| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Departments of Radiology (W.J.H.V., L.J.M.K., J.G.) and Medical Statistics (B.J.A.M.), C2S, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, the Netherlands. Received November 26, 2003; revision requested February 10, 2004; final revision received July 14; accepted August 26. Address correspondence to W.J.H.V. (e-mail: w.j.h.veldkamp@lumc.nl).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
MATERIALS AND METHODS: The chest digital radiography system is a digital charge-coupled device (CCD) chest imaging unit that uses slot-scan technology. A contrast-detail test object was used in combination with a phantom that simulates the primary and scatter transmission for the lungs and mediastinum. Twelve phantom images were obtained with each modality (ie, CCD digital radiography and AMBER and Bucky screen-film radiography) and were judged by six observers. CCD digital radiography soft-copy reading was compared with AMBER hard-copy reading. To measure image quality, contrast-detail curves were constructed from the observer data. The Wilcoxon signed rank test was used for statistical analysis.
RESULTS: For the lung configuration, contrast-detail curves showed lower threshold depth for hard-copy images obtained with CCD digital radiography than for those obtained with Bucky screen-film radiography. For hard-copy images, the difference between contrast-detail curves for CCD digital radiography and those for Bucky screen-film radiography was statistically significant (P < .006). No significant difference was found between CCD digital radiography and AMBER for hard-copy images obtained in either the lung or mediastinum configuration. For the lung configuration, a lower threshold depth was observed for CCD digital radiography soft-copy reading than for AMBER hard-copy reading, with significantly different contrast-detail curves for CCD digital radiography soft copy and AMBER hard copy (P < .006). No significant difference was found between either system for the mediastinum configuration.
CONCLUSION: Contrast-detail curves indicate that image quality for the CCD chest system provides a digital alternative to AMBER and Bucky screen-film radiography.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Recently, a new digital radiography system (ThoraScan; Nucletron-Oldelft, Veenendaal, the Netherlands) was introduced for chest imaging. The ThoraScan unit contains a charge-coupled device (CCD)–based detector that is built from an array of detector elements. CCD detectors offer low noise and high sensitivity (3,4). With the ThoraScan unit, a slot-scan technique is used. Advantages of the slot-scan concept include dose-efficient scatter rejection and the possibility of using small detectors to image large areas without demagnification.
Contrast-detail studies and assessment of physical imaging characteristics, such as the modulation transfer function (MTF), provide important indications of image quality (5,6). Such studies are recommended as a first step in the assessment of new medical imaging equipment (6). Thus, the purpose of our study was to evaluate the image quality and performance of a digital radiography system for chest imaging (ThoraScan) and to compare this with the image quality and performance of advanced multiple-beam equalization radiography (AMBER) and Bucky screen-film radiography systems.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
AMBER and Bucky Screen-Film Radiography
The CCD digital radiography system was compared with AMBER (Nucletron-Oldelft) and Bucky wall stand (Optimus S/F; Philips Medical Systems, Best, the Netherlands) screen-film radiography systems. The AMBER unit uses a scanning x-ray beam, with beam intensity modulated in response to measurements from a detector located behind the film cassette. The system adjusts beam intensity to compensate for reduced optical density in dense regions, such as the mediastinum, with improved visualization of mediastinal structures and retrocardiac and retrodiaphragmatic areas (7).
Contrast-Detail Test Object
A contrast-detail radiographic test object (University Medical Center Nijmegen, Nijmegen, the Netherlands) was used for the contrast-detail study. This test object consisted of a polymethyl methacrylate tablet (area, 26.5 x 26.5 cm; thickness, 10.0 mm) with cylindrical holes of varying diameter and depth in a 15 x 15 matrix, which yielded a total of 225 cells (8). In each cell, either one or two holes are present. Cells in the first three rows contain only one central hole, while cells in the other rows contain two identical holes—one in the middle and one in a randomly selected corner (Fig 1). In the horizontal direction, the hole depth increases logarithmically from 0.3 to 8.0 mm. In the vertical direction, the hole diameter increases logarithmically from 0.3 to 8.0 mm.
