HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kroft, L. J. M. Articles by Geleijns, J. Search for Related Content PubMed PubMed Citation Articles by Kroft, L. J. M. Articles by Geleijns, J.

Detection of Simulated Nodules on Clinical Radiographs: Dose Reduction at Digital Posteroanterior Chest Radiography¹

¹ From the Departments of Radiology (L.J.M.K., W.J.H.V., J.P.A.v.D., J.G.) and Medical Statistics (B.J.A.M.), C2S, Leiden University Medical Center, Albinusdreef 2, 2333 ZA Leiden, the Netherlands. From the 2005 RSNA Annual Meeting. Received August 8, 2005; revision requested October 17; revision received November 14; accepted December 12; final version accepted February 3, 2006. Address correspondence to L.J.M.K. (e-mail: l.j.m.kroft{at}lumc.nl).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
Purpose: To determine to what extent dose reduction results in decreased detection of simulated nodules on patient digital posteroanterior (PA) chest radiographs.

Materials and Methods: Raw data from 20 clinical digital PA chest images that were reported as having normal findings and that were obtained with a slot-scan charge-coupled device system were used. For research protocol that concerns data with patient identities concealed, institutional review board approval is not required. One hundred twenty nodules varying in size and signal intensity were digitally simulated and added to the chest images. Hard copies were printed to represent a 100% dose and, by adding noise, to represent simulated patient doses of 50%, 25%, and 12%. Four radiologists reviewed images. Each lesion was registered as "detected" or "not detected." A semiparametric logistic regression model was used for statistical analysis.

Results: The decrease in radiation dose from 100% to 50%, 25%, or 12% had no effect on lesion detection in the lungs. The decrease in radiation dose had an effect on lesion detection in the mediastinum, as probabilities deteriorated from the 100% dose to the 50%, 25%, and 12% dose with each step. Probabilities of smaller detection rates when compared with that of the reference category (100% dose) were 0.97 (95% confidence interval [CI]: –0.86, 0.012) for the 50% dose, 1 (CI: –0.59, –0.61) for the 25% dose, and 1 (CI: –2.41, –1.22) for the 12% dose. CIs for the effects were on the log(odds). Detection probability decreased with smaller and lower signal intensity lesions.

Conclusion: At clinical digital radiography, dose reduction resulted in decreased observer detection of simulated nodules in the mediastinum but not in the lungs.

© RSNA, 2006

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
At imaging of the chest, performance of digital chest radiography in patients is equal or even superior to that of conventional screen-film radiography, even at dose reductions of 33%–59% (1–4). Consequently, an attractive option is to lower patient radiation dose during clinical acquisitions (1,3,4). The results of these investigations, however, should be interpreted with care because they were based on observer preference studies rather than on diagnostic performance studies.

Lesion detection studies provide better assessment of diagnostic accuracy than do observer preference studies and have been performed on chest phantoms with simulated chest disease (5–10). Results of these studies also showed better performance with digital radiography when compared with that of screen-film radiography when the same radiation dose was used (5,9) or similar performance when doses were reduced to 35% (6) or 50% (7,8,10). On the other hand, a recently performed lesion detection study showed decreased diagnostic performance with digital chest radiography when compared with performance at a dose level equal to that of screen-film radiography with 400 speed systems for mediastinal areas of an anthropomorphic phantom at a reduced dose (11). One controlled clinical study with computed tomography demonstrated that the detection of various chest abnormalities was not significantly different between the dose of 100% and the dose reduced to 35% (12).

The effects of dose reduction for imaging of chest nodules in patients are not known. Obtaining this information seems crucial in regard to whether radiation dose should be reduced for digital chest radiography. Accordingly, the purpose of our study was to determine to what extent dose reduction at digital posteroanterior (PA) chest radiography results in decreased detection of simulated nodules on patient images.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
Patients and Radiographic Imaging
The raw data, with patient identities concealed and without any pre- or postprocessing, of digital chest radiographs obtained in 21 patients with a consecutively normal finding at chest radiographic examination (eg, without intrathoracic abnormalities) were used for the study. All patients had undergone chest imaging for clinical reasons and were selected from patients who had undergone chest radiographic examination on one of three arbitrary afternoons chosen (April 22, 26, and 27, 2004). The policy of our institutional review board (IRB) is that informed consent of patients is not required if the research protocol concerns data with patient identities concealed, as was the case in our study. Accordingly, our IRB does not require its approval for such a study, as was confirmed by the IRB in response to our query.

