Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2253011736
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Suryanarayanan, S.
Right arrow Articles by D’Orsi, C. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Suryanarayanan, S.
Right arrow Articles by D’Orsi, C. J.
(Radiology 2002;225:801-807.)
© RSNA, 2002

Experimental Studies

Flat-Panel Digital Mammography System: Contrast-Detail Comparison between Screen-Film Radiographs and Hard-Copy Images1

Sankararaman Suryanarayanan, MS2, Andrew Karellas, PhD2, Srinivasan Vedantham, PhD2, Hetal Ved, MS, Stephen P. Baker, MSc, PH and Carl J. D’Orsi, MD2

1 From the Departments of Radiology (S.S., A.K., S.V., C.J.D.) and Information Services (S.P.B.), University of Massachusetts Medical School, Worcester, Mass; and Department of Biomedical Engineering, Worcester Polytechnic Institute, Worcester, Mass (H.V.). Received October 24, 2001; revision requested January 4, 2002; revision received March 20; accepted May 13. Supported by National Institutes of Health, National Cancer Institute grant RO1-CA88792. Address correspondence to A.K., Department of Radiology, Emory University School of Medicine, 1364 Clifton Rd NE, Rm A133A, Atlanta, GA 30322 (e-mail: akarell@emory.edu).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX
 REFERENCES
 
PURPOSE: To compare the contrast-detail (CD) characteristics of screen-film (SF) and postprocessed digital images by using a phantom-based method.

MATERIALS AND METHODS: Images of a CD phantom with polymerized methyl methacrylate were acquired with SF and full-field digital mammography systems at matched exposure conditions. A four-alternative forced-choice experiment was conducted with seven observers participating in the study. Each observer was required to identify randomly located disks in phantom images from which detection curves were computed. The CD diagrams for the SF and digital systems were estimated from the detection curves and compared at 50% and 62.5% threshold levels. Furthermore, a theoretic model was used to estimate the CD performance of the SF and digital systems.

RESULTS: Analysis of covariance for mixed models was used with the natural logarithm of disk thickness as the dependent variable, the natural logarithm of disk diameter as the covariate, and the observer as a random factor. The results of statistical analysis indicated significant differences between the CD characteristics of SF and digital mammographic images at both 50% (P < .001) and 62.5% (P < .001) detection thresholds.

CONCLUSION: The authors conclude that digital CD curves, on average, exhibit threshold contrast characteristics that are lower (better) than those of SF mammography.

© RSNA, 2002

Index terms: Breast neoplasms, radiography, 00.32, 00.111, 00.121 • Breast radiography, comparative studies, 00.111, 00.121 • Breast radiography, technology, 00.111, 00.121 • Phantoms • Radiography, digital, 00.121


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX
 REFERENCES
 
The advent of digital mammography has catalyzed the development of a variety of detector technologies. Findings in studies show certain promising aspects of digital mammography and its potential effect on the early diagnosis and management of breast cancer (1–3). Although screen-film (SF) systems offer excellent spatial resolution and images of diagnostic quality, they are limited in their dynamic range and contrast characteristics. Recent developments in large-area flat-panel detector arrays have further promoted the case for digital mammography. The advantages of electronic detection are well known and include high detection efficiency, high dynamic range, capability for contrast enhancement (4), and postprocessing capabilities such as computer-aided diagnosis (5,6).

Findings in comprehensive investigations of the physical characteristics of an amorphous silicon–based flat-panel detector have demonstrated favorable spatial resolution and detective quantum efficiency characteristics (7,8) but have not addressed the contrast-detail (CD) performance in a rigorous manner. CD performance is a widely used quality control tool to assess clinical imaging systems. Results of a CD study comparing an SF and flat-panel digital chest system demonstrated improved observer performance when images were acquired with the digital system (9) and similar results with a prototype flat-panel digital detector (10). In other observer experiments, the CD characteristics of a mammographic small-field charge-coupled device were reported to have a substantial advantage over SF mammography (11).