|
Table 1 gives parameter settings for the three chest imaging systems. Antiscatter grids were used for the screen-film radiography systems. For AMBER, a moving grid with a ratio of 12:1 and 36 lines per centimeter was used. For Bucky screen-film radiography, a moving grid with a ratio of 12:1 and 36 lines per centimeter was used. All images were acquired with tube voltage and filtration set according to the clinical protocols. Tube voltage and charge settings for each chest system are given in Table 2. Protocols were established by radiologists, physicists, and manufacturers together. Because of the scanning properties of CCD digital radiography and AMBER, exposure times and, therefore, tube charges were much larger for these systems than for Bucky screen-film radiography. Local exposure times, however, were short to avoid movement artifacts. The settings were chosen automatically for both the lung and the mediastinum configurations. Window width and window level settings for hard-copy images obtained with CCD digital radiography corresponded to the default clinical settings for an average-sized patient. The multiscale image contrast amplification postprocessing settings, used for both soft-copy and hard-copy images, were as follows: crossover, –2.0; edge contrast, 0.0; latitude reduction, 0.0; multiscale contrast, 3.0; and noise reduction, 1.0.
|
|
For image reading, three radiologists and three physicists participated as observers. Among the radiologists was one author (L.J.K.M.). The radiologists had a mean experience of 16 years in reading chest images (range, 8–20 years). Among the physicists were two authors (W.J.H.V., J.G.). The physicists had a mean experience of 13 years in medical physics (range, 8–15 years). All 36 images were used for reading. For the lung configuration, the following paradigm was used: Each set of six lung images (corresponding to CCD digital radiography, AMBER, or Bucky screen-film radiography) was equally distributed among the six observers; that is, each observer scored two different images per set, and each image was scored two times by two different observers. Because reading was performed for all possible six permutations of the three modalities, the order in which the observers had to judge images from the different modalities cancelled out any learning effect.
For the mediastinum configuration, the Bucky screen-film images were initially left out of the scheme owing to poor image quality (ie, underexposed films), which reflected image characteristics in the mediastinal region on Bucky screen-film images. Image quality in this region is poor because of the small dynamic range of Bucky screen-film radiography systems compared with the wider dynamic range of AMBER systems and the much wider dynamic range of CCD digital radiography systems. Each of the remaining two sets of six images (CCD digital radiography and AMBER images) was distributed among the six observers in the same way as for the lung configuration. The order of images from the two modalities in which reading was performed was evenly alternated among the observers.
A four-alternative forced-choice detectability experiment was performed. The first three rows on the phantom were not evaluated because these rows contained only a central hole. Observers were instructed to mark the dot-containing corner in each cell even if they were uncertain of the dot’s location. All observers were familiarized with the experimental procedure prior to initiating each test. All images were judged on the same regular light box (Rotolux Planilux; Gerätebau F. Schulte, Warstein, Germany), and a pencil was used to mark the detected location of the dot on the hard-copy images. The observers were free to choose their viewing distance. They were not allowed to move or lift the film to improve reading. Mediastinum images were presented first, followed by lung images. With this sequence, performance bias owing to learning effect was minimized because the mediastinum images had an overall worse image quality.
After these sessions, three observers, including two authors (W.J.H.V., J.G.), additionally scored the mediastinum images obtained with the Bucky screen-film radiography system. The corresponding set of six images was evenly distributed among the observers, and each observer scored two different images.
To compare CCD digital radiography soft-copy reading with AMBER film hard-copy reading, the same three observers scored the CCD digital radiography mediastium and lung images more than 2 months later from the monitor. These images were the same as those that were scored from hard copy by the observers. The observers indicated the dot-containing corners by using a pencil and a template paper. The images were fully displayed by using the approximate full size of the monitor. Observers were not allowed to use other soft-copy tools. For the lung setting, the window width and window level were the same as those used for printing hard-copy images. This setting was the default window width and level used in clinical practice for an average-sized patient. Also, for the mediastinum setting, a clinically relevant window width and window level were chosen (ie, the window width was kept unchanged, but the window level was adjusted so that it was equal to the mean pixel value for these images). Observers were not allowed to adjust the window width or level settings. For dot detection, it would have been advantageous to choose a very small window width because the image background is uniform. The use of such a small window width, however, is not clinically relevant, and the corresponding results would therefore be biased in comparison with screen-film radiography results.
MTF Measurements
Measurements of the presampled MTF were performed by two authors (W.J.H.V., J.G.) who used the method described by Samei et al (11). An attenuating, thin-edge test device was fabricated for measurements. A 0.2-mm-thick tungsten layer (Goodfellows, Cambridge, England) was used as the attenuating material. Tungsten was preferred instead of lead because tungsten has a higher attenuation coefficient. With tungsten, a thinner layer can be used, which reduces the influence of penumbra effects. The layer measured 5 x 10 cm, with a purity of 99.95%. For such thickness, the attenuation was assessed to be 84% for a typical 133-kVp polychromatic x-ray beam by using a spectrum generator (Institute of Physics and Engineering in Medicine, York, England) (12). The edge was created with wire electrical discharge machining and resulted in an accurate straight edge. The edge was polished with light pressure and high precision by using a wet stone. The edge was placed with a small vertical and horizontal angle at the center of the detector. A 2.5-cm-thick aluminum filter was inserted just behind the x-ray tube. The same filter was used in the complete calibration procedure.
Contrast Measurements
The dynamic range and optical density were measured with image contrast, 
C, as a function of hole depth. Image contrast was measured by one author (W.J.H.V.) for both the lung and mediastinum configurations by using a densitometer (X-Rite, Grand Rapids, Mich). Contrast was computed as C = 1 – 10–OD, where OD is the differential optical density of the hole with the largest diameter (2). Mean contrast values were obtained from two AMBER and two Bucky screen-film radiography hard-copy images and from two CCD digital radiography hard-copy images.
Radiation Exposure
Dose measurements were obtained by two authors (W.J.H.V., J.G.). Entrance dose was calculated with a 15-cm3 ionization chamber (Keithley Instruments, Cleveland, Ohio). In one experiment, we derived the time-dependent entrance dose from the dose-area product with respect to the CCD digital radiography system when scanning an average-sized patient (length, 170 cm; weight, 70 kg) and including the prescan. The dose-area product was measured with a dosimeter (Diamentor M4; PTW-Freiburg, Freiburg, Germany). For AMBER, entrance doses for the lung and mediastinum configurations were measured separately. This was necessary because the AMBER system increases the radiation dose to the patient when scanning the mediastinum.
Data and Statistical Analysis
Image evaluation was performed by one author (W.J.H.V.) by placing a matching reference standard paper over the annotated images. The detection probability for each cell was determined from the six different scoring results for each system and for either simulated configuration (lung or mediastinum). We used a model-based interpolation scheme to fit a curve through the observer data, as described by Karssemeijer and Thijssen (13). With this model, the probability of detecting a hole with certain diameter as a function of its depth is described by means of a psychometric curve with probability range from 0.25 (chance) to 1.00. For hole diameter, a threshold depth was found that corresponded to 62.5% correct responses. This threshold was just halfway up the psychometric curve. Finally, a contrast-detail curve was fitted through the threshold depth values by using a second-order polynomial fit.
Differences between estimated threshold depths in the contrast-detail curves were calculated for each hole diameter. Pair-wise testing was performed for the lung and mediastinum configurations by using the Wilcoxon signed rank test to evaluate the performance of CCD digital radiography with respect to either AMBER or Bucky screen-film radiography and to assess contrast-detail observations and the differences in contrast range between hard-copy images obtained with CCD digital radiography and AMBER (14). Testing was performed by a statistician (B.J.A.M.). We evaluated eight tests in total. The Wilcoxon signed rank test is a nonparametric test. This means that a conservative approach is used. For each comparison, the P value, the Wilcoxon signed rank test statistic (V), and the number of paired samples (n) were determined. To correct for multiple testing, we suggest applying a Bonferroni correction to the results. Because we present eight tests, this implies a significance level of .006 (for each individual test) instead of .05 (for example) to maintain a global .05 significance level across all tests included. With respect to this adjustment, readers should note that the Bonferroni method is a rather conservative approach to adjust for the multiple testing problem. We also remind readers that the Bonferroni threshold (ie, .006) should, of course, not be used as a rigid cutoff value, just as the usual level (ie, .05) should never be taken as fixed decision boundary in individual tests. Rather, the .006 value is presented here to give some guidance on the order of magnitude of adjustment needed in interpreting P values and, thus, in helping readers with their own interpretation of the data analysis.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
|
|
|
|
|
MTF Measurements
The MTF results for images obtained with CCD digital radiography were measured with the thin-edge device (Fig 4). The Nyquist frequency, which is the highest frequency that a digital image can unambiguously represent, is depicted in Figure 4. The Nyquist frequency of the CCD digital radiography system is 3.1 cycles per millimeter for 162-µm pixel size. In both the scanning and slot direction, the MTF values differ only slightly. For the lowest frequency (0.0–1.0 cycles per millimeter), the slot direction MTF seems to be somewhat better, whereas, for the higher frequency (>2.5 cycles per millimeter), the scanning direction MTF seems slightly superior.
|
C became too small to be measured properly. The C values for hard-copy images obtained with CCD digital radiography were lower than those for hard-copy images obtained with conventional AMBER for both the lung and mediastinum configurations (P = .020 and P = .060, respectively).
|
|
|
C values of hard-copy images obtained with CCD digital radiography were lower than the C values of hard-copy images obtained with conventional Bucky screen-film radiography (P = .020). For the mediastinum configuration, the C values of hard-copy images obtained with Bucky screen-film radiography were lower than the C values of hard-copy images obtained with CCD digital radiography (P = .020).
