BJR
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
First published online February 16, 2009
British Journal of Radiology (2009) 82, 705-710
© 2009 British Institute of Radiology
doi: 10.1259/bjr/27942950
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
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
Google Scholar
Right arrow Articles by MOORE, C S
Right arrow Articles by BEAVIS, A W
PubMed
Right arrow PubMed Citation
Right arrow Articles by MOORE, C S
Right arrow Articles by BEAVIS, A W

Investigating the exposure class of a computed radiography system for optimisation of physical image quality for chest radiography

C S MOORE, BSc, MSc 1 J R SAUNDERSON, BSc, MSc 1,2 and A W BEAVIS, BSc, PhD 1,2,3

1 Radiation Physics Department, Hull and East Yorkshire Hospitals, Oncology Centre, Castle Hill Hospital, Cottingham, East Yorkshire, 2 Postgraduate Medical Institute, University of Hull, Kingston Upon Hull, 3 Faculty of Health and Wellbeing, Sheffield Hallam University, City Campus, Sheffield, UK

Correspondence: C S Moore, Radiation Physics Department, Hull and East Yorkshire Hospitals, Oncology Centre, Castle Hill Hospital, Cottingham, East Yorkshire, UK. E-mail: craig.moore{at}hey.nhs.uk


   Abstract
Top
 Abstract
Introduction
 Methods and materials
 Results and discussion
 Conclusions
 References
 
The purpose of this study was to investigate whether the exposure (speed) class (EC) of an Agfa computed radiography (CR) system could be used to optimise chest radiography. The frequency-dependent normalised noise–power spectra (NNPS(f)) were determined for a range of EC settings (25–1200) for a receptor dose of 4 µGy. Signal-to-noise ratios (SNRs) were measured in the lung, heart and diaphragm areas of a chest phantom with ECs of 400 and 600 at four tube voltages (60, 75, 90 and 125 kVp). As anatomical background can be a factor in detection of lung nodules, a tissue to rib ratio (TRR), which measures the ratio of pixel values in the nodule to that of rib, was measured in the lung region of the phantom to assess the suppression of the rib at ECs of 400 and 600. The NNPS(f) at ECs lower than 400 was relatively high. The NNPS(f) at EC 600 was found to be 7% lower when averaged over all frequencies than that at EC 400. The statistical significance of this difference was verified. The EC 800 and EC 1200 settings offered no extra advantages in terms of lowering frequency-dependent noise. The EC 600 setting offered improvements in SNR of between 10% and 18% in the lung, 11% and 16% in the heart, and 15% and 20% in the diaphragm compared with EC 400. Statistical analysis verified the significant difference. The EC 600 setting increased the TRR, thereby helping to suppress rib. This work indicates that an exposure class setting of 600 is the most appropriate for standard chest radiography, but clinical verification is required.


   Introduction
Top
 Abstract
 Introduction
Methods and materials
 Results and discussion
 Conclusions
 References
 
In optimising chest radiography, two important aspects need to be addressed. The first is to produce clinical images that are acceptable to the reporting radiologist, with enough information present to solve the diagnostic challenge and, therefore, to successfully manage the patient's clinical indications [1]. The second aspect is to minimise the stochastic radiation risk to the patient by keeping the effective dose to the patient as low as reasonably practicable (ALARP).

Many studies in the literature have concentrated on optimising tube voltage [2–5], with conflicting results. Tingberg and Sjostrom [6] imaged an anthropomorphic phantom at different tube voltages with a fixed effective dose. They used a visual grading analysis and concluded that the optimum tube voltage for computed radiography (CR) is lower than that recommended for film–screen systems [6]. Their work verified a study carried out by Chotas et al [7] in 1993. Samei et al [8] have recently reported that additional copper filtration, over and above the inherent filtration of the system, may provide optimal conditions for digital X-ray imaging of the chest. Work carried out by Moore et al [9] suggests that 0.1 mm of additional copper enhances the image quality of chest images acquired with a CR system at a given receptor dose.