To our knowledge, there is virtually no precedent of detailed scientific studies for evaluation of the CD characteristics of flat-panel-detector–based full-field digital mammography. Obenauer et al (12) performed an evaluation of a full-field digital mammography system with CD and receiver operating characteristic analysis. Findings in their study indicated that the digital system exhibited detection rates equivalent or higher than those for an SF system (12). The purpose of this study was to compare the CD characteristics of SF and postprocessed digital images by using a phantom-based method.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX
 REFERENCES
 
System Description
A clinical full-field digital mammography system (Senograph 2000D; GE Medical Systems, Milwaukee, Wis) was used to acquire the digital images in this study. The system comprises a columnar cesium iodide scintillator coupled to an amorphous silicon photodiode array with 1,900 x 2,304 elements and a pixel pitch of 100 µm, which provides a field of view of 19 x 23 cm. Light created as a result of the interaction of x-ray photons in the cesium iodide scintillator travels down the columnar crystalline structure of the scintillator to the detector pixels and is then converted to an electric signal. Each detector element (pixel) in the array is an individually addressable light detector, and the electric signals from the pixels are individually read and digitized to 14 bits per pixel by means of special-purpose low-noise electronics. The mammographic unit for the SF imaging was a clinical system (DMR; GE Medical Systems). Other than the digital detector, the x-ray–generating components of the two units are essentially equivalent.

CD Phantom
A commercially available CD phantom (CDMAM; Nuclear Associates, Carle Place, NY) was used in this study as the test object to compare the performance of the two systems. The phantom consists of a 0.5-mm-thick aluminum base, 15 x 23 cm in area, containing circular gold disks that are logarithmically sized from 0.10 to 0.80 mm in diameter and from 0.05 to 1.6 µm in thickness. The disks are centrally placed within a matrix of squares that form 16 rows and 16 columns. Within a row, the disk thickness is constant with logarithmically varying diameter. In addition, a disk that has the same thickness and diameter as those of the central disk corresponding to that square is randomly placed at one of the corners of each square (Fig 1).



View larger version (115K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Radiograph of the phantom acquired with the full-field digital mammography system illustrates the structured patterns in the phantom. The portion of the phantom that was used in this study was restricted to the area to the right of (below) the thick white line when the images were reviewed. This line does not exist physically in the phantom and is shown here to illustrate the region of interest.

 
For the purpose of this study, 4.0 cm of additional acrylic was added to bring the total thickness of the phantom to 4.5 cm. The half-value layer of the x-ray beam past the 4.5-cm phantom was equal to 0.6 mm of aluminum at 27 kVp and at 225 mAs with a molybdenum target–molybdenum filter combination, which matched the half-value layer with 5.0-cm-thick polymerized methyl methacrylate without the aluminum base as measured in identical conditions. Furthermore, with automatic exposure settings, the 4.5-cm phantom resulted in the same kilovolt peak, milliampere second, and target-filter combinations as those with 5.0-cm-thick polymerized methyl methacrylate. This thickness is consistent with the mean compressed breast thickness of 5.1 cm computed for 4,510 cases in our mammography database. The thickness is also supported by recent data on mean and median compressed breast thicknesses on the order of 5.1 cm that have been reported by Kruger and Schueler (13).

Image Acquisition
All SF images were acquired with an SF system (Min-R 2000; Eastman Kodak, Rochester, NY), and three images of the phantom were acquired at 27 kVp and 225 mAs with a molybdenum target– molybdenum filter combination. The exposure technique was obtained by allowing the system to automatically choose the kilovolt peak and milliampere second settings in the most commonly used "contrast-auto mode," which is designed to generate an exposure with optimal contrast. This mode of acquisition, with an emphasis on image quality rather than radiation dose, was used to obtain SF images with desirable exposure conditions. The recorded milliampere second value was consistent with what was expected for this equivalent thickness. An identical exposure technique was used to acquire three digital images of the phantom with the window and level settings adjusted collaboratively by two board-certified experienced radiologists (C.J.D.) before they were printed with a dry film printer (DryView 8600; Eastman Kodak).

The average background optical densities of the SF images and printed digital hard copies were 1.63 and 2.28, respectively, which met the minimum requirement for optical density of 1.5. Contrast matching of the targets (disks) was not attempted between the SF and digital images because the objective was to operate each modality in its optimal configuration. A mere matching of the contrasts of the SF and digital images would have led to a gross underutilization of the image enhancement capabilities of the digital system. Furthermore, we used disk thickness as a measure of stimulus rather than as an absolute measure of contrast and thereby obviated contrast matching between the targets.

Display adjustment with window and level functions was used to enhance the digital images before the hard copies were printed. This type of image enhancement is used routinely in a clinical setting when digitally acquired images are printed. Hard copy images were used in this investigation to circumvent the problems associated with the limitations of gray-scale cathode ray tube monitors, such as temporal instability and calibration issues, and to provide a common display medium for comparison with SF radiographs. Six images were used to compare the CD characteristics with the SF system versus those with the digital mammography system.