Radiation Exposure
The entrance dose for CCD digital radiography is about 40% higher than the entrance dose for Bucky screen-film radiography (0.089 vs 0.064 mGy) (Table 2). The entrance doses for AMBER and Bucky screen-film radiography were similar in the lung configuration; entrance doses for AMBER, however, were higher in the mediastinum configuration. The time-dependent entrance dose measurement for the CCD digital radiography system was also determined (Fig 6). The entrance dose of the prescan is 7% (2.77 mGy/cm2) of the total entrance dose (40.10 mGy/cm2).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
MTF values for the CCD digital radiography system show good resolution capability in the scanning direction, as well as in the slot direction. Values of 0.70 at 1 cycle per millimeter and 0.35 at 2 cycles per millimeter for CCD digital radiography are greater than MTF values for conventional screen-film radiography. To compare, MTF values for images obtained with a 400-speed screen-film combination (Lanex regular screens and T-mat-G 65500 film; Eastman Kodak) is roughly 0.60 at 1 cycle per millimeter and 0.30 at 2 cycles per millimeter (4). Thus, the resolution capability of the chest CCD digital radiography system is at least comparable to that of screen-film radiography systems.
Findings from previous studies on the performance of digital flat-panel chest imaging systems versus conventional chest radiography systems have shown equal or better results with the digital detector systems (2,17). Aufrichtig (2) demonstrated that, compared with standard chest hard-copy images, unprocessed digital hard-copy images had improved contrast-detail detectability at a similar radiation dose. Chotas and Ravin (17) found better contrast-detail results with postprocessed digital hard-copy images than with conventional film images, with small signals of low inherent subject contrast levels detected more often on the digital images, even at reduced exposure levels. In the present study, CCD digital radiography results were also better than the conventional Bucky screen-film radiography results and were at least comparable to AMBER results. Results of studies in which digital radiography systems have been compared with AMBER systems for image quality have, as far as we know, not been published before.
In the present study, the specific attenuation and scatter characteristics of lungs and mediastinum were taken into account separately. This study design was chosen for optimal comparison between the imaging modalities. The new CCD digital radiography unit was compared not only with a standard conventional screen-film radiography (Bucky) unit but also with the AMBER system. This was done because the AMBER system was specially designed for improved visualization of the mediastinal area in the chest (18). Good performance of the CCD digital radiography system was also expected in the mediastinal area because of the relatively large dynamic range of the digital system. Accordingly, contrast-detail curves showed a lower threshold depth for CCD digital radiography than for Bucky screen-film radiography in the lung and mediastinum configuration.
For the mediastinum configuration, the difference was found not to be statistically significant, but this was probably because of the smaller number of paired data points (ie, six) that could be derived from the contrast-detail test object for Bucky screen-film radiography compared with CCD digital radiography for this configuration. This limited the statistical power of the Wilcoxon signed rank test. The result for AMBER hard-copy reading was better than the result for CCD digital radiography hard-copy reading in the mediastinum configuration. This can be explained by adjustment of the radiation dose during scanning with the AMBER system, which was doubled compared with the radiation dose of the CCD digital radiography system. This higher dose causes an improved signal-to-noise ratio for AMBER hard-copy images, with consequently better contrast-detail performance. The superior performance for AMBER hard-copy reading versus CCD digital radiography hard-copy reading, however, disappeared with CCD digital radiography soft-copy reading. CCD digital radiography soft-copy reading, which is the preferred medical reporting method in clinical practice, was found to be better than CCD digital radiography hard-copy reading. Also, CCD digital radiography soft-copy reading was found to be superior to AMBER hard-copy reading for the lung configuration. Moreover, no difference was found between the CCD digital radiography soft-copy reading and AMBER hard-copy reading techniques for the mediastinum configuration, despite the double entrance dose for AMBER with this configuration. This effect was expected because C was significantly lower for hard-copy images obtained with CCD digital radiography than for hard-copy images obtained with AMBER. The limitation of reduced contrast in digital hard copy disappears when reading digital soft copy. Accordingly, we have shown that the diagnostic potential improves with CCD digital radiography soft-copy reading.