Imaging was performed with a system for digital chest radiography that featured a slot-scanning detector composed of a cesium iodide scintillator and charge-coupled devices (ThoraScan; Delft Imaging Systems, Veenendaal, the Netherlands). Technical details of the system have been described elsewhere (13,14). The standard clinical acquisition parameter settings were as follows: tube current (in milliamperes) with automatic exposure control; tube voltage, 133 kV; focus to detector distance, 183 cm; and total filtration, 3-mm aluminum equivalent and 0.3-mm copper equivalent. The image matrix was 2736 x 2736 and had a pixel size of 162 µm, which resulted in a maximum image size of 44 x 44 cm. A 0.022-mSv effective dose at PA imaging was applied for a normal-size patient, which is an intermediate dose compared with those of other digital radiographic systems (15).

Lesion Simulation
Simulated lesions representing nodules (eg, metastases) were digitally added to the digital raw data of the PA chest radiographs. Chest nodules were simulated by computer programming spherical masses (16) (W.J.H.V., J.P.A.v.D.) with custom software (Matlab, Matrix Laboratory 6.5; MathWorks, Natick, Mass). These nodules varied in size with a radius of 25, 50, or 100 pixels (representing nodule diameters in the plane of projection of 0.81 cm, 1.62 cm, and 3.24 cm, respectively). For uniformity in this article, the term nodule is applied for all sizes. In addition, various degrees of attenuation were simulated by varying the signal intensity by scaling the arbitrary profile, where the linear attenuation coefficient for muscle (µ = 0.2 cm^–1) at mean photon energy of 70 keV was taken to represent low-density tissue. The signal intensity within the lesions varied in such a way that they represented three-dimensional nodules by modeling a semisphere with maximum attenuation in the lesion center and zero attenuation at the lesion border. For the lungs, the signal intensity varied between 0.8 and 1.6 in 0.1 increments to obtain lesions with maximum central attenuation in the direction of transmission equivalent to tissue thickness of 1.6–5.5 cm. For the mediastinum, the signal intensity varied between 0.4 and 1.1 in 0.1 increments to obtain lesions with attenuation equivalent to tissue thickness of 1.0–3.5 cm. In addition to basic modeling (16), the edges of the simulated three-dimensional lesions were blurred for improved visual representation of nodules.

Among 20 radiographs, 60 lesions were added to the lungs and 60 lesions were added to the mediastinum. Lesions were considered to be located in the mediastinum if they projected over the heart, spine, aorta, central pulmonary vessels, or diaphragm. The distribution of the number and location of these lesions within the lungs and within the mediastinum was random. The number of simulated lesions per patient varied between two and 13. For each patient, a standard-of-reference image was produced to provide for precise localization of each simulated lesion.

Simulation of Dose Reduction
Reduced-dose images were simulated by digitally adding zero-mean Gaussian white noise to the raw data by using custom software (Matlab; MathWorks) (W.J.H.V., J.P.A.v.D., J.G.). Simulating dose reduction was done on a pixel-by-pixel basis: To each pixel a certain number was added, which was derived from the zero-mean Gaussian distribution with a certain standard deviation, depending on the original pixel value (ie, the number of photons reaching the detector) (17).

