| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, University of Cologne, Medical School,
Joseph-Stelzmann-Str. 9, 50924 Cologne, Germany.
2 Philips Medical Systems, Röntgenstraße 24, Hamburg 22335,
Germany.
3 Department of Medical Statistics, Informatics and Epidemiology, University of
Cologne, Medical School, 50924 Cologne, Germany.
4 Department of Urology, University of Cologne, Medical School, 50924 Cologne,
Germany.
Received September 16, 2002;
accepted after revision April 21, 2003.
Address correspondence to M. Zähringer.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Fifty patients who had a clinical indication for urography underwent unenhanced abdominal imaging that was alternately performed using flat-panel silicon X-ray detector radiography or computed radiography. For patients who required a second radiograph with contrast medium, the examination modality was changed to avoid exposing the patients to excess radiation. The images obtained on flat-panel X-ray detector radiography were obtained at half the radiation dose of the images obtained on computed radiography (800 speed vs 400 speed). The resulting 50 pairs of images were interpreted by four independent observers who rated the detectability of structures of bone and the efferent urinary tract relevant to diagnosis and compared the image quality.
RESULTS. At half the radiation dose, digital flat-panel X-ray detector radiography provided equivalent image quality of the liver and spleen, lumbar vertebrae 2 and 5, pelvis, and psoas margin on abdominal radiographs. The image quality obtained with digital flat-panel X-ray detector radiography of the kidneys, the hollow cavities of the upper efferent urinary tract, and the urinary bladder was judged to be statistically better than those obtained with computed radiography.
CONCLUSION. With half the exposure dose of computed radiography, the flat-panel X-ray detector produced urograms with an image quality equivalent to or better than computed radiography.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Recent developments in digital projection radiography have produced active matrix flat-panel detectors with electronic transistor switches. Materials suitable as X-ray detectors include thallium-doped cesium iodide or amorphous selenium [5, 6] in conjunction with an amorphous silicon layer [7–12]. The electronic amorphous silicon-based flat-panel detector with a cesium iodide scintillator allows direct acquisition of digital radiographs. No intermediate steps involving optical or mechanical scanning are necessary because of a large-area semiconductor layer with integrated transistors. Arrays of amorphous silicon thin-film transistors are at the heart of the new flat-panel detector. Within the scintillator, cesium iodide–area X-ray quanta interacting with the crystalline structure generate light flashes, the intensity of which depends on the quantity of X-ray quanta. The detector consists of a matrix with light-sensitive elements that store the electrical charges in individual sensors. Each sensor corresponds to a pixel in the radiograph. The use of amorphous silicon is necessary to create detector areas that are large enough to capture the anatomic images.
Silicon is not sensitive enough to detect the radiation energy required for radiologic diagnostics, so a cesium iodide image conversion layer is applied over the amorphous silicon layer to absorb the X-ray photons and convert them into light photons that are more easily detected by the photodiodes of the silicon. The needlelike crystals of the cesium iodide work like light guides, preventing scatter effects that reduce the resolution of other phosphors [5, 13]. The quantum detection efficiency of these new detectors based on cesium iodide scintillators at low spatial frequencies is approximately 60%—more than twice as great as those of analog film-screen radiography and computed radiography systems. This high efficiency makes it conceivable that a radiograph can be acquired with less radiation and with no loss of image quality [14].
The aim of our clinical study was to investigate whether it is possible to reduce radiation exposure by a factor of two by using a flat-panel silicon X-ray detector instead of a computed radiography system and still obtain at least an equivalent (subjectively determined) image quality in excretory urograms.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The patients were randomly assigned to undergo unenhanced abdominal radiography performed on either a digital flat-panel detector radiography (Digital Diagnost, Philips Medical Systems, Hamburg, Germany) or computed radiography (ADC Compact, Agfa Medical Systems, Cologne, Germany) system. To spare patients excessive exposure to radiation, we avoided using double exposure. Therefore, we switched to the alternative technique for the first contrast-enhanced (Imeron 300, Byk Gulden, Konstanz, Germany) image and switched again, if clinically required, to obtain the next image. The radiation dose to which the patients were exposed during examinations on the flat-panel detector radiography system was half the dose that patients received during examinations on computed radiography system (computed radiography = 400-speed film-screen system; flat-panel detector radiography = 800-speed film-screen system).