The correlation between physical (signal-to-noise ratio – SNR) and clinical (observer studies) image quality has recently been investigated by Sandborg et al [10]. They concluded that there may be a positive correlation between SNR and clinical image quality in chest radiography, but their method did not include anatomical noise. The role of patient thickness [11] and the effect of an anti-scatter grid [7] on image quality in chest radiography with CR have also been investigated using Monte Carlo studies. It is therefore clear that most of the literature is focused on optimising X-ray exposure factors for chest CR. However, we feel that the optimisation of CR processing has, to some extent, been neglected.

In this study we attempt to answer part of this problem by investigating whether changes to the EC of an Agfa 75.0 CR system (Agfa, Peissenberg, Germany) can be used to increase a physical measure of image quality (SNR) and a measure of nodule detectability (tissue to rib ratio – TRR) over the clinical effective dose range. We also investigate the effect of ECs on the normalised noise–power spectra (NNPS) of the system.

It is important not to confuse the ECs used here with the traditional film–screen terminology of speed. Although they have similar names, film speed determines the amount of exposure required for a given level of optical density, whereas CR speed (or exposure class in the case of Agfa) indicates to the system the magnitude of the signal expected on the CR imaging plate, as discussed below.

If the user-selectable EC chosen is low (i.e. 25–200), then the system will decrease its sensitivity; gain on the photomultiplier tube (PMT) is decreased and therefore the signal recorded is low. This is because the expected signal reaching the imaging plate is high (i.e. high incident X-ray beam intensity). Conversely, if the EC chosen is high (i.e. 800–1200), the expected signal reaching the imaging plate is weaker (i.e. lower incident X-ray beam intensity), and the scanner increases its sensitivity (PMT gain is increased).

The scanner has a dynamic range of 4096 pixel values (0–4095) but images are always displayed over a "useful" pixel value range, and this range is positioned by the EC setting to encompass the image signal anticipated from the imaging plate.

If a higher EC is chosen, the useful pixel value range is compressed and it is positioned higher on the dynamic range axis to encompass the anticipated latent image signal that will be detected. Examples of this process are demonstrated in GoGoFigures 1–3Go. Figure 1Go shows a histogram of pixel values for a low EC setting (e.g. 200), Figure 2Go shows a medium EC setting (e.g. 400) and Figure 3Go shows a high EC setting (e.g. 1200). It can be seen that, as the EC is increased, the useful range of pixel values is compressed and shifted to the higher end of the dynamic range.


Figure 1
View larger version (7K):
[in this window]
[in a new window]

  		Figure 1. Histogram of pixel values for a low EC setting.

 

Figure 2
View larger version (6K):
[in this window]
[in a new window]

  		Figure 2. Histogram of pixel values for a medium EC setting.

 

Figure 3
View larger version (6K):
[in this window]
[in a new window]

  		Figure 3. Histogram of pixel values for a high EC setting.

 
The default exposure class in our radiology department for chest examinations is 400, so we investigated whether any improvement in physical image quality is achieved by using other exposure classes.


   Methods and materials
Top
 Abstract
 Introduction
 Methods and materials
Results and discussion
 Conclusions
 References
 
All tests were performed in a general-purpose X-ray room equipped with a Philips Optimus Diagnost TH (Philips Medical Systems, Surrey, UK) ceiling-suspended X-ray system, with a total inherent filtration equivalent to 3 mm of aluminium, and using an Agfa 75.0 CR reader with MD4.0 plates (35 cm x 43 cm, effective pixel pitch of 0.167 mm). It has been reported recently [12, 13] that the use of anti-scatter grids with digital imaging is not justified because of the resulting increase in patient dose without a corresponding increase in image quality. Following this, and in line with the current protocol used for chest imaging in our radiology department, an anti-scatter grid was not used. Also, only manually selected exposure factors were used for this investigation.