Experimental Design and Observer Study
Seven observers participated in this study; each observer independently reviewed all the images in a dark room in one session. A transillumination image viewer was used, and a photometer (Photo.Meter LX; Quantum Instruments, Garden City, NY) was used to measure the light output from the viewer. The mean output luminance was about 3,000 candela per square meter during all observation sessions, and background illuminance was 10 lux. The region around the image area was masked to prevent any stray light from affecting observer performance. An area that encompassed disk diameters ranging from 0.10 to 0.80 mm and thickness ranging from 0.05 to 1.6 µm was marked on each image (Fig 1), and the observers were asked to restrict their observations to this region. This region provided an adequate data range for comparison and also helped to maintain reasonable observation times, which ranged from 2 to 3 hours on average. The observers were not told about the objective of the study, and the images were presented to the observers in a random fashion to reduce systematic errors.

The phantom provided an opportunity to conduct a four-alternative forced-choice experiment because four alternative choices were presented with each square as a result of the randomly located corner disk. The observers were asked to read one column at a time, starting with the disk with the largest diameter and proceeding toward the smallest perceivable disk in that column. Since this was a forced-choice study, they were asked to arrive at their "best estimate" for the location of the disk in each square in situations where the disks were not perceivable. A template was provided that represented the portion of the phantom image under observation, and the observers were asked to mark the location of the corner disk in each square on this template. The observers were asked to refrain from looking at previously marked sheets to circumvent learning effects. A magnifying glass was provided to the observers to aid them in their observations, as is done in standard mammographic image reading.

Computation of CD Characteristics from Experimental Data
For each disk diameter, the "percentage correct," or the fraction of correct detection data, was computed for every observer, as described by several authors (9,10,14,15). It is well known that data from experiments to measure the probability of correct response on the basis of human observers can be fit to a Gaussian integral model (14,15). We incorporated a similar approach and used a maximum-likelihood parameter estimation scheme to fit each observer’s data to a Gaussian integral function for all the diameters (14,15). The logarithmic value of each disk thickness was used as the stimulus level (SL), which is defined as the following:

where the parameter {alpha} represents the stimulus intensity at which the slope of the function is maximum, ß represents the steepness of the function, and x is the input stimulus value. A Gaussian integral function of the following type,

was used to fit the mean values of the percentage correct detection [P(x)] data by setting {gamma}, the probability of correct response due to chance alone, to 0.25. as this was a four-alternative forced-choice experiment. The data values below 0.25 were ignored by the algorithm because the probability due to chance alone was 0.25 for this experiment. Suitable ranges were given to {alpha} and ß to encompass observer data and iteratively fit them. A complete description of the fitting function and its operation has been described by Harvey (16,17).

One hundred forty detection curves were obtained (10 diameters x 2 modalities x 7 observers) after all the observer data were analyzed. Threshold levels of 50% and 62.5% were chosen to obtain the CD data. A threshold level of 50% is commonly used in perception experiments, whereas a 62.5% threshold gives a level midway between chance (0.25 for a four-alternative forced-choice experiment) and 100% detection. For each observer, the CD characteristics for each modality (SF or digital) were generated by means of vertical projection of the data points on each detection curve at a given threshold (eg, 50%) onto the stimulus axis (Fig 2). The individual CD curves were then averaged to obtain mean CD data for each modality.



View larger version (24K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Line graph shows the detection characteristics for one observer with the digital system and three disk diameters. The horizontal line represents a 50% threshold level. The projection of the vertical dotted lines on the x axis gives disk depths that correspond to different disk diameters. Detection characteristics were obtained for all disk diameters and were used to generate the CD characteristics. A similar procedure was adopted to obtain the CD characteristics generated by all the observers with both SF and digital systems at threshold levels of both 50% and 62.5%.

 
Theoretic CD Characteristics
Statistical decision theory has been widely used to predict the dependence of threshold signal contrast on object size. Many authors have used nonprewhitening matched-filter models to understand the CD behavior of imaging modalities (10,14,18–20). We used a non-prewhitening matched-filter method to model the CD performance characteristics (Fig 3) of the SF and digital mammography systems investigated in this study as

A detailed description of the model is provided in the Appendix.