The 50 image pairs (unenhanced and enhanced images) that were acquired allowed a direct comparison of the detectability of anatomic bone structures. In 20 of the 50 patients, only one contrast-enhanced image was necessary to confirm the diagnosis. However, in 30 patients, two or more exposures were obtained after administration of the contrast medium, giving us the opportunity to compare the detail revealed on the radiographs of the efferent urinary tract (Fig. 1A, 1B) and of the bone structures. Of the 30 patients who underwent urography, 14 underwent contrast-enhanced digital flat-panel detector radiography as the first examination, and 16 underwent contrast-enhanced computed radiography as the first examination. Only one contrast-enhanced image pair per patient was compared. To ensure that differences in the patients' excretory function did not affect quality ratings of the images obtained, we recorded serum creatinine values before starting the examination to ensure the patients had comparable functioning.
|
|
Examination Technique
We determine the pixel size of the flat-panel detector from the distance
between the sensor elements. In the case of a 43 x 43 cm detector, the
distance is 143 µm, resulting in a matrix of 3000 x 3000 pixels. This
pixel size leads to a spatial resolution of more than 3.5 line pairs per
millimeter (lp/mm), sufficient for most projection radiographic studies
[4,
16–19].
The quantum detection efficiency of the detector at low spatial frequencies is
approximately 60%.
The technology and physics of computed radiography are well known [20–22]. The Agfa ADC Compact system that we used has an image format of 35 x 43 cm overlaying a matrix of 1760 x 2140 pixels, equivalent to a pixel size of 0.2 mm. Its spatial resolution is 2.5 lp/mm at a pixel depth of 12 bits per pixel. The dynamic range is 1:40000, and the quantum detection efficiency is approximately 25%.
The following settings were used for the examinations on both flat-panel detector radiography and computed radiography systems: film-focus distance, 110 cm; 77 kV; and maximum image size, 43 x 43 cm for Digital Diagnost and 35 x 43 cm for ADC Compact. We were able to reduce the radiation exposure for patients undergoing digital flat-panel detector radiography by halving the milliampere-seconds product required for computed radiography. Mean milliampere-second values for computed radiography were 6.9 mAs and for flat-panel detector radiography, 3.4 mAs. Because milliampere-second is linearly related to exposure dose, the assumed reduction in the radiation dose with digital flat-panel detector radiography is approximately 49%. The exposure values corresponded to those of a 400-speed film-screen system with computed radiography and to an 800-speed film-screen system with digital flat-panel detector radiography.
Dosimetric studies were conducted using a semiconductor detector (Dosimax, Wellhöfer, Schwarzenbruck, Germany) before the patients underwent radiography. The image processing of digital flat-panel detector radiography and computed radiography was based on the sets of image-processing parameters supplied by the manufacturers. Before the formal evaluation of image quality began, the sets of images obtained in several patients were processed using varying parameters that were then judged by a group of three experienced radiologists so as to optimize the brightness, contrast resolution, and detailed enhancement of the images and to standardize the parameters. After the optimal parameters had been determined, postprocessing in both systems was completed and the image data from the flat-panel detector radiography and the computed radiography systems were sent to a joint workstation (Easy-Vision, Philips Medical Systems, Hamburg, Germany). To ensure that the film layouts gave no clues as to which technique was used, the images produced by digital flat-panel detector and computed radiography systems were printed out from this workstation on a laser printer (Matrix LR 3300 [Agfa Medical Systems] and film, Scopix LT 2B Dayl.A1 [Agfa Medical Systems], reproduction matrix, 4256 x 5174; 315 dots per inch).
Evaluation
We separated the image pairs and deleted patient-identifying information on
the radiographs but ensured that we could identify the pairs of images that
belonged together after the image evaluation was completed by coding the
images with randomly generated numbers. The 100 images were interpreted by
four independent observers experienced in uroradiology (two radiologists and
two radiology residents in their fifth year of training). Their written
evaluations were recorded on a report form that included all imaging criteria.
These reviewers were required to rate the detectability of various anatomic
structures in each image pair using a 5-point quality scale (1, an anatomic
structure was well visualized in all sections; 2, the structure was well
visualized in some sections; 3, satisfactory visualization; 4, insufficient
visualization; and 5, the structure could not be detected on the image). The
images were observed by viewers under standardized conditions in dedicated
observer-search rooms in an isolated research environment. Room lights were
off, and no brightening lamps were used.
For the analysis, the scores for all 50 pairs of images submitted by the four observers were separated according to the image technique. Sum scores were produced that encompassed the associated anatomic structures (image quality of the liver and spleen, the kidneys and efferent urinary tract, lumbar vertebrae 2 and 5, and pelvis with the approximate femur). Two to six typical organ features were described, depending on the structure being reviewed. Because of this procedure, these scores were not in the original range of 1–5. The median of the four radiographers' scores was then calculated for the two methods so that we could compare the detectability of bone structures in all 50 patients. Visualization of the kidneys and the efferent urinary tract were directly comparable in 30 of the 50 patients, and the sections with the best contrast of the upper cavity system were evaluated. Using Wilcoxon's signed rank sum test, we tested the medians of the sum scores for differences between the imaging techniques. The null hypothesis was that the image quality of an anatomic structure did not differ between the two imaging techniques or that computed radiography produced better quality.
Figure 2A, 2B, 2C, 2D, 2E, 2F, 2G, 2H depicts the medians of the sum scores for flat-panel detector radiography (x-axis) and the computed radiography (y-axis) as point clouds. We also analyzed the data using the Bland-Altman plot for assessing clinical agreement [23, 24] between flat-panel detector radiography and computed radiography. The average of both methods was plotted on the x-axis (mean of the medians from the sum score). The difference between the two values (median flat-panel detector radiography minus median computed radiography) was plotted on the y-axis. The mean and the mean ± 2 SDs were calculated. Dividing the mean ± 2 SDs by the number of typical features of the organ or other structure produces a value that can be interpreted by applying the original 1–5 rating scale. The interval mean ± 2 SDs (limits of agreement) can be used to decide whether observed deviations between the image quality of the two imaging techniques are within a clinically acceptable range.
|
|
|
|
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Kidneys (Sum Score with an Interval of 2–10)
The imaging quality rating scores for the left and right kidneys were added
to produce a sum score of between 2 and 10
(Fig. 2B). A statistically
significant difference was found in favor of digital flat-panel detector
radiography for the quality of images of both kidneys (p = 0.001).
Therefore, the Bland-Altman plot for both kidneys was not necessary.
Hollow Cavities of the Upper Efferent Urinary Tract and the Urinary
Bladder (Sum Score with an Interval of 4–20)
The individual image quality scores for the renal collecting system, the
proximal and distal views of the ureter, and the urinary bladder contours were
added and produced a sum score for both of the tested radiographic techniques
of between 4 and 20 (Fig. 2C).
The difference between the sum score medians using Wilcoxon's rank sum tests
proved to be statistically remarkable (p < 0.0005). The image
quality produced by flat-panel detector radiography was rated better overall
than the computed radiography. Testing the results by Bland-Altman plots was
not necessary.
Lumbar Vertebrae 2 and 5 (Sum Score with an Interval of
6–30)
The sum score of the lumbar vertebrae 2 and 5 combines visualization of the
upper and lower endplates of the vertebrae, spinous and transverse processes,
vertebral pedicles, and trabecular structures in the cancellous bone.
Therefore, the sum score fell between 6 and 30 for each procedure (Figs.
2D and
2E). Wilcoxon's rank sum test
showed that the difference between the sum score medians for lumbar vertebrae
2 and 5 was not statistically remarkable (L2, p = 0.136; L5,
p = 0.560). Additionally, the Bland-Altman plots for L2 and L5
(Fig. 2F) showed a comparable
image quality for computed radiography and flat-panel detector radiography
(Table 1).
Bilateral Psoas Margin (Sum Score with an Interval of
2–10)
Adding the scores obtained for images on both sides of the psoas margin
resulted in an interval of 2–10 (Fig.
2G). Comparisons of the image quality of depictions of the psoas
margin stripe produced by the two radiography systems reveal no statistically
significant difference (p = 0.744).
Pelvis, Sacroiliac Joint, and Head of Femur (Sum Score with an
Interval of 4–20)
Adding the sum score medians produced an interval of 4–20
(Fig. 2H). According to the
results of Wilcoxon's rank sum test, no statistically significant difference
was found between the image quality produced with the flat-panel detector
radiography and computed radiography for visualization of the bone structure
of the pelvis, femoral head, and sacroiliac joint (p = 0.251). The
limits of agreement are in acceptable range of clinical agreement (–0.82
to 0.72).
The values of the Bland-Altman plot for different anatomic regions (Table 1) showed a comparable image quality for computed radiography and digital flat-panel detector radiography, with a narrow 95% confidence limit.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Recent advances have made digital flat-panel detector radiography a viable new option in medical imaging [10, 14, 25]. Compared with computed radiography, flat-panel detector radiography based on amorphous silicon speeds up the readout process considerably by means of active transistor matrix arrays. Immediate image access is thereby possible, and handling the cassette systems, a critical time factor that hampers workflow in both the analog film-screen and computed radiography systems, is no longer necessary.
The international radiology literature abounds with articles describing the theoretic principles on which digital flat-panel detector radiography is based [5, 14, 25]. Likewise, many reports have been published of phantom studies investigating the detectability of artificial bone lesions, simulated pulmonary nodules, and foreign bodies [8–11], as well as initial clinical studies in bone diagnostics [11]. The compiled evidence strongly suggests that digital flat-panel detector radiography is a suitable procedure for bone and chest diagnostics that offers the possibility of reducing the radiation dose with no loss of image quality. Zähringer et al. [26] and Okamura et al. [27] showed that the image quality of digital radiography with a flat-panel detector in abdominal imaging before and after IV contrast medium administration was superior to that of film-screen or computed radiography systems using the same exposure parameters.
Compared with analog film-screen radiography systems, computed radiography provides better contrast resolution, a broader dynamic range, contrast enhancement, and more flexible options for reducing the dose in certain cases. Gradation processing remains consistent because the contrast and blackening settings are automatically optimized, independent of dose or exposure parameters, virtually eliminating the possibility of poor exposures. Compared with analog film-screen radiography, computed radiography has a lower limiting spatial resolution of 2.5–5.0 lp/mm, depending on the image format [20, 21]. The quantum detection efficiency of the most recent (fifth) generation of computed plates has caught up with that of analog film-screen radiography [22]. Optimum image postprocessing options are also largely responsible for the good image quality of digitally stored images [28].
Controlled clinical studies have established that computed radiography and analog film-screen systems using identical radiation doses perform equally well in producing urograms for uroradiologic diagnostics [29–31]. In pediatric uroradiology, computed radiography allows the dose to be reduced by 75% in children and adolescents. In small children and infants, images produced by dose-reduced computed techniques of the third generation were considered at the time to yield diagnostically insufficient results [32]. Experimental studies have shown that the digital quantum detection efficiency of digital flat-panel detector radiography is approximately a factor of two higher than that of computed radiography [13, 14]. This finding may lead to a better contrast resolution depending on the image processing methods applied and the parameters used. Physical tests show a better low-contrast resolution with digital flat-panel detector radiography. In our study, the same calibration methods have been applied for both systems to control the impact on image quality by converting the digital image files to film. The resulting image quality was checked during the acceptance tests and by constancy testing during routine operations.