Effect of exposure class on the normalised noise–power spectrum
The NNPS is a widely used method in determining the noise characteristics of digital imaging systems [14–18]. We therefore decided to investigate what effect changing the exposure class had on the NNPS. The frequency-dependent NNPS (NNPS(f)) is defined as the variance per frequency bin of a stochastic signal (random signal owing to noise sources in the image) in the spatial frequency domain. It must be derived from linear, uniform, non-processed images [19–21] so a uniform Perspex phantom 10 cm thick was used to provide attenuation and scatter similar to that of an adult chest [2]. The phantom was positioned in front of an imaging plate and a focus to imaging plate distance of 180 cm was used. The phantom was exposed at 90 kVp and with a sufficiently high current (mA) to provide 4.00 µGy (± 0.02; mean ± 1 SD) of dose incident at the imaging plate. After a delay of 1 min, the imaging plate was read out using the EC 400 setting and flat field (without post-processing – i.e. all MUSICA (multiscale image contrast amplification) parameters are set to zero). The NNPS(f) was subsequently calculated using the standard method described by Dobbins et al [14] from the resulting image after it had been corrected for the logarithmic amplification by the photomultiplier tube in the CR system (i.e. the image was linearised). The anode heel effect was corrected in the flat-field image using in-house software. This was repeated for images processed with exposure classes 25, 50, 100, 400, 600, 800 and 1200.

Effect of exposure class on the SNR in the phantom using clinical post-processing
As the NNPS(f) examines noise properties of images without any post-processing, they are somewhat removed from the clinical situation, since all clinical images are evaluated with post-processing. It was therefore felt necessary to study the effect of exposure class on the important areas of the chest using an anthropomorphic equivalent phantom (see below) with clinically processed images.

Clinical phantom used for the study
A radiograph of the phantom used for the study is shown in Figure 4Go. It was based upon the LucAl phantom (Standard Dosimetric/Calibration Phantom; Centre for Devices and Radiological Health, Carson, CA, USA), described by Conway et al [22], with additional anatomical inserts as developed by Vassileva [23]. The LucAl phantom consisted of 300 mm x 300 mm plates of polymethylmethacrylate (PMMA) and 1100 alloy aluminium (Al). The overall thickness of the phantom was 267.1 mm, with a total of 4.1 mm Al, 73 mm PMMA and a 190 mm air gap. For a complete description of the phantom and its subsequent validation, the reader should refer to references 22–25.


Figure 4
View larger version (75K):
[in this window]
[in a new window]

  		Figure 4. A radiograph of the phantom. The aluminium strip and wax insert for the tissue-to-rib ratio calculation are visible in the lung region. Region of interests (ROIs) are shown in blue. This is where the SNR indices were measured.

 
Unless otherwise stated in subsequent sections, the phantom was exposed under typical conditions in accordance with local practice, i.e. with a focus to cassette distance of 180 cm and with the CR plate placed 5 cm behind the phantom in the cassette holder. The X-ray field was collimated to the edges of the phantom. Since different individual imaging plates do not have exactly matching sensitivities, a single CR cassette was used throughout the study. After exposure, a 1 min delay was observed before the CR cassette was read out using the following digitiser acquisition settings as standard for chest radiography:

The following MUSICA parameters were used to process each image:

These parameters are used as standard in our radiology department for chest examinations, and were set to these values at commissioning by Agfa.

Four tube voltage values across the diagnostic energy range were investigated: 60, 75, 90 and 125 kVp. These tube potentials range from the lowest used in our radiology department (60 kVp) to that recommended by CEC [1] for film–screen radiography (125 kVp). We therefore felt it prudent to investigate this range, rather than just the kilovoltages used in our institution, as it is likely that others will still be using the tube voltage recommended by CEC.

