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


     

DOI: 10.1148/radiol.2343031805
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 Metz, S.
Right arrow Articles by Link, T. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Metz, S.
Right arrow Articles by Link, T. M.
(Radiology 2005;234:776-784.)
© RSNA, 2005

Experimental Studies

Chest Radiography with a Digital Flat-Panel Detector: Experimental Receiver Operating Characteristic Analysis1

Stephan Metz, MD, Petra Damoser, Regina Hollweck, Rudolf Roggel, MD, Christoph Engelke, MD, Klaus Woertler, MD, Bernhard Renger, MS, Ernst J. Rummeny, MD and Thomas M. Link, MD

1 From the Department of Diagnostic Radiology (S.M., P.D., R.R., C.E., K.W., B.R., E.J.R., T.M.L.) and Institute of Medical Statistics and Epidemiology (R.H.), Technische Universität München, Ismaninger Str 22, 81675 Munich, Germany. Received November 11, 2003; revision requested February 3, 2004; revision received April 21; accepted June 2. Address correspondence to S.M. (e-mail: smetz@roe.med.tu-muenchen.de).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
PURPOSE: To evaluate the influence of different detector radiation doses and peak kilovoltage settings on diagnostic performance and radiation dose at posteroanterior (PA) chest radiography performed with an amorphous silicon flat-panel detector (FPD).

MATERIALS AND METHODS: All examinations were performed by using a digital FPD. PA chest radiographs of an anthropomorphic chest phantom were obtained with detector radiation doses of 2.50 µGy (system speed, 400), 1.56 µGy (speed, 640), and 1.25 µGy (speed, 800) and with peak kilovoltage values of 100, 120, and 140 kVp. Four types of simulated lesions—nodules of different sizes, polylobulated lesions, interstitial-nodular lesions, and interstitial-reticular lesions—were superimposed on the phantom. After four radiologists assessed all of the images, receiver operating characteristics analysis was performed. In addition, the entrance surface dose was measured and the effective dose was calculated.

RESULTS: Reduced detector dose led to significantly decreased diagnostic performance in overall lesion detection (P < .05). However, over pulmonary areas only, this effect could not be seen. With use of the same kilovoltage values, reducing the detector dose, even to 1.25 µGy (speed, 800), did not lead to significantly decreased lesion detectability. In terms of diagnostic performance and effective dose, 120 kVp was the most effective.

CONCLUSION: Standard PA chest radiographs should still be acquired at a detector dose of 2.50 µGy (speed, 400) with 120 kVp to yield the highest diagnostic performance. However, when the present analysis was focused on the lung fields only, no significant loss in diagnostic performance could be demonstrated, even after a 50% reduction in radiation dose.

© RSNA, 2005


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
The results of numerous studies have shown the diagnostic performance of digital detector systems at chest radiography to be equal or superior to that of other systems used in routine clinical practice, such as conventional screen-film and storage-phosphor radiographic systems (1–8). Flat-panel detectors (FPDs) are highly sensitive to radiation and are capable of responding to a wide range of radiation exposure levels—that is, they have wide latitude—while maintaining high local contrast. FPDs are also characterized by a detective quantum efficiency that is high and superior to that of the two other described systems (9). However, the spatial resolution of FPDs is lower than that of screen-film radiographic systems (10–12).

The results of some previously performed studies indicate that using FPDs at posteroanterior chest radiography should yield benefits in terms of the superior image quality achieved with the same radiation dose and/or the reduced radiation dose to the patient without loss of diagnostic performance, as compared with the image quality achieved andthe radiation dose required with the other systems (6–8,13,14). However, the ability to use substantially reduced radiation doses with FPDs without decreasing diagnostic performance remains a controversial topic (7,13–16). To our knowledge, there have been no studies on the influence of different peak kilovoltage values on diagnostic performance.