In addition to the 100% dose images that had been derived at clinical chest imaging, three images, representing a 50%, a 25%, and a 12% patient dose, were created for each patient. The window-width and window-level settings for reduced-dose images were similar for all dose levels to provide for equal visual presentation and identical postprocessing techniques irrespective of dose. The usual clinical postprocessing system (MUSICA [multiscale image contrast amplification]; Agfa, Ridgefield, NJ) was used without equalization or enhancement of the mediastinum. For each patient, the 100%, 50%, 25%, and 12% radiographs were printed as hard copies on laser imaging film (DryView 35 x 43; Eastman Kodak, Rochester, NY) with a laser printer (DryView 8700; Kodak Nederland, Vianen, the Netherlands).

Image Reading
The 84 radiographs (four from one of the 21 patients were used for training) with added simulated lesions were presented to four independent observers, all of whom were senior radiologists with a mean experience of 17 years in reading chest images (range, 2–25 years). The radiographs were presented on a film viewing box (Rotolux Planilux; Philips Medical Systems, Best, the Netherlands). The observers were not informed about the addition of noise, the number of lesions simulated, or the distribution of lesions among the radiographs.

Image reading was performed in four sessions for each observer. Per session, 20 radiographs were read (ie, one image for each patient). To compensate for learning bias, the images were presented in random sequences and the dose level varied randomly among the images presented per session; five images of each dose level were presented per session. The observers were asked to read each image as if metastases were to be detected and to draw the contour of all lesions they detected on the radiographs. Reading sessions were performed within at least 1-week intervals. After all four sessions were completed, each observer had read images of all dose levels for each patient.

Training was performed ahead of each reading session, and the observers were presented with four chest radiographs of different lesions and dose levels and with the corresponding standard-of-reference radiographs. This allowed the observers to become accustomed to the visibility of the lesions and to the annotation of nodules. The training images were simulated by using data from one patient, and they were not among the images being evaluated.

Data Evaluation
The annotated radiographs were evaluated in consensus by two outcome readers (L.J.M.K. and W.J.H.V., 6 and 3 years of experience in reading PA chest images, respectively) by placing the annotated images over the corresponding standard-of-reference images.

When compared with the reference standard images, each lesion annotated was registered as "detected" (true-positive finding) and each lesion not annotated as "not detected" (false-negative finding). Annotated contours not containing lesions were registered as false-positive findings.

Statistical Analysis
Statistical analysis was performed by using a logistic regression model for the probability to detect the lesion. The model adjusts for dose level, signal intensity, and size, in addition to patient and observer normal random effect. Analysis was performed with a software package (WinBUGS version 1.4.1, 1996–2004; Imperial College and Medical Research Council Biostatistics Unit, Cambridge, England). Although comparison across relative dose levels is the primary outcome and research question, adjustment for lesion size and signal intensity is required and important in order to account for considerable variability of lesion detection due to these effects. Adjustment for patient and observer random effect is necessary in order to correct for correlation in the evaluation due to these effects. The analysis was performed for lesion detection in the lungs and mediastinum separately. For evaluation of false-positive annotations, analysis was performed with a Poisson model, which is a standard model for comparing observed differences in count data.

A (full Bayes) probabilistic implementation of the model was used for the evaluation. Hence, all model-based summary statistics, including probability measures presented, are based on the a posteriori distributions calibrated within the full probabilistic modeling approach. The results presented and discussed are the estimated effects and associated standard errors and confidence intervals, which are relevant for interpretation. To aid the interpretation of these measures, we include the a posteriori probabilities of the fitted detection probabilities as being smaller when compared with those at the reference level. These probability measures should not be confused with the traditional so-called P value.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 
Lesion Detection
An example of simulated chest nodules and reduced dose imaging is shown in Figure 1. Lowering the dose had no effect on the number of lesions detected in the lungs (Fig 2a). For the mediastinum, the lower the dose, the fewer lesions were detected (Fig 2b).