Measurements of air kerma at the cassette were made behind the lung region of the phantom, as chest examinations are traditionally controlled by exposure to the lung fields. The CR plate was removed and replaced with a calibrated 6 cm3 ionisation chamber (Radcal Corporation, Monrovia, CA, USA) positioned 5 cm behind the lung region using the seven mAs settings to encompass the required air kerma range. The mAs settings were selected to give measured air kerma values at the cassette between approximately 2 µGy and 15 µGy at each tube voltage, as these are typically encountered in our institution for manually derived exposure factors. The mAs settings chosen ranged from 3–50 mAs at 60 kVp, from to 1–32 mAs at 75 kVp, from 1–20 mAs at 90 kVp and from 1–13 mAs at 125 kVp. A number of exposures of the test cassette were then made using the same range of mAs values. The dose–area product (DAP) was measured for each exposure in order to calculate the effective dose [26]. After each of these exposures, the cassette was read and processed with the chest CR acquisition parameters and the 400 exposure class. Using in-house software [27], the SNR was calculated in the lung, heart/spine and diaphragm regions of the chest using the equation SNR  =  S/{sigma}, where S is the mean pixel value in the region of interest and {sigma} is the standard deviation. The relationship between the SNR and effective dose at each peak kilovoltage was examined. This methodology was repeated for exposure class 600. Only exposure class settings of 400 and 600 were examined as these were deemed to be the most appropriate with respect to the results derived from the NNPS investigation.

As described in the introduction, increasing the EC can compress and shift the clinically useful data higher up the dynamic range of the CR system. If one is not careful, the use of high ECs for a given exposure may result in data being "clipped". To investigate how the EC 600 setting compresses and shifts the clinically useful data compared with the EC 400 setting, the phantom was exposed twice with a tube potential of 90 kVp and a DAP sufficient to ensure an effective dose of approximately 0.02 mSv (i.e. a typical adult effective dose from a chest examination). After each exposure the cassette was read as stated above, first with EC 400, then with EC 600. The histogram of pixel values was plotted for each subsequent image and compared.

Effect of exposure class on rib contrast in the lung
It has been reported by Hoeschen et al [28] that anatomical background is a factor in observers' detection of nodules. Assuming that ribs are a part of the anatomical background that interferes with the detection of nodules, it would be an advantage to suppress the contrast of the ribs, but only if this is not going to affect other image details. Anatomical noise includes not only rib but overlying tissue that can lead to details being disguised. However, purely from the point of view of rib suppression, we felt it worth investigating. A 3 mm thick aluminium strip [23] was added to the lung region of the phantom to simulate a rib, and a 2 x 2 x 1 cm block of paraffin wax to simulate a soft-tissue nodule [29]. This nodule size is similar to those typically encountered clinically [30]. The ratio of mean pixel value in wax to that of aluminium (the TRR) was evaluated at the 400 and 600 exposure class settings for each of the tube voltages investigated. The TRR was found to be a function of effective dose at each setting and tube voltage.


   Results and discussion
Top
 Abstract
 Introduction
 Methods and materials
 Results and discussion
Conclusions
 References
 
Effect on exposure class on the normalised noise–power spectrum
The NNPS(f) as a function of spatial frequency for exposure classes 25, 50, 100, 400 and 600 is shown in Figure 5Go. For clarity, ECs 800 and 1200 are not displayed, but neither offered further advantages in terms of reducing frequency-dependent noise with respect to EC 600.


Figure 5
View larger version (18K):
[in this window]
[in a new window]

  		Figure 5. Plot of the frequency-dependent normalised noise–power spectrum for each image processed with exposure classes 25 (thin solid line), 50 (dotted line), 100 (uneven dashed line), 400 (even dashed line) and 600 (thick dashed line).

 
It is evident from Figure 5Go that images read with low exposure classes (25, 50 and 100) have a higher noise content than those read with higher exposure classes. EC 600 has slightly lower noise characteristics than the image processed with the default 400 EC setting. The NNPS(f) is between 5% and 10% lower at EC 600 than EC 400, and approximately 7% lower when averaged over all spatial frequencies for the same dose to the imaging plate. To test for significant difference, Student's two-tailed t-test was performed on the data measured with ECs 400 and 600. A p-value of <0.1 x 10–5 was calculated and therefore noise at EC 600 is statistically significantly different from that at EC 400.