View larger version (18K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Schematic shows the theoretic model used to model the signal and noise transfer processes that occur when an observer perceives spatial information (Appendix).

 
Statistical Analysis
The experimental CD data from the observers were pooled to compute mean disk thickness values. Distribution characteristics of perceived disk thickness were evaluated by means of the Kolmogorov-Smirnov one-sample test for normality and graphic plotting of histograms of residuals. The linearity of the association between the logarithm of both disk thickness and disk diameter was also assessed by means of comparisons among candidate transformation with maximal coefficients of determination. Transformation with natural logarithms resulted in the best conformity to normal distribution and linearity in the relationship. Analysis of covariance for mixed models was used with the natural logarithm of disk thickness as the dependent variable, the natural logarithm of disk diameter as the covariate, and the observer as a random factor. The statistical analysis was performed with software (SAS, version 8.0; SAS Institute, Cary, NC). A P value of less than .05 was considered to indicate a statistically significant difference.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 APPENDIX
 REFERENCES
 
The percentage correct detection characteristics for all seven observers with disk diameters of 0.80 and 0.40 mm are summarized in Figures 4 and 5. A comparison of sample percentage correct detection data for one observer that corresponds to diameters of 0.80 and 0.40 mm with both digital and SF systems is shown in Figure 6. Significant differences were observed between the experimental CD characteristics of the SF and digital mammography systems at the 50% (P < .001) and 62.5% (P < .001) threshold levels. For the SF system, perceived disk thickness data that corresponded to disk diameters of 0.10 and 0.13 mm could not be computed because the algorithm did not converge as a result of excessive variability in these data sets. Lower (better) threshold contrast characteristics were observed with the digital modality at both the 50% and 62.5% threshold levels, as shown in Figures 7 and 8, respectively. Similar results were obtained with the model-predicted CD performance.



View larger version (33K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4a. Line graph shows the percentage correct detection characteristics for all observers with 0.80-mm disk diameter for (a) digital and (b) SF mammography. Markers represent experimental data points for each observer, while the smooth lines represent estimated detection characteristics.

 


View larger version (32K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4b. Line graph shows the percentage correct detection characteristics for all observers with 0.80-mm disk diameter for (a) digital and (b) SF mammography. Markers represent experimental data points for each observer, while the smooth lines represent estimated detection characteristics.

 


View larger version (34K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5a. Line graph shows the percentage correct detection characteristics for all observers with 0.40-mm disk diameter for (a) digital and (b) SF mammography. Markers represent experimental data points for each observer, while the smooth lines represent estimated detection characteristics.

 


View larger version (33K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5b. Line graph shows the percentage correct detection characteristics for all observers with 0.40-mm disk diameter for (a) digital and (b) SF mammography. Markers represent experimental data points for each observer, while the smooth lines represent estimated detection characteristics.

 


View larger version (25K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6a. Line graph shows the percentage correct detection characteristics for one observer with disk diameters of (a) 0.80 and (b) 0.40 mm for both digital and SF mammography. The observer achieved better detection characteristics with the digital system.

 


View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6b. Line graph shows the percentage correct detection characteristics for one observer with disk diameters of (a) 0.80 and (b) 0.40 mm for both digital and SF mammography. The observer achieved better detection characteristics with the digital system.

 


View larger version (25K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7. Line graph shows the mean CD characteristics obtained at the 50% threshold level for both SF and digital systems. On average, the digital system exhibits threshold contrast characteristics that are lower (better) than those for the SF system, which indicates higher fidelity for the digital images. The smooth lines indicate model-predicted CD diagrams for the SF and digital systems. The error bars indicate SD; depth = thickness.

 


View larger version (26K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 8. Line graph shows mean CD characteristics obtained at the 62.5% threshold level for both SF and digital systems. On average, the digital system exhibits threshold contrast characteristics that are lower (better) than those for the SF system, which indicates higher fidelity for the digital images. The smooth lines indicate model-predicted CD diagrams for the SF and digital systems. The error bars indicate SD; depth = thickness.

