrast index.
|
|
View larger version (91K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7c. Matching contrast. (a, b) Video waveform of the step phantom (a) and fluoroscopic image of an anthropomorphic skull phantom (b) show the overall contrast of a newly installed imaging system. By benchmarking one of the vendor’s dedicated neuroangiography systems with the step phantom, adjustments to both the TV camera gamma curve and the display window width were made. (c, d) Video waveform of the step phantom (c) and fluoroscopic image of the skull phantom (d) show the effects of the adjustments, which resulted in a 25% (70-mV) increase in the contrast index.
|
|
View larger version (169K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7d. Matching contrast. (a, b) Video waveform of the step phantom (a) and fluoroscopic image of an anthropomorphic skull phantom (b) show the overall contrast of a newly installed imaging system. By benchmarking one of the vendor’s dedicated neuroangiography systems with the step phantom, adjustments to both the TV camera gamma curve and the display window width were made. (c, d) Video waveform of the step phantom (c) and fluoroscopic image of the skull phantom (d) show the effects of the adjustments, which resulted in a 25% (70-mV) increase in the contrast index.
|
|
The solution involved modifying the nonlinear amplification of the TV camera video signal, or gamma curve. The gamma curve, which is designed to optimize the video system for a particular imaging application, varies considerably between vendors and even within one vendor’s product line (Fig 8). Although the optimum gamma curve for a given application is generally determined by the vendor, it is often possible for the service engineer to select a different curve. For the installation in question, the solution involved selecting a different gamma curve and also reducing the digital fluoroscopic window width from a value of 1,023 to 900. The result was a contrast index increase of 70 mV (Fig 7c, 7d), from 280 mV to 350 mV. It is generally not advisable to simply narrow the window width to gain more image contrast. This type of change, made in isolation, will most likely result in increased image flaring due to decreased dynamic range.
View larger version (20K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 8. Graphic representation of the TV camera gamma curve selections available from one vendor. The choice of the appropriate gamma curve depends on the particular imaging application. On cardiac imaging systems, for example, a curve is typically selected that will amplify the low video signal level areas of the image (ie, the iodinated contrast material-filled coronary vessels) while suppressing the high video signal level areas (ie, the lung fields) to reduce the amount of image flaring. Fluoro = fluoroscopy.
|
|
Comparing Contrast before and after Component Replacement
Images were recorded before and after replacement of a 7-year-old II and TV camera pickup tube on a vascular angiography system (Fig 9). There was a marked increase in the video signal voltage levels (ie, brightness) for steps 3 and 4, which represented a 50-mV increase in the contrast index. If it is assumed that the old II and TV camera originally had the same 360-mV contrast index as the new ones, this result represents a 14% loss of contrast over the life of the system.
View larger version (92K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9a. Effect of component replacement. (a, b) Video waveforms of the step phantom before (a) and after (b) the II and TV camera pickup tube were replaced on a vascular imaging system. The contrast index increased 14%, from 310 mV to 360 mV. (c, d) Fluoroscopic images of a phantom designed for subjective measurement of low contrast show little difference before (c) and after (d) component replacement, illustrating its limitation in measurement of overall system contrast. (e, f) Digital subtraction angiograms of an arterial vessel phantom obtained before (e) and after (f) component replacement show contrast improvements that agree well with those demonstrated by the step phantom.
|
|
View larger version (84K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9b. Effect of component replacement. (a, b) Video waveforms of the step phantom before (a) and after (b) the II and TV camera pickup tube were replaced on a vascular imaging system. The contrast index increased 14%, from 310 mV to 360 mV. (c, d) Fluoroscopic images of a phantom designed for subjective measurement of low contrast show little difference before (c) and after (d) component replacement, illustrating its limitation in measurement of overall system contrast. (e, f) Digital subtraction angiograms of an arterial vessel phantom obtained before (e) and after (f) component replacement show contrast improvements that agree well with those demonstrated by the step phantom.
|
|
View larger version (161K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9c. Effect of component replacement. (a, b) Video waveforms of the step phantom before (a) and after (b) the II and TV camera pickup tube were replaced on a vascular imaging system. The contrast index increased 14%, from 310 mV to 360 mV. (c, d) Fluoroscopic images of a phantom designed for subjective measurement of low contrast show little difference before (c) and after (d) component replacement, illustrating its limitation in measurement of overall system contrast. (e, f) Digital subtraction angiograms of an arterial vessel phantom obtained before (e) and after (f) component replacement show contrast improvements that agree well with those demonstrated by the step phantom.