View larger version (155K): [in this window] [in a new window] [Download PPT slide]   Figure 1a: (a–d) Clinical PA digital radiographs of simulated chest nodules with (a, c) 100% radiation dose and with (b, d) simulated 12% dose. Images c and d show right lower part of radiographs a and b, respectively. Diameter and attenuation of lung lesion were 3.24 and 4.00 cm, respectively (arrow in a, b). For mediastinal lesions, these were 1.62 and 2.25 cm, 0.81 and 1.38 cm, 0.81 and 1.13 cm (arrow 1, 2, and 3, respectively). Note difference in noise levels between a and b and between c and d. Lung lesion is easily appreciated in radiographs a and b, whereas subtle mediastinal lesions seem more difficult to appreciate at lower dose (d).

View larger version (153K): [in this window] [in a new window] [Download PPT slide]   Figure 1b: (a–d) Clinical PA digital radiographs of simulated chest nodules with (a, c) 100% radiation dose and with (b, d) simulated 12% dose. Images c and d show right lower part of radiographs a and b, respectively. Diameter and attenuation of lung lesion were 3.24 and 4.00 cm, respectively (arrow in a, b). For mediastinal lesions, these were 1.62 and 2.25 cm, 0.81 and 1.38 cm, 0.81 and 1.13 cm (arrow 1, 2, and 3, respectively). Note difference in noise levels between a and b and between c and d. Lung lesion is easily appreciated in radiographs a and b, whereas subtle mediastinal lesions seem more difficult to appreciate at lower dose (d).

View larger version (127K): [in this window] [in a new window] [Download PPT slide]   Figure 1c: (a–d) Clinical PA digital radiographs of simulated chest nodules with (a, c) 100% radiation dose and with (b, d) simulated 12% dose. Images c and d show right lower part of radiographs a and b, respectively. Diameter and attenuation of lung lesion were 3.24 and 4.00 cm, respectively (arrow in a, b). For mediastinal lesions, these were 1.62 and 2.25 cm, 0.81 and 1.38 cm, 0.81 and 1.13 cm (arrow 1, 2, and 3, respectively). Note difference in noise levels between a and b and between c and d. Lung lesion is easily appreciated in radiographs a and b, whereas subtle mediastinal lesions seem more difficult to appreciate at lower dose (d).

View larger version (152K): [in this window] [in a new window] [Download PPT slide]   Figure 1d: (a–d) Clinical PA digital radiographs of simulated chest nodules with (a, c) 100% radiation dose and with (b, d) simulated 12% dose. Images c and d show right lower part of radiographs a and b, respectively. Diameter and attenuation of lung lesion were 3.24 and 4.00 cm, respectively (arrow in a, b). For mediastinal lesions, these were 1.62 and 2.25 cm, 0.81 and 1.38 cm, 0.81 and 1.13 cm (arrow 1, 2, and 3, respectively). Note difference in noise levels between a and b and between c and d. Lung lesion is easily appreciated in radiographs a and b, whereas subtle mediastinal lesions seem more difficult to appreciate at lower dose (d).

View larger version (15K): [in this window] [in a new window] [Download PPT slide]   Figure 2a: (a, b) Graphs of number of simulated lesions detected in (a) lungs and (b) mediastinum at 100% dose and at reduced dose levels accumulated for all four observers. Note difference between lungs and mediastinum for effect of dose reduction.

View larger version (13K): [in this window] [in a new window] [Download PPT slide]   Figure 2b: (a, b) Graphs of number of simulated lesions detected in (a) lungs and (b) mediastinum at 100% dose and at reduced dose levels accumulated for all four observers. Note difference between lungs and mediastinum for effect of dose reduction.

Note that the probabilities (Table 1) may be interpreted as the area under the receiver operating characteristic curve (18) and hence give information on the extent to which the detection probabilities differ after accounting for other information, such as lesion size, signal intensity, and observer and patient effects.

A stronger positive effect for lesion signal intensity was shown for the lungs than for the mediastinum. The estimate for signal intensity was the increase in log(odds) for each increment increase in signal intensity. The positive coefficient indicates increasing detection probabilities with increasing signal intensity. Likewise, for lesion size, detection probability decreased from those observed at the 100-pixel radius size downward (Table 2). The effect for lesion size was much stronger for the lungs than for the mediastinum (Table 1).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

The main finding of our study was that lowering the radiation dose from 100% to 50%, 25%, and 12% resulted in decreased observer detection at each step for nodules projected over the mediastinum, whereas no effect was found for detection in the lungs.