In this study, the clinical protocols of the systems, which were established by radiologists, physicists, and manufacturers together, reflect common clinical practice. The use of matched entrance doses (and equal beam quality) instead of clinical protocols is not feasible because AMBER modulates the beam intensity in response to measurements from a detector located behind the film cassette. Moreover, the effective dose for CCD digital radiography (0.014 mSv) has been shown to be comparable to that of AMBER (0.016 mSv) in previous studies (19). Both modalities have a higher radiation exposure compared with Bucky screen-film radiography (0.009 mSv) (19). The purpose of AMBER is to provide optimal recording of the chest on hard-copy images without under- or overexposed areas. Establishing Bucky screen-film radiography with the same effective dose as AMBER would cause suboptimal imaging with overexposed areas.
During the implementation of CCD digital radiography in our hospital, the effective dose level that was chosen for CCD digital radiography was comparable to that of AMBER. In this way, we attempted to achieve appropriate image quality that was comparable to that of AMBER. In later stages, dose optimization with CCD digital radiography will be investigated.
Study Limitation
The Bucky screen-film images in the mediastinum configuration were initially left out of the evaluation because of the inherently poor image quality; these images were later scored for comparison with CCD digital radiography and AMBER images. This may have resulted in a performance bias owing to learning effect, which would have been to the advantage of the Bucky screen-film images. The results for these images, however, remained poor compared with the results for CCD digital radiography and AMBER. Thus, the delayed evaluation of Bucky screen-film images in mediastinum configuration had no effect on the overall study outcome.
In this study, a number of systems were evaluated on a pair-wise basis to investigate the performance of CCD digital radiography compared with that of other systems. An omnibus test could have been used in this study to look at all comparisons. It is crucial to note, however, that the problem discussed in this article is for the specific comparison of CCD systems and other machines. Thus, a precise set of specific hypotheses tests is known and specified in advance. In this context, an omnibus test would be a distraction of the approach in this study. To correct for multiple testing in this study, a Bonferroni correction was applied.
We found lower contrast values for hard-copy images obtained with CCD digital radiography than for those obtained with AMBER and Bucky screen-film radiography. We used a clinically relevant window width and window level setting, as determined by the CCD digital radiography system. The window width, however, was rather wide (ie, moderately overestimated by the system) to ensure that all information would be inside the window. This resulted in moderately lower contrast. In daily clinical practice, the window width and window level setting will eventually be optimized by hand. Furthermore, the fact that 8-bit printing was used may have also limited optimal use of the lookup table for printing.
Practical application: To the best of our knowledge, the dedicated ThoraScan chest system used in the present study is the first digital full-field chest imaging system that is based on a CCD detector without optical demagnification. The contrast-detail study showed good imaging quality for the CCD digital radiography system when hard-copy images were used. The performance of the CCD digital radiography system was found to be superior to the performance of the conventional Bucky screen-film radiography system. For soft-copy reading, CCD digital radiography performance was comparable to that of state-of-the-art AMBER. Compared with conventional techniques, the CCD digital radiography system has potential for improved chest imaging.
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
The physical properties of the CCD detector system are presented in Table A1. The detector is an assembly of eight CCD chips that are implemented side by side. The CCD detector is sensitive at a width of 44.0 cm and at a height of 10.8 mm. The basic detector elements in the CCD measure 27 x 27 µm, but downsampling decreases the final image resolution. The standard resolution mode features a pixel size of 162 x 162 µm. For future implementation, the advantages of a high-resolution mode of 81 x 81 µm will be evaluated. The CCD chips are covered with a fiberoptic plate that protects the CCDs against the x-rays and optically couples the phosphor (scintillator) with the CCD. The scintillator layer that converts the x-rays into visible light is deposited on the fiberoptic plate and consists of thallium-doped cesium iodide.
|
Time-Delay Integration
The CCD readout operates in time-delay integration mode. With this technique, a charge "packet" is transferred down a CCD column from cell to cell. At each CCD cell, the charge packet accumulates additional signal during a short time, 
. This process of accumulation and transfer is repeated until the charge packet is transferred down n detector rows (eg, with n equals 400, there are 400 27-µm detector elements along the 10.8-mm detector width). The CCD is also moved at a velocity equal to the average transfer rate of the charge packet but in the opposite direction. As a result, the charge packet accumulates signal originating from an approximately fixed point. Time-delay integration acquisition allows shorter scanning time and, thus, lower tube load than simple linear array acquisition. For time-delay integration, the total integration time is n. To achieve the same signal level for a pixel with a linear array, the scanning time would have to be a factor n longer than would be required for a multilinear detector operating in time-delay integration mode (22).
Image Processing
Three basic types of corrections are applied to the raw image data, namely, direct current voltage offset, dark current correction, gain correction, and edge correction. With direct current voltage offset and dark current correction, a correction matrix is constructed that captures the characteristics of the detector columns with respect to voltage offset and thermally generated dark current in the CCDs. This compensation needs to be performed before every scan because the correction matrix depends on the internal temperature of the detector. After correction for dark current and direct current voltage offset, gain correction is needed to equalize the output from all columns. Gain correction is performed by taking samples from the detector output when exposed to a uniform level of radiation during the daily calibration procedure. Finally, edge correction is needed to compensate for the fact that the detector elements of the eight tiled CCDs are not in perfect physical contact at the edges of the CCDs. The resulting seven gaps are filled by interpolation with signal values from adjacent pixels. For final postprocessing, the ThoraScan system uses multiscale image contrast amplification (MUSICA; Agfa-Gevaert, Mortsel, Belgium) (20).
MTF Curve
The MTF curve describes the resolution capability of an imaging system. It is the ratio of output contrast to input contrast of an imaging system. Each step in the imaging process can contribute to a loss of contrast (21). Because the MTF of a scanning system might not be symmetric, a separate analysis has to be carried out in the scanning and slot (row) directions. The MTF in the slot direction and the MTF in the scanning direction are both determined by using the pixel aperture, the thallium-doped cesium iodide scintillator, and the fiberoptic plate. The MTF in the slot direction may also be influenced by imperfect colinearity of the detector columns in the scanning direction and by the charge transfer process of the serial register. The MTF in the scanning direction degrades further as a result of cesium iodide decay characteristics, imperfect column transfer efficiency, and asynchronism between scanning velocity and charge transfer timing (3,4).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, W.J.H.V.; study concepts, W.J.H.V., J.G., L.J.M.K.; study design, all authors; literature research, W.J.H.V., L.J.M.K.; experimental studies, W.J.H.V., J.G.; data acquisition, W.J.H.V., J.G., L.J.M.K.; data analysis/interpretation, all authors; statistical analysis, B.J.A.M.; manuscript preparation, W.J.H.V., L.J.M.K., J.G.; manuscript definition of intellectual content, all authors; manuscript editing, W.J.H.V., L.J.M.K.; manuscript revision/review, W.J.H.V., L.J.M.K., J.G.; manuscript final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  L J M Kroft, W J H Veldkamp, B J A Mertens, J-P A van Delft, and J Geleijns Dose reduction in digital chest radiography and perceived image quality Br. J. Radiol., December 1, 2007; 80(960): 984 - 988. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Suryanarayanan, A. Karellas, S. Vedantham, I. Sechopoulos, and C. J. D'Orsi Detection of Simulated Microcalcifications in a Phantom with Digital Mammography: Effect of Pixel Size Radiology, July 1, 2007; 244(1): 130 - 137. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Korner, C. H. Weber, S. Wirth, K.-J. Pfeifer, M. F. Reiser, and M. Treitl Advances in Digital Radiography: Physical Principles and System Overview RadioGraphics, May 1, 2007; 27(3): 675 - 686. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. J. M. Kroft, W. J. H. Veldkamp, B. J. A. Mertens, J. P. A. van Delft, and J. Geleijns Detection of simulated nodules on clinical radiographs: dose reduction at digital posteroanterior chest radiography. Radiology, November 1, 2006; 241(2): 392 - 398. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
), AMBER (
), and Bucky screen-film radiography (
). Hole sizes are (a) 1.0 mm and (b) 1.3 mm. Individual data points are derived from scoring data of six observers. Threshold depth corresponds to 62.5% (arrow) correct responses. Threshold 1 corresponds to CCD digital radiography, threshold 2 to AMBER, and threshold 3 to Bucky screen-film radiography. Note that a curve shift to the left indicates better detectability.
) are shown.