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
APPENDIX
 REFERENCES
 
Use of the CD method to evaluate clinical imaging systems is widely accepted. The focus of this study was to develop and use a technique to reliably estimate the physical and imaging quality characteristics of clinical x-ray imaging systems, especially mammography, and not to determine detection in the presence of structured noise. Alternative forced-choice methods provide effective means of performing psychophysical measurements and have been used widely with synthetic images (10,14,15,20–23). Alternative forced-choice techniques are convenient to use in experiments such as these, in which the objective is to compare or rank modalities. Many observer experiments are performed with a receiver operating characteristic technique, which is economical when images are scare and cannot be easily reproduced. In experiments such as the present study, however, it has been reported that alternative fixed choice is preferred because the experiments can be carefully controlled and a wider range of signal-to-noise ratios can be investigated (21).

Because signal location was randomized in the present study, we believe that it could have mitigated observer learning to a great extent compared with that with more simplistic CD experiments in which the location of the signal is fixed. Furthermore, the multitude of disks in the phantom makes it difficult to memorize locations. Forced-choice methods also provide an opportunity to evaluate the CD characteristics at multiple detection threshold levels, which is not possible with conventional CD experiments. We found that in general, the predictions of the theoretic model agreed well with the experimentally obtained CD data. The CD characteristics obtained with the SF system were higher than predicted, which could be due to the wider window settings of the SF images that decrease the displayed image noise and consequently increase the contribution of the internal noise of the observer (23). Furthermore, we used one visual response function in this study and ignored the fact that different observers have slightly different responses. Measurement of individual visual response functions was beyond the scope of this study. However, we noticed that if the visual response function was varied so that it peaked at higher frequencies, then the agreement between the predicted and experimental CD data improved (14).

Detection characteristics in the presence of structured noise have been investigated (20,24,25). Findings in other studies show that structured noise affects detection and thereby increases threshold contrast levels (15). In fact, an increase in threshold contrast with increasing lesion diameter has been reported for mammography in the presence of structured noise (20,24). Structured noise could potentially interfere with subtle attenuation differences that degrade contrast sensitivity, but image contrast can be enhanced with digital images, which has been shown to increase detection in other studies (23). This indicates promising potential for digital mammography because a variety of image enhancement operations can be performed easily. However, the effects of image enhancement techniques in terms of detection and threshold contrast for digital mammography in the presence of structured noise needs to be investigated further.

The superior CD characteristics of the digital system can be attributed to its high detective quantum efficiency and dynamic range (7), which can be expected to have a direct effect on image quality and affect the contrast characteristics of the resulting images. The lower (better) threshold contrast characteristics exhibited with the digital mammography system might lead to acquisition of images with improved diagnostic quality with no increase in radiation dose. With further improvements, a reduction in radiation dose may be attainable.

Practical application: We describe a technique that could be used to estimate and compare the CD characteristics of clinical x-ray imaging systems from a physical standpoint. Study of CD characteristics of an imaging system by using phantoms provides a powerful objective method for gaining useful insight into the information capacity of an imaging system (26). It is well known that this type of performance evaluation is not a substitute for clinical trials (27), but it serves an important role in the quantitative understanding of imaging system performance. The CD method can be extremely useful as a precursor study to determine system performance or compare the performance among different systems before clinical trials (28–31). CD evaluation may reveal strengths or deficiencies in the system that may need to be remedied before the clinical studies begin. Moreover, studies such as these may be used as an effective quality assurance tool or as a means of comparing various imaging technologies.


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX
REFERENCES
 
A block diagram of the model used in this investigation is shown in Figure 3. For a signal-known-exactly problem, the threshold signal-to-noise ratio (SNRT) can be described as

This formula can be simplified as

where CT is the threshold contrast, S(u,v) is the frequency response of the circular targets. The latter is computed as

where d is the disk diameter, J1 is the first-order Bessel function, and MTF(u,v) is the two-dimensional spatial frequency response of the imaging system obtained from the one-dimensional response, assuming rotational symmetry, and HVS(u,v) is the frequency response of the human visual system modeled as HVS(f) = fe-ßf, with a peak value around four cycles per degree (20) calculated by assuming a viewing distance of 40 cm. DQE(u,v) is the detective quantum efficiency, A is the large area signal, and q is the incident x-ray fluence. Hence, the threshold contrast can be computed as the following:

The MTF and detective quantum efficiency of the digital system were obtained from a prior investigation (7), while the SF data (32) were obtained from the manufacturer. The model-predicted CD diagrams were scaled by a constant to match the experimental SF and digital CD data (10).