Effect of exposure class on the SNR in the phantom using clinical post-processing
As exposure classes lower than 400 generate images with a relatively high noise content (NNPS(f)), and those images generated with ECs greater than 600 offer no advantage in terms of lowering the NNPS(f), only ECs 400 and 600 were investigated with the "clinical" phantom.

GoGoGoFigures 6–9Go show a logarithmic (loge) fit of the SNR in the lung, heart/spine and diaphragm "anatomical compartments" of the phantom as a function of effective dose at each tube potential.


Figure 6
View larger version (19K):
[in this window]
[in a new window]

  		Figure 6. Plot of the signal-to-noise ratio at 60 kVp in the lung (400, uneven dashed line; 600, dots), heart (400, thick solid line; 600, thick dashed line) and diaphragm (400, thin solid line; 600, thin dashed line) regions of the phantom.

 

Figure 7
View larger version (21K):
[in this window]
[in a new window]

  		Figure 7. Plot of the signal-to-noise ratio at 75 kVp in the lung (400, uneven dashed line; 600, dots), heart (400, thick solid line; 600, thick dashed line) and diaphragm (400, thin solid line; 600, thin dashed line) regions of the phantom.

 

Figure 8
View larger version (21K):
[in this window]
[in a new window]

  		Figure 8. Plot of the signal-to-noise ratio at 90 kVp in the lung (400, uneven dashed line; 600, dots), heart (400, thick solid line; 600, thick dashed line) and diaphragm (400, thin solid line; 600, thin dashed line) regions of the phantom.

 

Figure 9
View larger version (20K):
[in this window]
[in a new window]

  		Figure 9. Plot of the signal-to-noise ratio at 125 kVp in the lung (400, uneven dashed line; 600, dots), heart (400, thick solid line; 600, thick dashed line) and diaphragm (400, thin solid line; 600, thin dashed line) regions of the phantom.

 
As can be seen from these figures, the SNR (at all effective doses) in the images processed with EC 600 (SNR600) is greater than in those images processed at EC 400 (SNR400) in each compartment of the chest. Table 1Go shows a summary of the percentage increase in SNR in each region of the phantom at each tube potential. The SNRs found at the two ECs were compared using Student's two-tailed t-test to test for significance. All observations were found to be statistically significant at the >95% confidence limit (p<0.05). The largest p-value calculated was at 90 kVp in the diaphragm (p = 0.04).


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

  		Table 1. Mean increase in SNR in each region of the phantom with images processed at EC 600 rather than EC 400

 
Interestingly, although processing algorithms can affect pixel values and noise in a way that may be hard to predict, these results suggest that there is a logarithmic relationship between the SNR and the effective dose in each region of the chest (all R2 >0.95), for a given set of pre-processing and MUSICA parameters.

Figure 10Go shows the histograms of the phantom with images exposed at ECs 400 and 600. As can be seen, processing the image with EC 600 shifts the useful clinical data only slightly higher up the dynamic range and compresses the data very little.


Figure 10
View larger version (22K):
[in this window]
[in a new window]

  		Figure 10. Histogram plots of the phantom. The plot above is derived from the image processed at exposure class (EC) 400 and the plot below is derived from the image processed at EC 600.

 
Effect of exposure class on rib contrast in the lung
Figure 11Go demonstrates the effect of exposure class on the TRR at each tube potential investigated. It can quite clearly be seen that EC 600 slightly increases the TRR at each tube voltage compared with EC 400. The increase is <1% at each tube voltage so it is likely that clinical benefits are negligible, but it demonstrates that EC 600 produces TRRs equal to or greater than (i.e. no worse) those produced by EC 400.


Figure 11
View larger version (20K):
[in this window]
[in a new window]

  		Figure 11. Plot of the effect of exposure class (EC) setting on the tissue to rib ratio at each tube voltage investigated. 60 kVp and EC 400/600 are represented by a diamond marker and solid/dashed lines. 75 kVp and EC 400/600 are represented by triangle markers and solid/dashed lines. 90 kVp and EC 400/600 are represented by square markers and solid/dashed lines. 125 kVp and EC 400/600 are represented by cross markers and solid/dashed lines.