|
|
View larger version (164K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9d. Effect of component replacement. (a, b) Video waveforms of the step phantom before (a) and after (b) the II and TV camera pickup tube were replaced on a vascular imaging system. The contrast index increased 14%, from 310 mV to 360 mV. (c, d) Fluoroscopic images of a phantom designed for subjective measurement of low contrast show little difference before (c) and after (d) component replacement, illustrating its limitation in measurement of overall system contrast. (e, f) Digital subtraction angiograms of an arterial vessel phantom obtained before (e) and after (f) component replacement show contrast improvements that agree well with those demonstrated by the step phantom.
|
|
View larger version (174K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9e. Effect of component replacement. (a, b) Video waveforms of the step phantom before (a) and after (b) the II and TV camera pickup tube were replaced on a vascular imaging system. The contrast index increased 14%, from 310 mV to 360 mV. (c, d) Fluoroscopic images of a phantom designed for subjective measurement of low contrast show little difference before (c) and after (d) component replacement, illustrating its limitation in measurement of overall system contrast. (e, f) Digital subtraction angiograms of an arterial vessel phantom obtained before (e) and after (f) component replacement show contrast improvements that agree well with those demonstrated by the step phantom.
|
|
View larger version (193K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 9f. Effect of component replacement. (a, b) Video waveforms of the step phantom before (a) and after (b) the II and TV camera pickup tube were replaced on a vascular imaging system. The contrast index increased 14%, from 310 mV to 360 mV. (c, d) Fluoroscopic images of a phantom designed for subjective measurement of low contrast show little difference before (c) and after (d) component replacement, illustrating its limitation in measurement of overall system contrast. (e, f) Digital subtraction angiograms of an arterial vessel phantom obtained before (e) and after (f) component replacement show contrast improvements that agree well with those demonstrated by the step phantom.
|
|
It is also possible to evaluate contrast differences by using phantoms with threshold contrast detection test objects or simulated blood vessels. Images of these types of phantoms were acquired before and after component replacement on the system described in the preceding paragraph for comparison with the step phantom measurements. Images of a threshold contrast detection phantom (model 07-645 Fluoro-Test tool; Nuclear Associates, Carle Place, NY) obtained before (Fig 9c) and after (Fig 9d) component replacement showed little difference between the two images as viewed in photographs of last-image-hold images. The similarity between the images was even more apparent when they were viewed under live fluoroscopic conditions. However, digital subtraction angiograms of an arterial vessel phantom (model 76-700-1150 digital subtraction angiography phantom; Nuclear Associates) obtained before (Fig 9e) and after (Fig 9f) component replacement showed subtle improvement in both the contrast and sharpness of the simulated vessels. The noticeable increase in image noise seen after component replacement (Fig 9f) was primarily a result of the increased gain of the new II, which caused the system to select a 37% lower exposure, and less a result of the increased contrast.
Acceptance Testing of Equipment
Quantified evidence of poor system performance during acceptance testing can be provided to an equipment vendor by using the step phantom and VWFM method. During preinstallation acceptance testing of a new angiography system at the vendor’s manufacturing facility, it was found that the overall contrast of the new system (Fig 10a) was significantly less than that of our currently installed equipment of the same model (Fig 9b). The manufacturer replaced the TV camera in an attempt to correct the problem and then shipped the system to us. After installation at our clinical site, it was discovered that the contrast had not been improved.
View larger version (100K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10a. Documentation of contrast improvements made to an imaging system that was not performing to the same level as a like piece of equipment from the same vendor. Video waveforms of the step phantom obtained at installation (a), after service by local engineers (b), and after service by national technical support engineers (c) show incremental improvements in contrast.
|
|
View larger version (89K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10b. Documentation of contrast improvements made to an imaging system that was not performing to the same level as a like piece of equipment from the same vendor. Video waveforms of the step phantom obtained at installation (a), after service by local engineers (b), and after service by national technical support engineers (c) show incremental improvements in contrast.
|
|
View larger version (92K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 10c. Documentation of contrast improvements made to an imaging system that was not performing to the same level as a like piece of equipment from the same vendor. Video waveforms of the step phantom obtained at installation (a), after service by local engineers (b), and after service by national technical support engineers (c) show incremental improvements in contrast.
|
|
The local field service engineers were able to achieve only a small improvement by adjusting some of the setup parameters of the video chain while staying within the vendor’s documented specifications (Fig 10b). The system was then turned over for clinical use, with almost immediate complaints by radiologists of inability to see the guide tips of microcatheters and adequately visualize guide wires, particularly when imaging over the patient’s spine. The vendor responded by having the system serviced by one of their national service support technicians, who compared the video signal being supplied by the video out-put boards of the two systems and then made several parameter adjustments to the new system. The result of these adjustments was an improvement in the contrast index from 240 mV to 350 mV, so that the contrast of the new system (Fig 10c) was now similar to that of the old system (Fig 9b).
|
Discussion
|
|---|
Our experience suggests that measurement of fluoroscopic image contrast can be a valuable tool for evaluating equipment performance, configuring new systems, and routine quality control. When measurements of the overall contrast indicate a deficiency in the imaging system, it is important to examine each component of the imaging chain to determine how contrast performance can be improved. Corrective action may require adjustment or replacement of the II, TV camera, or other video system components. This test method has been instrumental in detecting poor adjustment of the TV system on several occasions. Other reports describing performance assessment of fluoroscopic systems have also con-cluded that the TV system was the most likely component to be misadjusted (4).
We have found that a system with a contrast index of less than 300 mV in a 20–23-cm field of view will elicit complaints of low contrast from physicians. Indexes between 300 and 350 mV may also be perceived as being too low for some applications or by the more discerning physician. Our experience indicates that radiologists can detect contrast changes in clinical images that are approximately equal to a 15% change in the contrast index. Because of the eye’s logarithmic response to changes in brightness, contrast changes in the blacks tend to affect contrast perception more than changes in the whites (5).
It is possible to measure video signal levels by using a VWFM, an oscilloscope, or digital image analysis software. We prefer using a VWFM instead of an oscilloscope because of ease of use and the ability to integrate 14 scan lines of video into a signal composite waveform, thereby improving the observer’s ability to interpret the noise characteristics of the waveform as part of a complete system evaluation. There is a marked difference in signal strength when a single line is selected (Fig 9b) versus when 14 lines are selected (Fig 10c). Unfortunately, VWFMs do not have the means to output the contents of their display screens; therefore, documentation of test results must be done with a scope camera. Oscilloscopes offer the advantages of RS-232 digital output, storage of multiple waveforms, triggering off of any video signal source, and onboard waveform analysis tools. Use of digital image analysis to measure pixel values of the contrast steps can be quite convenient; however, the necessary software tools are not available on all systems and digital image transfer to an external workstation may not be practical or possible.
The amount of brightness falloff and anode-cathode orientation should be considered when one is making comparisons between different systems or between planes on a biplane system. The close, central arrangement of the steps in the phantom generally results in minimal variation from brightness falloff. However, brightness levels from a uniform region of the phantom image (just above or below the steps) may be used to correct brightness levels for falloff. To eliminate the nonuniformity due to the heel effect, it may be necessary to rotate the phantom 90° and make the measurements vertically across the entire video field, instead of horizontally along the raster scan lines.
The test method described herein measures the image information that is sent to the fluoroscopic display monitor. Therefore, the contrast of the monitor itself is not characterized. It is important to optimize the monitor settings for clinical viewing by properly adjusting contrast and brightness as part of a complete system evaluation (6)–(8). Variations in monitor contrast complicate the performance of system comparisons that reflect clinical presentation. By means of a photometer, at least one vendor uses the monitor as the final contrast adjustment on its systems. This uncontrolled variable must be considered if the test method is used to compare systems with different monitors. We are evaluating the usefulness of photometer measurements of monitor brightness over each contrast step as a method of quantifying complete system contrast.
Fluoroscopic contrast resolution can also be assessed by using several different test objects, including the Fluoro-Test tool (9), Leeds test object TO.N3 (10), and the image quality test object of the Center for Devices and Radiological Health (CDRH) of the Food and Drug Administration (11). Each of these test objects contains a series of steps of varying contrast. A measure of the threshold contrast of the system is obtained by viewing the fluoroscopic image of the test object and determining the lowest contrast step that is just perceptible to the viewer. The contrast index value measured with the technique described in this article is related to the ability of an imaging system to resolve low-contrast objects. However, the contrast index provides a measure of the typical brightness range of the video signal instead of the contrast present at a single brightness level. In addition, the contrast index provides a direct quantitative value that is not affected by the subjective nature of threshold contrast detection. Although contrast resolution test objects provide valuable information about low-contrast resolution performance, they are not designed to measure the overall contrast of the imaging system.
The contrast of II-TV digital systems can also be investigated by using various techniques to measure the characteristic curve, which relates the output pixel value to the input relative x-ray intensity. Characteristic curves can be derived from a series of exposures at various input exposure levels (12) or a single image of a step wedge (13), allowing quantitative analysis of digital gray-level information. For either method, the automatic exposure control system must be disabled and the digital gray-level data must be recovered. The contrast measurement method described herein provides a simplified assessment of image contrast that is useful for monitoring and comparison of the system response of both analog and digital systems. However, the method is not designed for quantitative determination of the characteristic curve.
This method has also proved useful in evaluating new flat-panel detectors for fluoroscopy. Since the gray scale of images produced by flat-panel detection systems is determined by variable lookup table parameters, the contrast scale is not constrained by limitations present in conventional video amplification components. The results of the step phantom measurements provide a benchmark for making comparisons to II-based systems and provide baseline quality control measurements for the detector.
|
Summary
|
|---|
We have developed a simple method for measuring image contrast in fluoroscopic imaging systems. The method is useful for optimization of video parameters, quality control monitoring, and characterization of differences in contrast between systems. Our experience suggests excellent correlation with the physician’s perception of image contrast. A change of 15% in the contrast index is clinically visible, and systems that have a contrast index below 300 mV will be perceived as having poor contrast.
|
Footnotes
|
|---|
Abbreviations: ANSI = American National Standards Institute, II = image intensifier, TV = television, VWFM = video waveform monitor
|
References
|
|---|
-
. Test standards for the determination of the system contrast ratio and the system veiling glare index of an x-ray image intensifier system. NEMA XR 16 Rosslyn, Va: National Electrical Manufacturers Association, 1996.
-
. . EIA standard RS-170 Arlington, Va: Electronic Industries Association, November, 1957.
-
. AAPM electronic imaging in medicine, Medical Physics Monograph no 11. College Park, Md: American Institute of Physics, 1984; 424-437.
-
Poletti JL, Le Heron JC. Performance assessment of x-ray image intensified television fluoroscopic systems in New Zealand. Br J Radiol 1988; 61:393-400.[Abstract/Free Full Text]
-
Cornsweet TN. Visual perception New York, NY: Harcourt Brace Jovanovich, 1970; 246-251.
-
. Barium enema quality control manual, 1998 Reston, Va: American College of Radiology, 1998; 86.
-
Gray JE, Lisk KG, Haddick DH, Harshbarger JH, Oosterhof A, Schwenker R. Test pattern for video displays and hardcopy cameras. Radiology 1985; 154:519-527.[Abstract/Free Full Text]
-
. Specification acceptance testing and quality control of diagnostic x-ray imaging equipment. Medical Physics Monograph no 20. College Park, Md: American Institute of Physics, 1991; 499-568.
-
Wagner AJ, Barnes GT, Xizeng W. Assessing fluoroscopic contrast resolution: a practical and quantitative test tool. Med Phys 1991; 18:894-899.[Medline]
-
Cowen AR, Clark OF, Coleman NJ, Craven DM, McArdle S, Hay GA. Leeds x-ray test objects instruction manual Leeds, England: University of Leeds,; .
-
Suleiman OH, Conway BJ, Quinn P, et al. National survey of fluoroscopy: radiation dose and image quality. Radiology 1997; 203:471-476.[Abstract/Free Full Text]
-
MacIntyre WJ, Pavlicek W, Gallagher JH, Meaney TF, Buonocore E, Weinstein MA. Imaging capability of an experimental digital subtraction angiography unit. Radiology 1981; 139:307-313.[Abstract/Free Full Text]
-
Fujita H, Doi K, Giger ML, Chan H. Investigation of basic imaging properties in digital radiography. V. Characteristic curves of II-TV digital systems. Med Phys 1986; 13:13-18.[Medline]
Top
Abstract
Introduction