At chest radiography, the detection of subtle nodules is limited by the image background (19) and is highly dependent on the location in the chest; effects are on contrast-detail and nodule diameters (20). Recently, various aspects concerning detectability of 10-mm nodules at digital radiography have been investigated thoroughly by the Radiological Imaging Unification Strategies (RADIUS) chest trial group. The authors found that anatomic background (ie, the projected anatomy that is not anatomic noise) acts largely as noise and is the major image component affecting the detectability of nodules. Of less importance is the disturbance caused by anatomic noise (ie, scatter noise and structural noise), although this component is still larger than that of the system noise (ie, quantum and detector noise) (21).

In our study, the only varying parameter was the simulated dose (ie, simulation of reduced quanta available for acquiring the image) which resulted in higher relative quantum noise. The technical factor "detector" was not changed, and the projected anatomic conditions per lesion (ie, the anatomic background and the anatomic noise) were constant during the investigation because for each patient, the same raw data set was used for simulating lower doses. Moreover, the displayed image contrast was set equal for all doses presented. Therefore, for lesions projected over the mediastinum, the technical factor "quantum noise" must have been the limiting factor for reduced nodule detection.

The dose we used at imaging in clinical practice with the digital system is roughly similar to that used at screen-film radiography (400 speed) (14). The exposure level was sufficient for the lungs and could even be considered superfluous, because nodule detection did not decrease at reduced dose, even to a level of 12%. Decreased nodule detection for lesions projected over the mediastinum at reduced dose, however, implies suboptimal exposure for the mediastinum with these reduced doses. Although the added value of digital radiography when compared with screen-film radiography has been found advantageous mainly at imaging of the mediastinum (9,10,22), the decreased diagnostic imaging quality we found in the mediastinum at reduced dose lessens the possible added value of digital radiography. This may form an essential limitation for lowering the patient dose.

Lesion size and signal intensity were important predictor variables for detection probability within the lungs and within the mediastinum, and these effects were stronger in the lungs than in the mediastinum. Our signal intensity findings may fit in with findings from the RADIUS chest study (23). These authors found a higher level of conspicuity for certain object contrasts in the lungs than in mediastinal areas. This was because of the combination of a low scatter-to-primary ratio (a relatively high number of quanta used reached the detector) together with a relative low image contrast (that was low because of the relative anatomic background homogeneity in the lungs) (23). In our study, the higher level of conspicuity for certain object contrasts in the lungs might be explained by the greater effect of signal intensity differences in the lungs than in the mediastinum. It is possible that the concept of higher level of conspicuity explains the greater effect of lesion size in the lungs, as well. The exact relation of the observational effect between signal intensity, size, and the visibility of lesions in various chest areas, however, is difficult to discern and its cause still has to be determined.

A limitation of our study is that only digital PA chest radiographs were evaluated. In these radiographs, the signal intensity used for the simulated nodules in the lungs differed slightly from that in the mediastinum. This was done because the nodules with the highest signal intensity in the mediastinum appeared sharply delineated when the images were being prepared for the study, and this effect was reduced by decreasing the signal intensity in the nodules. Therefore, the number of lesions detected in the lungs and the number detected in the mediastinum were not directly comparable, and this may have additionally contributed to a decreased number of lesions detected in the mediastinum. With the statistical model we used, however, using different signal intensities for nodules in the mediastinum and lungs caused no other effects on the study results especially not for differences between dose levels. Although soft-copy reading is usually performed for digital images in clinical practice, for study purposes we chose hard-copy reading. We presume that the study outcome would not have been different if soft-copy reading had been used instead. Another limitation is that calculated patient effective doses and detector characteristics are rather variable between different digital systems used in clinical practice (15). Consequently, the results found in our study may not be directly extrapolated to other digital systems.