 

   Conclusions
Top
 Abstract
 Introduction
 Methods and materials
 Results and discussion
 Conclusions
References
 
This study has shown that processing images with ECs lower than 400 increases their frequency-dependent noise (NNPS(f)) content. However, processing images at EC 600 slightly reduces the NNPS(f) of the CR system (compared with EC 400 images) and improves the physical image quality indices, SNR and TRR, in the chest when images of a clinical relevant phantom are processed with the chest processing parameters of the system. Processing images with ECs of 800 and 1200 offers no further advantages in terms of reducing noise. The results suggest that processing clinical chest images with EC 600, rather than EC 400, would appear to be optimum over a wide range of effective doses (0.05–0.8 mSv). Such a change in clinical exposure class setting would require clinical validation before going into routine clinical use.

Received for publication May 2, 2008. Revision received September 12, 2008. Accepted for publication September 15, 2008.


   References
Top
 Abstract
 Introduction
 Methods and materials
 Results and discussion
 Conclusions
 References
 

  1. Council of European Communities. European guidelines on quality criteria for diagnostic radiographic images, Rep EUR 16260 EN, 1996.
  2. Honey ID, MacKenzie A, Evans DS. Investigation of optimum energies for chest imaging using film-screen and computed radiography. Br J Radiol 2005;78:422–7.[Abstract/Free Full Text]
  3. Nobuhiro O, Hajime N, Seiichi M, Kunihiro T, Katsumi N, Akira Y. Optimal beam quality for chest radiography. Invest Radiol 1996;31:126–31.[CrossRef][Medline]
  4. Dobbins JT, Rice JJ, Beam CA, Ravin CE. Threshold perception performance with computed radiography: implications for chest radiography. Radiology 1992;183:179–87.[Abstract/Free Full Text]
  5. Tingberg A, Sjostrom D. Search for optimal tube voltage for image plate radiography. Invest Radiol 1996;31:126–31.[CrossRef][Medline]
  6. Tingberg A, Sjostrom D. Optimisation of image plate radiography with respect to tube voltage. Radiat Prot Dosimetry 2005;114:286–93.[Abstract/Free Full Text]
  7. Chotas HG, Floyd CE, Dobbins JT, Ravin CE. Digital chest radiography with photostimulable storage phosphors: signal-to-noise ratio as a function of kilovoltage with matched exposure risk. Radiology 1993;186:395–8.[Abstract/Free Full Text]
  8. Samei E, Dobbins III JT, Lo JY, Tornai MP. A framework for optimising the radiographic technique in digital X-ray imaging. Radiat Prot Dosimetry 2005;114:220–9.[Abstract/Free Full Text]
  9. Moore CS, Beavis AW, Saunderson JR. Investigation of optimum X-ray beam tube voltage and filtration for chest radiography with a Computed Radiography system. Br J Radiol 2008;81:771–7.[Abstract/Free Full Text]
  10. Sandborg M, Tingberg A, Gustaf U, Dance DR, Carlsson GA. Comparison of clinical and physical measures of image quality in chest and pelvis computed radiography at different tube voltages. Med Phys 2006;33:4169–75.[CrossRef][Medline]
  11. Gustaf U, Sandborg M, Dance DR, Hunt R, Carlsson GA. The influence of patient thickness and imaging system on patient dose and the physical image quality in digital chest imaging. Radiat Prot Dosimetry 2005;114:294–7.[Abstract/Free Full Text]
  12. Roberts JA, Evans SC, Rees M. Optimisation of imaging technique using direct digital radiography. J Radiol Prot 2006;26:287–99.[CrossRef][Medline]
  13. Bernhardt TM, Rapp-Bernhardt U, Hausmann T, Reichel G, Krause UW, Doehring W. Digital selenium radiography: anti-scatter grid for chest radiography in a clinical study. Br J Radiol 2000;73:963–8.[Abstract]
  14. Dobbins JT, Samei E, Ranger NT, Chen Y. Intercomparison of methods for image quality characterisation. II. Noise power spectrum. Med Phys 2006;33:1466–75.[CrossRef][Medline]
  15. Hillen W, Schiebel U, Zaengel T. Imaging performance of a digital storage phosphor system. Med Phys 1987;14:744–51.[CrossRef][Medline]
  16. Dobbins JT, Ergun DL, Rutz L, Hinshaw DA, Blume H, Clark DC. DQE(f) of four generations of computed radiography acquisition devices. Med Phys 1995;22:1581–93.[CrossRef][Medline]
  17. Kengyelics SM, Launders JH, Cowen AR. Physical imaging performance of a compact computed radiography acquisition device. Med Phys 1998;25:354–60.[CrossRef][Medline]
  18. Redlich U, Hoeschen C, Doehring W. Assessment and optimisation of image quality of chest radiography systems. Radiat Prot Dosim 2005;114:264–8.[Abstract/Free Full Text]
  19. Mackenzie A, Honey ID. Characterization of noise sources for two generations of computed radiography systems using powder and crystalline photostimulable phosphors. Med Phys 2007;34:3345–57.[CrossRef][Medline]
  20. Flynn MJ, Samei E. Experimental comparison of noise and resolution for 2k and 4k storage phosphor radiography systems. Med Phys 1999;26:1612–23.[CrossRef][Medline]
  21. Samei E, Flynn MJ. An experimental comparison of detector performance for direct and indirect digital radiograph systems. Med Phys 2003;30:608–22.[CrossRef][Medline]
  22. Conway BJ, Butler BF, Duff JE, Fewell TR, Gros RE, Jenning RJ, et al. Beam quality independent attenuation phantom for estimating patient exposure from X-ray automatic exposure controlled chest examinations. Med Phys 1984;11:827–32.[CrossRef][Medline]
  23. Vassileva J. A phantom for dose-image quality optimization in chest radiography. Br J Radiol 2002;75:837–42.[Abstract/Free Full Text]
  24. Servomaa A, Tapiovaara M. Patient equivalent phantoms in chest radiography. In: Moores B, Stieve F, Eriskat H, Schibilla H, editors. Technical and Physical Parameters for Quality assurance in Medical Diagnostic Radiology: Tolerances, Limiting Values and Appropriate Measuring Methods. BIR Report 18. London: British Institute of Radiology, 1989.
  25. Shrimpton PC, Wall BF, Fisher FS. The tissue-equivalence of the Alderson-Rando anthropomorphic phantom for X-rays of diagnostic qualities. Phys Med Biol 1989;26:133–9.[CrossRef]
  26. Le Heron JC. XDOSE X-ray Radiography Dosimetry Program using NRPB SR-262 Organ Doses. Christchurch, New Zealand: National Radiation Laboratory, Ministry of Health, 1994.
  27. Moore CS, Liney GP, Beavis AW, Saunderson JR. A method to optimise the processing algorithm of a computed radiography system for chest radiography. Br J Radiol 2007;80:724–30.[Abstract/Free Full Text]
  28. Hoeschen C, Tischenko O, Buhr E, Illers H. Comparison of technical and anatomical noise in digital thorax X-ray images. Radiat Prot Dosimetry 2005;114:75–80.[Abstract/Free Full Text]
  29. International Commission on Radiological Units and Measurements. Tissue Substitutes in Radiation Dosimetry. ICRU Report 44. Bethesda, MD:ICRU, 1989.
  30. Reeves AP, Chan AB, Yankelevitz DF, Henschke CI, Kressler B, Kostis WJ. On measuring the change in size of pulmonary nodules. IEEE Trans Med Imaging 2006;25:435–50.[CrossRef][Medline]




This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Services
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
Google Scholar
Right arrow Articles by MOORE, C S
Right arrow Articles by BEAVIS, A W
PubMed
Right arrow PubMed Citation
Right arrow Articles by MOORE, C S
Right arrow Articles by BEAVIS, A W

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
BJR DMFR IMAGING  ALL BIR JOURNALS