The most commonly used method of assessing the radiation dose to the patient is that of measuring the entrance surface dose (ESD). Higher peak kilovoltage settings are applied in many radiology departments because the ESD decreases as the peak kilovoltage increases (17). However, the relative associated risk is not factored into the ESD as it is into the effective dose, which is regarded as a more appropriate indicator of the risk associated with different peak kilovoltage settings (18). Launders and co-workers (17) compared various peak kilovoltage values and found that the minimum effective radiation dose was reached at settings of 90–110 kVp. In addition, the detective quantum efficiency increased as the peak kilovoltage was reduced (9,17).

The purpose of this experimental study was to evaluate the influence of different detector radiation doses and peak kilovoltage settings on diagnostic performance and radiation dose at posteroanterior chest radiography performed with an amorphous silicon FPD.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Chest Phantom
Thin wires were used to divide an anthropomorphic chest phantom (Alderson, Long Beach, Calif) into six pulmonary (left superior, left middle, left lower third, right superior, right middle, and right lower third), three mediastinal (superior, middle, and lower third), and three subphrenic (right, central, left) areas, which together comprised the entire phantom (Fig 1). The thin wires were present during imaging and therefore clearly delineated the 12 areas on the radiographs (Fig 1). To simulate various types of lesions, four different structures were created (by S.M.): (a) nodules of various sizes (5–15 mm) that were composed of different dilutions of paraffin and bee’s wax, (b) polylobulated lesions made of silicone, (c) interstitial-nodular lesions composed of clusters of millet grains, and (d) interstitial-reticular lesions simulated by gauze soaked in different dilutions of contrast material (iotrolan, Isovist 300; Schering, Berlin, Germany) and saline (Fig 2).



View larger version (143K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Posteroanterior radiograph of whole phantom. Thin wires clearly demarcate the 12 areas.

 


View larger version (165K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2. Radiographs of simulated lesions (arrows) obtained with a detector dose of 2.50 µGy and 120 kVp. A, Nodule over a mediastinal area. B, Interstitial-nodular lesion over a pulmonary area. C, Polylobulated lesion over a pulmonary area. D, Interstitial-reticular lesion over a mediastinal area.

 
These artificial lesions were randomly taped (by S.M.) onto 20 2-mm-thick plastic plates that corresponded to the 12 areas of the phantom. On each plate, zero to all 12 of the defined areas could be encompassed within one lesion, and each plate could contain one to all four kinds of lesions. Ultimately, however, 120 (50%) of the 240 defined areas were covered with lesions, and each lesion was placed in 30 areas—50% over pulmonary areas and 50% over mediastinal and subphrenic areas. Each of the 20 plates was then superimposed on the phantom, and radiographs of the phantom were obtained (by S.M.) by using different radiation exposure parameters. Defined rotations of the phantom—10° to the left and 10° to the right—were performed during the acquisition of half the images to change the appearance of the chest anatomy on the radiographs.

Detector System
The digital radiography system used in this study was a commercially available unit that consisted of a large-area FPD (Revolution XQ/i; GE Medical Systems, Milwaukee, Wis) with a spatial resolution of 2.5 line pairs per millimeter. The performance of this FPD system has been previously described in detail (19). The FPD is fabricated on a single monolithic glass substrate with an active image area of 41 x 41 cm. Amorphous silicon is layered onto the glass, thin-film transistor array, which, in turn, is overlaid with a structured cesium iodide scintillator and a protective coating. When the scintillator is exposed to x-rays, visible light is emitted and converted to an electric signal in photodiodes in the thin-film transistor array. Each pixel is then analyzed by means of onboard amplifiers and sampling electronics and then converted to a digital value to form a raw image that is then corrected for nonuniformities by using system-specific calibration data. The pixels are square, with dimensions of 200 µm, and yield an image matrix of 2048 x 2048 pixels. Pixel intensities are acquired at 14 bits per pixel. An antiscatter grid (ratio, 13:1; 70 lines per centimeter) was used for digital exposures.

The digital images were postprocessed with a standardized algorithm supplied by the manufacturer (XQ/i M3; GE Medical Systems) that had been adjusted to yield image appearances similar to those generated with the screen-film system used in our department. The postprocessing procedure included a multi–spatial resolution algorithm to perform edge enhancement and dynamic range reduction and an applied asymmetric contrast curve to match the appearance of the screen-film system. All images were printed on film hard copies (Kodak Dry View 8700; Eastman Kodak, Rochester, NY).

Exposure Parameters
The anthropomorphic chest phantom was positioned so that it was posteroanterior to the FPD. Imaging parameters were as follows: source-to-object distance, 1.8 m; filtration, 2.5 mm Al; focal spot size, 0.6 mm; and tube current, 200 mA. All exposures were controlled by the automatic exposure control (part of Revolution XQ/i FPD). Radiation exposure was terminated when the dose reached the preset level in the ionization chambers. The influence of varying peak kilovoltage values (100, 120, and 140 kVp) on the detectability of four different artificial lesions and on the ESD was studied while keeping the detector radiation dose constant at 2.50 µGy (system speed, 400). A detector dose of 2.50 µGy and a peak kilovoltage value of 120 kVp are routinely used in our department. Then, we examined the influence of lower detector doses—specifically, 1.56 µGy (speed, 640) and 1.25 µGy (speed, 800)—with each peak kilovoltage value on image quality, ESD, and effective dose. Hence, a total of nine parameter settings were evaluated.

Dose Measurements
ESD.—The energy dose (in grays) was measured (by P.D.) by using a semiconductor detector (DIADOS; PTW, Freiburg, Germany) that was calibrated by the manufacturer. The detector was fixed on the posterior part of the phantom (focus side), and for each parameter setting (detector dose and peak kilovoltage), measurements in nine defined regions of the radiation field encompassing the entire phantom were evaluated. The mean of the energy dose measurement for all nine regions was calculated as the ESD (in milligrays) and used to calculate the effective dose (in microsieverts) (by P.D.).

Effective dose.—The organ dose (Dorgan, in sieverts) was calculated as follows: Dorgan = ESD · CF. The conversion factors (CFs), calculated as mean organ dose/ESD (in microsieverts/milligrays), were determined by using virtual organ detectors and Monte Carlo techniques. These factors are published in tables for selected examinations (posteroanterior chest radiography performed with a source-to-object distance of 1.8 m; 2.5-mm Al filtration; and 90, 125, and 140 kVp) (20). On the basis of these data, the values for 100 and 120 kVp were interpolated. The effective dose (in sieverts) is the sum of the product of each organ dose and a tissue-weighting factor of biologic effects (Worgan) and is calculated as follows: {Sigma}(Dorgan · Worgan). The tissue-weighting factors of biologic effects were derived from published recommendations of the International Commission of Radiological Protection (18).

Image Analysis
Four experienced radiologists (R.R. with 3, C.E. with 5, K.W. with 10, and T.M.L. with 10 years of experience) independently assessed all 180 images (20 plastic plates by nine parameter settings) explicitly for the presence or absence of a lesion over each of the 12 defined areas. This analysis resulted in a total of 8640 observations (180 images by 12 areas by four observers)—960 observations per exposure parameter.

All images were randomly presented, and not more than 45 images were shown during one session. To prevent learning bias, the interval between each reading was at least 3 weeks. None of the observers was involved in the preparation of the phantoms. No time constraints were imposed. All images were viewed on a quality-controlled light box with moveable shutters under subdued ambient light. The exposure parameters used were not indicated on the film images. To prevent learning curve–induced changes, film hard-copy images of thephantoms without any simulated structures and examples of simulated lesions were presented to the observers before the image analysis.

Up to four kinds of lesions could be seen on one plate. We therefore used a modified evaluation method that was adapted for our experimental model: (a) The observers were asked to grade the presence or absence of a lesion according to a decision scale with five levels of confidence. Grade 5 meant definitely positive for the presence of a lesion; grade 4, probably positive; grade 3, uncertain; grade 2, probably negative for the presence of a lesion; and grade 1, definitely negative. (b) If a lesion was graded as positive (grade 5 or 4), the type of lesion (nodule, polylobulated lesion, interstitial-nodular lesion, or interstitial-nodular lesion) had to be indicated.