Practical application: Because the depiction of lesions can be obscured with anatomic background and various noise sources, which is especially the case for the mediastinum, the known advantage of digital radiography over screen-film radiography for better observer detection of lesions projected over the mediastinum (9,10,22) seems to be a desirable diagnostic improvement if the dose is not decreased. The results of our study with simulated nodules and clinical digital radiographs show that lowering the radiation dose leads to decreased observer detection of lesions projected over the mediastinum. We therefore suggest some reservation about lowering the radiation dose at digital radiography for nodule presence or evaluation.

	ADVANCES IN KNOWLEDGE

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 

• Dose reduction at clinical digital posteroanterior chest radiography results in decreased observer detection of simulated nodules in the mediastinum but not in the lungs.
  
• Detection probability decreases with smaller lesions and with lower signal intensity lesions.
  
• Lesion size and lesion signal intensity are independent variables for detection probability, and the effects are stronger in the lungs than in the mediastinum.
  

	ACKNOWLEDGMENTS

 
We acknowledge radiologists S. J. G. C. Frerichs, MD, A. de Roos, MD, PhD, G. J. Vielvoye, MD, PhD, and H. M. Zonderland, MD, PhD, from the Leiden University Medical Center, the Netherlands, for participating as observers in the study.

	FOOTNOTES

 

Abbreviations: PA = posteroanterior

Authors stated no financial relationship to disclose.

Author contributions: Guarantor of integrity of entire study, L.J.M.K.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; literature research, L.J.M.K., W.J.H.V., J.G.; experimental studies, L.J.M.K., W.J.H.V., J.P.A.v.D., J.G.; statistical analysis, B.J.A.M.; and manuscript editing, L.J.M.K., W.J.H.V., B.J.A.M., J.G.

	References

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION ADVANCES IN KNOWLEDGE References

 

1. Fink C, Hallscheidt PJ, Noeldge G, et al. Clinical comparative study with a large-area amorphous silicon flat-panel detector: image quality and visibility of anatomic structures on chest radiography. AJR Am J Roentgenol 2002;178:481–486.[Abstract/Free Full Text]
2. Strotzer M, Völk M, Fründ R, Hamer O, Zorger N, Feuerbach S. Routine chest radiography using a flat-panel detector: image quality at standard detector dose and 33% dose reduction. AJR Am J Roentgenol 2002;178:169–171.[Abstract/Free Full Text]
3. Ganten M, Radeleff B, Kampschulte A, Daniels MD, Kauffmann GW, Hansmann J. Comparing image quality of flat-panel chest radiography with storage phosphor radiography and film-screen radiography. AJR Am J Roentgenol 2003;181:171–176.[Abstract/Free Full Text]
4. Bacher K, Smeets P, Bonnarens K, De Hauwere A, Verstraete K, Thierens H. Dose reduction in patients undergoing chest imaging: digital amorphous silicon flat-panel detector radiography versus conventional film-screen radiography and phosphor-based computed radiography. AJR Am J Roentgenol 2003;181:923–929.[Abstract/Free Full Text]
5. Schaefer-Prokop CM, Prokop M, Schmidt A, Neitzel U, Galanski M. Selenium radiography versus storage phosphor and conventional radiography in the detection of simulated chest lesions. Radiology 1996;201:45–50.[Abstract/Free Full Text]
6. van Heesewijk HP, van der Graaf Y, de Valois JC, Feldberg MA. Effects of dose reduction on digital chest imaging using a selenium detector: a study of detecting simulated diffuse interstitial pulmonary disease. AJR Am J Roentgenol 1996;167:403–408.[Abstract/Free Full Text]
7. Strotzer M, Gmeinwieser JK, Völk M, Fründ R, Seitz J, Feuerbach S. Detection of simulated chest lesions with normal and reduced radiation dose. Invest Radiol 1998;33:98–103.[CrossRef][Medline]
8. Rapp-Bernhardt U, Roehl FW, Gibbs RC, Schmidl H, Krause UW, Bernhardt TM. Flat-panel x-ray detector based on amorphous silicon versus asymmetric screen-film system: phantom study of dose reduction and depiction of simulated findings. Radiology 2003;227:484–492.[Abstract/Free Full Text]
9. Kroft LJ, Geleijns J, Mertens BJ, Veldkamp WJ, Zonderland HM, de Roos A. Digital slot-scan charged coupled device chest radiography versus AMBER and Bucky screen-film radiography: detection of simulated chest nodules and interstitial disease using a chest phantom. Radiology 2004;231:156–163.[Abstract/Free Full Text]
10. Bernhardt TM, Otto D, Reichel G, et al. Detection of simulated interstitial lung disease and catheters with selenium, storage phosphor, and film-based radiography. Radiology 1999;213:445–454.[Abstract/Free Full Text]
11. Metz S, Damoser P, Hollweck R, et al. Chest radiography with a digital flat-panel detector: experimental receiver operating characteristic analysis. Radiology 2005;234:776–784.[Abstract/Free Full Text]
12. van Heesewijk HP, Casparie HW, de Valois JC, van der Graaf Y. Effect of dose levels on the diagnostic performance of a selenium-based digital chest system. Invest Radiol 2001;36:455–459.[CrossRef][Medline]
13. Samei E, Saunders RS, Lo JY, et al. Fundamental imaging characteristics of a slot-scan digital chest radiographic system. Med Phys 2004;31:2687–2698.[CrossRef][Medline]
14. Veldkamp WJ, Kroft LJ, Mertens BJ, Geleijns J. Digital slot-scan charge-coupled device radiography versus AMBER and Bucky screen-film radiography: comparison of image quality in a phantom study. Radiology 2005;235:857–866.[Abstract/Free Full Text]
15. Kroft LJ, Veldkamp WJ, Mertens BJ, Boot MV, Geleijns J. Comparison of eight different digital chest radiography systems: variation in detection of simulated chest disease. AJR Am J Roentgenol 2005;185:339–346.[Abstract/Free Full Text]
16. te Brake GM, Karssemeijer N. Single and multiscale detection of masses in digital mammograms. IEEE Trans Med Imaging 1999;18:628–639.[CrossRef][Medline]
17. Veldkamp WJ, Kroft LJ, van Delft JP, Mertens BJ, Geleijns J. Techniques for feasibility of dose reduction with a digital chest radiography system. Biomed Tech (Berl) 2005;50(suppl):425–426.
18. Pepe MS. The receiver operating characteristic curve. In: Oxford Statistical Science Series: No. 3, The statistical evaluation of medical tests for classification and prediction. Oxford, England: Oxford University Press, 2003; 66–95.
19. Samei E, Flynn MJ, Eyler WR. Detection of subtle lung nodules: relative influence of quantum and anatomic noise on chest radiographs. Radiology 1999;213:727–734.[Abstract/Free Full Text]
20. Samei E, Flynn MJ, Peterson E, Eyler WR. Subtle lung nodules: influence of local anatomic variations on detection. Radiology 2003;228:76–84.[Abstract/Free Full Text]
21. Håkansson M, Båth M, Börjesson S, et al. Nodule detection in digital chest radiography: summary of the RADIUS chest trial. Radiat Prot Dosimetry 2005;114:114–120.[Abstract/Free Full Text]
22. Garmer M, Hennigs SP, Jäger HJ, et al. Digital radiography versus conventional radiography in chest imaging: diagnostic performance of a large-area silicon flat panel detector in a clinical CT-controlled study. AJR Am J Roentgenol 2000;174:75–80.[Abstract/Free Full Text]
23. Båth M, Håkansson M, Börjesson S, et al. Nodule detection in digital chest radiography: introduction to the RADIUS chest trial. Radiat Prot Dosimetry 2005;114:85–91.[Abstract/Free Full Text]

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]

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Kroft, L. J. M. Articles by Geleijns, J. Search for Related Content PubMed PubMed Citation Articles by Kroft, L. J. M. Articles by Geleijns, J.