   FOOTNOTES
 
2 Current address: Department of Radiology, Emory University School of Medicine, Atlanta, Ga. Back

Abbreviations: CD = contrast-detail, SF = screen-film

Author contributions: Guarantors of integrity of entire study, S.S., A.K.; study concepts, S.S., A.K., C.J.D.; study design, S.S., H.V.; literature research, S.S., A.K.; experimental studies, S.S., S.V., H.V.; data acquisition, S.S., H.V.; data analysis/interpretation, S.S., A.K., C.J.D.; statistical analysis, S.P.B., S.S.; manuscript preparation, S.S., S.V., H.V.; manuscript definition of intellectual content, A.K., S.S., C.J.D.; manuscript editing, S.S., A.K.; manuscript revision/review and final version approval, A.K., C.J.D.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX
 REFERENCES
 

  1. Yaffe MJ, Nishikawa RM, Maidment ADA, Fenster A. Development of a digital mammography system. SPIE Proc Med Imaging 1988; 914:182-188.
  2. Shtern F. Digital mammography and related technologies: a perspective from the National Cancer Institute. Radiology 1992; 183:629- 630.[Abstract/Free Full Text]
  3. Nishikawa RM, Mawdsley GE, Fenster A, Yaffe MJ. Scanned-projection digital mammography. Med Phys 1987; 14:717-727.[CrossRef][Medline]
  4. Bick U, Giger ML, Schmidt RA, Nishikawa RM, Doi K. Density correction of peripheral breast tissue on digital mammograms. RadioGraphics 1996; 16:1403-1411.[Abstract]
  5. Kallergi M, Carney GM, Gaviria J. Evaluating the performance of detection algorithms in digital mammography. Med Phys 1999; 26:267-275.[CrossRef][Medline]
  6. Kupinski MA, Giger ML. Automated seeded lesion segmentation on digital mammograms. IEEE Trans Med Imaging 1998; 17:510-517.[CrossRef][Medline]
  7. Vedantham S, Karellas A, Suryanarayanan S, et al. Full breast digital mammographic imaging with an amorphous silicon-based flat panel detector: physical characteristics of a clinical prototype. Med Phys 2000; 27:558-567.[CrossRef][Medline]
  8. Vedantham S, Karellas A, Suryanarayanan S, D’Orsi CJ, Hendrick RE. Breast imaging using an amorphous silicon based full-field digital mammography system: stability of a clinical prototype. J Digit Imaging 2000; 13:191-199.
  9. Chotas HG, Ravin CE. Digital chest radiography with a solid-state flat-panel x-ray detector: contrast-detail evaluation with processed images printed on film hard copy. Radiology 2001; 218:679-682.[Abstract/Free Full Text]
  10. Aufrichtig R. Comparison of low contrast detectability between a digital amorphous silicon and a screen-film based imaging system for thoracic radiography. Med Phys 1999; 26:1349-1358.[CrossRef][Medline]
  11. Liu H, Fajardo LL, Barrett JR, Baxter RA. Contrast-detail detectability analysis: comparison of a digital spot mammography system and an analog screen-film mammography system. Acad Radiol 1997; 4:197-203.[CrossRef][Medline]
  12. Obenauer S, Hermann KP, Schorn C, Fischer U, Grabbe E. Digitale vollfedmammographie: dosisabängige detektion von simulierten herdbefunden und mikrokalzifikationen. Experimentelle Radiologie 2000; 172:1052-1056.
  13. Kruger RL, Schueler BA. A survey of clinical factors and patient dose in mammography. Med Phys 2001; 28:1449-1454.[CrossRef][Medline]
  14. Ishida M, Doi K, Loo LN, Metz CE, Lehr J. Digital image processing: effect on detectability of simulated low-contrast radiographic patterns. Radiology 1984; 150:569-575.[Abstract/Free Full Text]
  15. Ohara K, Doi K, Metz CE, Giger ML. Investigation of basic imaging properties in digital radiography. 13. Effect of simple structured noise on the detectability of simulated stenotic lesions. Med Phys 1989; 16:14-21.
  16. Harvey LO, Jr. Efficient estimation of sensory thresholds. Behav Res Methods Instrum Comput 1986; 18:623-632.
  17. Harvey LO, Jr. Efficient estimation of sensory thresholds with ML-PEST. Spat Vis 1997; 11:121-128.[Medline]
  18. Wagner RF, Brown DG, Pastel M. Application of information theory to the assessment of computed tomography. Med Phys 1979; 6:83-94.[CrossRef][Medline]
  19. Seltzer SE, Judy PF, Swensson RG, Chan KH, Nawfel RD. Flattening of the contrast-detail curve for large lesions on liver CT images. Med Phys 1994; 21:1547-1555.[CrossRef][Medline]
  20. Burgess AE, Jacobson FL, Judy PF. Human observer detection experiments with mammograms and power-law noise. Med Phys 2001; 28:419-437.[CrossRef][Medline]
  21. Burgess AE. Comparison of receiver operating characteristics and forced choice observer performance measurement methods. Med Phys 1995; 22:643-655.[CrossRef][Medline]
  22. Herrmann C, Buhr E, Hoeschen D, Fan SY. Comparison of ROC and AFC methods in a visual detection task. Med Phys 1993; 20:805-811.[CrossRef][Medline]
  23. Ohara K, Chan HP, Doi K, Giger ML, Fujita H. Investigation of basic imaging properties in digital radiography. 8. Detection of simulated low-contrast objects in digital subtraction angiographic images. Med Phys 1986; 13:304-311.
  24. Chakraborty DP. Anomalous nodule visibility effects in mammographic images. Proc SPIE Med Imaging Image Perception Performance 2001; 4324:68-76.
  25. Eckstein MP, Whiting JS. Lesion detection in structured noise. Acad Radiol 1995; 2:249-253.[CrossRef][Medline]
  26. Kundel HL. Images, image quality and observer performance: New Horizons in Radiology lecture. Radiology 1979; 132:265-271.[Abstract]
  27. Lewin JM, Hendrick ER, D’Orsi CJ, et al. Comparison of full-field digital mammography with screen-film mammography for cancer detection: results of 4,945 paired examinations. Radiology 2001; 218:873-880.[Abstract/Free Full Text]
  28. Barrett HH, Abbey CK, Clarkson E. Objective assessment of image quality. III. ROC metrics, ideal observers, and likelihood-generating functions. J Opt Soc Am A Opt Image Sci Vis 1998; 15:1520-1535.
  29. Chakraborty DP. Quantitative versus subjective evaluation of mammography accreditation phantom images. Med Phys 1995; 2:133-143.
  30. Chakraborty DP. Computer analysis of mammography phantom images (CAMPI): an application to the measurement of microcalcification image quality of directly acquired digital images. Med Phys 1997; 8:69-1277.
  31. Hall TJ, Insana MF, Harrison LA, Soller NM, Schlehr KJ. Ultrasound contrast-detail analysis: a comparison of low-contrast detectability among scanhead designs. Med Phys 1995; 22:1117-1125.[CrossRef][Medline]
  32. Bunch PC. Advances in high-speed mammographic image quality 1999. Proc SPIE Med Imaging Phys Med Imaging 1999; 3659:120-130.