Statistical Analyses
After completion of the readings, the lesions depicted on the radiographs were characterized (by S.M.). If the evaluated lesion type (nodule, polylobulated lesion, interstitial-nodular lesion, or interstitial-nodular lesion) did not match the true simulated lesion on the plate, the observation was defined as negative (grade 1). However, this occurred in only 24 (0.5%) of a total of 4320 observations that revealed a lesion. Subsequently, to evaluate lesion detection, a standard receiver operating characteristics analysis was performed (by S.M. and R.H.) by using the MedCalc program (version 6.15.000; MedCalc Software, Mariakerke, Belgium) to calculate the area under the receiver operating characteristics curve (AUC) and the corresponding 95% confidence interval (21,22).

Subgroup analyses were performed separately for each observer, each lesion, and each of the nine radiation exposure parameter settings by calculating one AUC value. Furthermore, an analysis of overall lesion detection performance, as well as separate analyses of lesion detection in the mediastinal and subphrenic areas versus the pulmonary areas, was performed by summarizing the data for all lesions and all observers that were obtained by using each of the nine parameter settings. No adjustments for multiple comparisons were applied.

Statistical differences were evaluated by comparing two receiver operating characteristics curves by using a two-tailed paired t test (23), and P < .05 indicated significance. In addition, one of theauthors (R.H.) performed a retrospective power analysis for bivariate two-tailed comparisons of AUC values on the basis of the difference between the AUC and the standard error (exemplarily shown for a sample size of 480, which is the number of lesions over pulmonary areas and over mediastinal and subphrenic areas [Table 1]).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Retrospective Power Analysis Results

 

   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
Radiation Dose Measurements
The required milliampere second (mAs) values (mean values of 20 exposures), the ESD measurements, and the effective dose calculations for each of the nine parameter settings are shown in Table 2 and in Figures 3 and 4, respectively. The variation in peak kilovoltage values was less than 5%. At a constant detector radiation dose, the ESD decreased as the peak kilovoltage was increased. In contrast, the effective dose decreased as the peak kilovoltage was decreased. At each detector dose, the effective dose became approximately 3% lower as the peak kilovoltage decreased from 120 to 100 kVp, became approximately 7% lower as the peak kilovoltage decreased from 140 to 120 kVp, and became approximately 11% lower as the peak kilovoltage decreased from 140 to 100 kVp.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Detector Radiation Dose, Peak Kilovoltage, and Milliampere Second Values Evaluated

 


View larger version (17K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Graph shows ESDs measured at each detector dose-peak kilovoltage setting.

 


View larger version (17K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 4. Graph shows effective doses calculated at each detector dose-peak kilovoltage setting.

 
Receiver Operating Characteristics Analyses
AUC values for each observer (interobserver variability).—AUC values for each observer and with each of the nine radiation exposure parameter settings (Fig 5) were calculated. AUC values were similar for the lesion detection by observers 2, 3, and 4, whereas AUC values for observer 1 with most of the parameter settings were significantly (P < .05) higher. The smallest variability in AUC values was observed with a detector dose of 2.50 µGy, the use of which also yielded the highest AUC values for each observer.



View larger version (43K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 5. Graph illustrates interobserver variability. Separate AUC values (means and 95% confidence intervals) for each observer and each parameter setting are shown. The highest AUC values and the smallest variability in AUC values were observed at a detector dose of 2.50 µGy.

 
AUC values for each lesion (interlesion variability).—AUC values for each lesion and with each of the nine exposure parameter settings (Fig 6) were calculated. The results were comparable to those for interobserver variability (Fig 5). Again, the variability in AUC values was smaller with a detector dose of 2.50 µGy, the use of which also yielded the highest AUC values for each lesion. Overall, the average AUC value for the detection of nodules and polylobulated lesions was superior to that for the detection of interstitial-nodular and interstitial-reticular lesions.



View larger version (42K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 6. Graph illustrates interlesion variability. Separate AUC values (means and 95% confidence intervals) for each lesion and each parameter setting are shown. Again, the highest AUC values and the smallest variability in AUC values were observed at a detector dose of 2.50 µGy. IN = interstitial-nodular lesion, IR = interstitial-reticular lesion, N = nodule, PL = polylobulated lesion.

 
Overall performance.—AUC values were calculated for each exposure parameter setting (Table 3, Fig 7). There were no significant differences in AUC values based on the individual peak kilovoltage values used with each detector dose. Diagnostic performance in the detection of all simulated lesions was best with a detector dose of 2.50 µGy, with most of the differences in performance with the lower detector doses being significant (Fig 7). The evaluated AUC values did not vary significantly at radiation doses of between 1.56 and 1.25 µGy.


View this table:
[in this window]
[in a new window]

  		TABLE 3. Performance of Various Radiation Exposure Parameters at Chest Radiography

 


View larger version (28K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 7. Graph illustrates overall performance. AUC values (means and 95% confidence intervals) for each parameter setting are shown. The diagnostic performance in the detection of all simulated lesions was best at a detector dose of 2.50 µGy. ** = P < .01 for difference in values, as compared with those calculated at 2.50 µGy and 120 and 140 kVp. *** = P < .001 for difference in values, as compared with those calculated at 2.50 µGy and 120 and 140 kVp. x = P < .05 for difference in values, as compared with those calculated at 2.50 µGy and 100 kVp. xx = P < .01 for difference in values, as compared with those calculated at 2.50 µGy and 100 kVp.

 
Mediastinal and subphrenic areas versus pulmonary areas.—With use of averaged data for all lesions and all observers, the AUC values for each exposure parameter setting were calculated separately for lesion detection over the pulmonary areas (Table 3, Fig 8) and over the mediastinal and subphrenic areas (Table 3, Fig 9).



View larger version (23K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 8. Graph illustrates data for pulmonary areas. AUC values (means and 95% confidence intervals) for each parameter setting are shown. At the same detector dose, use of higher peak kilovoltages led to increased detection of all lesions. * = P < .05 for difference in values, as compared with those calculated at 2.50 µGy and 140 kVp and at 1.25 µGy and 140 kVp.

 


View larger version (24K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 9. Graph illustrates data for mediastinal and subphrenic areas. AUC values (means and 95% confidence intervals) for each parameter setting are shown. The diagnostic performance decreased significantly as the detector dose was reduced from 2.50 µGy. xx = P < .01 for difference in values, as compared with those calculated at 2.50 µGy and 120 kVp. xxx = P < .001 for difference in values, as compared with those calculated at 2.50 µGy and 120 kVp.

 
Over the pulmonary areas, the use of higher peak kilovoltage values led to increased detection of all lesions, with a significantly (P < .05) better performance at 140 kVp than at 100 kVp. The images in Figure 10 show a simulated noduleover a pulmonary area, as depicted at 100 and 140 kVp with the same detector dose. There is notably better visibility of the lesion with the higher peak kilovoltage value. With the same peak kilovoltage values, the detectability of no lesions decreased significantly as the detector dose was reduced from 2.50 µGy (Fig 11, A, B).



View larger version (93K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 10. Radiographs of simulated nodule (arrows) over a pulmonary area obtained at 100 kVp (A) and 140 kVp (B) by using the same detector dose of 2.50 µGy. Note the better depiction of the lesion at the higher peak kilovoltage.

 


View larger version (177K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 11. Radiographs of simulated lesions over pulmonary and mediastinal areas. A, B, Interstitial-reticular lesion (arrows) over a pulmonary area depicted on radiographs obtained at detector doses of 2.50 µGy (A) and 1.25 µGy (B) and at the same peak kilovoltage value of 120 kVp. Note that the detectability of the lesion is very similar on these two images. C, D, Interstitial-nodular lesion (arrows) over a mediastinal area depicted on radiographs obtained at detector doses of 2.50 µGy (C) and 1.25 µGy (D) and at the same peak kilovoltage value of 120 kVp. Note the different depictions of the lesions with the different detector radiation doses; the substantially lower amount of noise in C improves detectability.

 
Over the mediastinal (P < .01) and subphrenic (P < .001) areas, the diagnostic performance decreased significantly as the detector dose was reduced from 2.50 µGy (Fig 11, C, D). With regard to different peak kilovoltage values, a significant difference in diagnostic performance was noted only between 100 and 120 kVp at a detector dose of 1.56 µGy.

In Figure 11, examples of simulated interstitial-reticular lesions over pulmonary areas and of simulated interstitial-nodular lesions over mediastinal areas are depicted on radiographs obtained with different detector doses. There is a clear decrease in lesion depiction over the mediastinal areas only.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
REFERENCES
 
The results of our phantom study show that the detector radiation dose could not be reduced below 2.50 µGy (speed, 400), to 1.56 µGy (speed, 640) or 1.25 µGy (speed, 800), without causing a significant decrease in diagnostic performance in the overall detection of simulated lesions. In this context, different peak kilovoltage values (100, 120, and 140 kVp) did not influence the diagnostic performance. However, the diagnostic performance in the detection of lesions over the pulmonary areas was significantly improved with increases in peak kilovoltage at a constant detector dose. In addition, when the same peak kilovoltage values were used, no significant loss in the detectability of any lesion over the pulmonary areas could be demonstrated, even after a reduction in the detector dose from 2.50 µGy to 1.56 and 1.25 µGy.

Several research groups (7,13–16) have used different phantom models to study the effect of detector dose on diagnostic performance at chest radiography performed with digital detector systems. In general, new exposure parameters for chest radiography should be studied in vitro by using a model closest to clinical findings. Therefore, studies involving the use of an anthropomorphic chest phantom and different relevant lung abnormalities yield acceptable data, although clinical testing is still required to validate the results. Nevertheless, anthropomorphic chest phantom studies can be used to evaluate a wide range of imaging parameters that would not be assessable in a patient trial and to match the positioning and exposure parameters for each method tested. Therefore, they yield a better clinical study design by indicating which parameters are likely to have the largest effect on the diagnostic performance.

van Heesewijk and co-workers (15), in a receiver operating characteristic study with an anthropomorphic chest phantom and bird seed to simulate diffuse interstitial pulmonary disease, observed no significant reduction in the detectability of these lesions as the standard setting of 2.0 mAs (ESD, 0.049 mGy) was reduced to 1.0 mAs and subsequently to 0.5 mAs. We also observed no significant differences in diagnostic performance with the lower exposure doses of 1.56 and 1.25 µGy. However, in our study, a significant reduction in diagnostic performance was observed when the detector dose was lowered from 2.50 µGy. Differences between the van Heesewijk et al study results and our study results could be explained by the relatively low initial standard radiation dose setting and the subsequent further reductions in dose in the former study.

Comparing the diagnostic performance achieved at different system speed levels in a model similar to that used in our study, significantly decreasing diagnostic performance was observed in the other study (7) after altering the speed from 200 to 600, but there was no significant difference between the performance with a speed of 200 and that with a speed of 400 (7). Furthermore, consistent with our results were some significant differences between the detectability of single lesions at a speed of 400 and that at a speed of 600. Moreover, when the speed was increased to 600, a decrease in the detection of nodules, especially over regions with attenuation higher than that of the lung, was noted in the other study (7).

Physical data and contrast-detail phantoms cannot be applied to assess pulmonary lesions, but they may reveal some preliminary information. In one study (13), the observer performance of an FPD system using postprocessed film hard-copy images of a contrast-detail phantom obtained at 4.0 mAs were compared with the performance of those images obtained at 3.2 mAs. No differences between the two milliampere second groups were found, probably because of a similar high initial setting but a minimally reduced value compared with that used in our study. Geijer and co-workers (14), who evaluated image quality and radiation dose by using a contrast-detail phantom, had results comparable to ours. In their study, the image quality was significantly superior with use of the highest radiation dose setting (speed, 400; detector dose, 2.50 µGy), as compared with the image quality achieved at the intermediate (speed, 600; detector dose, 1.67 µGy) and lowest (speed, 800; detector dose, 1.25 µGy) dose settings.

A patient trial without proved lesions and without receiver operating characteristic analysis focused only on subjective image quality is probably not a reliable study to evaluate differences between different exposure parameters. In a study in which digital chest radiography was performed in patients at detector doses of 2.5 µGy (speed, 400) and 1.8 µGy (speed, 560), no significant differences were observed (16). In our study, however, the detector dose was further reduced to 1.56 µGy (speed, 640).

Our study results demonstrate the influence of different peak kilovoltage values on diagnostic performance. In addition, the level of this influence depended on the analyzed region in the current study. Over mediastinal and subphrenic areas, the detectability of lesions at the same detector dose tended to be better at lower peak kilovoltages. An explanation for this effect might be the better low-contrast resolution at lower energies. Furthermore, the detective quantum efficiency, and consequently the signal-to-noise ratio, increases with decreasing peak kilovoltage (9,17). This relationship confirms the better low-contrast resolution.

In contrast, over the pulmonary areas, lesion detectability at the same detector dose was significantly better as the peak kilovoltage values increased from 100 to 140 kVp. The reason for this might be the reduced visibility of the bones with higher beam energy. Although the signal-to-noise ratio decreases with increasing peak kilovoltage, in our study this effect did not substantially influence the detection of lesions over the high-contrast pulmonary areas.

In terms of limitations, in FPDs, the measurement of the dose in the ionization chamber is dependent on the tube voltage. For the detector used in the current study, the ionization chambers of the automatic exposure control were calibrated at 80 kVp and 3.0 µGy, resulting in a true detector radiation dose of 90% of the chosen dose at 100 kVp, of 85% of the chosen dose at 120 kVp, and of 83% of the chosen dose at 140 kVp. In a clinical setting similar to that simulated in our study, this influence cannot be eliminated. Thus, the detector doses used in our study (2.50, 1.56, and 1.25 µGy) do not represent true detector doses; rather, they are the doses that can be adjusted.

In addition, it has to be stressed that this was a phantom study with all of the inherent limitations. These include limited structured noise of the anthropomorphic chest phantom and the fact that the different physical sizes of patients were not simulated. Therefore, patient studies based on these findings are warranted to ensure that our study results will have an effect on clinical imaging. However, recent results indicate that it may be possible to reduce the radiation dose moderately—from 2.5 to 1.8 µGy—without decreasing image quality, even in the overall performance (16).

Practical application: In this experimental study focused on different radiation exposure parameters for posteroanterior chest radiography, significantly reduced diagnostic performance in the overall detection of simulated lesions was observed when the detector dose was decreased from 2.50 µGy to 1.56 and 1.25 µGy. Therefore, for standard posteroanterior chest radiography, a detector dose of 2.50 µGy still should be required. For the lung only—in example, for short-term follow-up examinations of interstitial lung disease or pulmonary metastases—a low-dose examination at a detector dose of 1.25 µGy (50% dose reduction) may be performed without decreasing diagnostic performance. By keeping the detector dose constant, one can reduce the effective dose by decreasing the peak kilovoltage. Hence, on the basis of these data and the described data on diagnostic performance, a peak kilovoltage value of 120 kVp was the most suitable.


   ACKNOWLEDGMENTS
 
We thank Gerhard Brunst, PhD, of GE Medical Systems, for his support and for providing the anthropomorphic chest phantom.


   FOOTNOTES
 
Abbreviations: AUC = area under receiver operating characteristic curve, ESD = entrance surface dose, FPD = flat-panel detector

Authors stated no financial relationship to disclose.

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


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 

  1. van Heesewijk HP, van der Graaf Y, de Valois JC, Feldberg MA. Digital chest imaging using a selenium detector: a simulated diffuse interstitial pulmonary disease detection study. Invest Radiol 1995; 30:300-305.[CrossRef][Medline]
  2. van Heesewijk HP, van der Graaf Y, de Valois JC, Vos JA, Feldberg MA. Chest imaging with a selenium detector versus conventional film radiography: a CT-controlled study. Radiology 1996; 200:687-690.[Abstract/Free Full Text]
  3. 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]
  4. Woodard PK, Slone RM, Gierada DS, Reiker GG, Pilgram TK, Jost RG. Chest radiology: depiction of normal anatomy and pathologic structures with selenium-based digital radiography versus conventional screen-film radiology. Radiology 1997; 203:197-201.[Abstract/Free Full Text]
  5. Woodard PK, Slone RM, Sagel SS, et al. Detection of CT proved pulmonary nodules: comparison of selenium-based digital and conventional screen-film chest radiographs. Radiology 1998; 209:705-709.[Abstract/Free Full Text]
  6. Strotzer M, Gmeinwieser JK, Völk M, Fründ R, Seitz J, Feuerbach S. Detection of simulated chest lesions with normal and reduced radiation dose: comparison of conventional screen-film radiography and a flat-panel x-ray detector based on amorphous silicon. Invest Radiol 1998; 33:98-103.[CrossRef][Medline]
  7. 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]
  8. 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]
  9. Floyd CE, Jr, Warp RJ, Dobbins JT, 3rd, et al. Imaging characteristics of an amorphous silicon flat-panel detector for digital chest radiography. Radiology 2001; 218:683-688.[Abstract/Free Full Text]
  10. Neitzel U, Maack I, Günther-Kohfhl S. Image quality of a digital chest radiography system based on a selenium detector. Med Phys 1994; 21:509-516.[CrossRef][Medline]
  11. Chotas HG, Floyd CE, Ravin CE. Technical evaluation of a digital chest radiography system that uses a selenium detector. Radiology 1995; 195:264-270.[Abstract/Free Full Text]
  12. Ravin CE, Chotas HG. Chest radiography. Radiology 1997; 204:593-600.[Free Full Text]
  13. 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]
  14. Geijer H, Beckmann KW, Andersson T, Persliden J. Image quality versus radiation dose for a flat-panel amorphous silicon detector: a phantom study. Eur Radiol 2001; 11:1704-1709.[CrossRef][Medline]
  15. 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]
  16. 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]
  17. Launders JH, Cowen AR, Bury RF, Hawkridge P. Towards image quality, beam energy and effective dose optimization in digital thoracic radiography. Eur Radiol 2001; 11:870-875.[CrossRef][Medline]
  18. International Commission on Radiological Protection. 1990 recommendations of the International Commission on Radiological Protection. Ann ICRP 1991; 21:1-201.
  19. Granfors PR. Performances characteristics of an amorphous silicon flat panel x-ray imaging detector. Proc SPIE 1999; 3659:480-490.[CrossRef]
  20. Williams G, Zankl M, Abmayr W, Veit R, Drexler G. The calculation of dose from external photon exposures using reference and realistic human phantoms and Monte Carlo methods. Phys Med Biol 1986; 31:449-452.[CrossRef][Medline]
  21. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978; 8:283-298.[Medline]
  22. Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 1993; 39:561-577.[Abstract/Free Full Text]
  23. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148:839- 843.[Abstract/Free Full Text]



This article has been cited by other articles:


Home page
 Br. J. Radiol.Home page
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]

Home page
 RadiologyHome page
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]

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 Metz, S.
Right arrow Articles by Link, T. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Metz, S.
Right arrow Articles by Link, T. M.

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