Our study shows that digital flat-panel detector radiography allows equivalent detectability of bone structures, which are the basis for diagnosing abnormal changes on abdominal radiographs, at half the exposure dose of computed radiography. The image quality of the contrast-enhanced radiographs obtained on digital flat-panel detector radiography of both kidneys, the renal pelvis, and the urinary collecting system, ureter, and urinary bladder was rated as significantly better than that of computed radiography. The technical precondition for the superior image quality is that the quantum detection efficiency of the new silicon detector system is higher than that of the computed radiography system. The pixel size of 143 µm produced a spatial resolution that exceeded 3.5 lp/mm, which is considered sufficient for rendering images of the quality required for uroradiologic diagnoses [30, 31].
No significant difference was found between the two imaging techniques in the depiction of the psoas margin stripe. The psoas margin is a fine low-contrast structure that is similar to the kidneys in that it contains high-spatial frequencies, which should be better displayed on digital flat-panel detector radiography than on other types of radiography. In practice, detectability of small low-contrast structures is greatly affected by the superposition of bowel gas, mucosal folds, or fecal material. This fact might be an explanation for the insignificant difference between the two systems for the detection of the psoas margin stripe.
As mentioned previously, urograms make up 4% of all radiographic examinations in a radiology department but represent 40% of all radiographic examinations performed in a urology department. Therefore, increased speed of data acquisition using digital flat-panel detector radiography with direct readout of image data and without the need for handling cassette systems might result in improved patient throughput, although we did not measure to what extent patient throughput was increased in our department. In addition, apart from bowel-contrast examinations, excretory urograms expose the gonads to the highest radiation dose of all projection radiographic techniques; thus, dose reduction is essential in uroradiology.
Our study had one methodologic limitation: To protect the patients from excessive radiation, we obtained no double exposures. Clinically indicated abdominal radiography was performed before and after IV contrast medium administration alternately using digital flat-panel detector radiography and computed radiography. Therefore, direct comparison of the image quality of the efferent urinary tract was not possible in all patients examined. In 20 patients, the results obtained after one contrast-enhanced radiograph were sufficient for the diagnosis. Hence, we refrained from contrast-enhanced imaging as the control technique in the interest of these patients.
It was the aim of our study to subjectively compare the quality of images produced with digital flat-panel detector radiography with those produced with computed radiography, rather than to compare the detectability of abnormalities. Future investigations should compare digital flat-panel detector radiography with computed radiography as a tool for revealing abnormal lesions.
The images were printed on film and interpreted on lightboxes. If the images had been interpreted on computer workstations with considerably less luminance, our results might have been a different. However, this statement would also apply if we had used a light box with reduced luminance. In addition, interpretation on a monitor allows individual adjustment of brightness and contrast. In this case, the improved signal-to-noise ratio and the larger dynamic range of the digital flat-panel detector radiography should result in even better scoring results for the technique.
In conclusion, our study should be regarded as a step toward appraising the use of digital flat-panel silicon X-ray detector technology in uroradiologic diagnostics. The introduction of digital flat-panel detector radiography in projection radiography gives us a technology that yields enhanced performance at half the conventional exposure dose used in computed radiography while maintaining an equivalent or better detectability of anatomic structures before and after IV contrast medium administration. Digital flat-panel detector radiography provides these advantages while obviating use of cassettes and perhaps allowing greater patient throughput.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  W. Huda The Current Concept of Speed Should Not Be Used to Describe Digital Imaging Systems Radiology, February 1, 2005; 234(2): 345 - 346. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