This article has been cited by other articles:


Home page
 RadiologyHome page
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]

Home page
 RadiologyHome page
M. Penedo, M. Souto, P. G. Tahoces, J. M. Carreira, J. Villalon, G. Porto, C. Seoane, J. J. Vidal, K. S. Berbaum, D. P. Chakraborty, et al.
Free-Response Receiver Operating Characteristic Evaluation of Lossy JPEG2000 and Object-based Set Partitioning in Hierarchical Trees Compression of Digitized Mammograms
Radiology, November 1, 2005; 237(2): 450 - 457.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
P. Skaane, C. Balleyguier, F. Diekmann, S. Diekmann, J.-C. Piguet, K. Young, and L. T. Niklason
Breast Lesion Detection and Classification: Comparison of Screen-Film Mammography and Full-Field Digital Mammography with Soft-copy Reading--Observer Performance Study
Radiology, October 1, 2005; 237(1): 37 - 44.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
S. Suryanarayanan, A. Karellas, S. Vedantham, S. M. Waldrop, and C. J. D'Orsi
Detection of Simulated Lesions on Data-compressed Digital Mammograms
Radiology, July 1, 2005; 236(1): 31 - 36.
[Abstract] [Full Text] [PDF]

Home page
 RadiologyHome page
P. Skaane and A. Skjennald
Screen-Film Mammography versus Full-Field Digital Mammography with Soft-Copy Reading: Randomized Trial in a Population-based Screening Program--The Oslo II Study
Radiology, July 1, 2004; 232(1): 197 - 204.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Suryanarayanan, S.
Right arrow Articles by D’Orsi, C. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Suryanarayanan, S.
Right arrow Articles by D’Orsi, C. J.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE