Optimizing Image Quality and Dose for Digital Radio

Technik und Medizinphysik

643

Optimizing Image Quality and Dose for Digital Radiography of Distal Pediatric Extremities Using the Contrast-to-Noise Ratio Optimierung von Bildqualität und Dosis für die digitale Radiografie distaler kindlicher Extremitäten auf Grundlage des Kontrast-Rausch-Verhältnisses Authors

R. Hess1, U. Neitzel2

Affiliations

1

University of Applied Science, Hamburg Philips Healthcare, Clinical Science Diagnostic X-ray, Hamburg

Key words

Zusammenfassung

Abstract

"

!

!

Ziel: Ziel der Untersuchung war es, den Einfluss von Röntgenröhrenspannung und Vorfilterung auf Bildqualität, ausgedrückt durch das KontrastRausch-Verhältnis (CNR), und die Dosis bei digitalen Röntgenaufnahmen distaler kindlicher Extremitäten zu untersuchen und die Bedingungen zu bestimmen, bei denen das beste Verhältnis von CNR und Patientendosis erreicht wird. Material und Methoden: Mittels eines Phantoms, welches die Absorptionseigenschaften kindlicher distaler Extremitäten simuliert, wurden das CNR und die damit verknüpfte Patientendosis für Röhrenspannungen von 40 bis 66 kV mit und ohne Zusatzfilter von 0,1 mm Cu/1 mm Al bestimmt. Das gemessene CNR diente als Maß für die Bildqualität, während die mittlere absorbierte Dosis (MAD) aus einer Kombination von Messung und Simulation als Maß für die Patientendosis ermittelt wurde. Ergebnisse: Das vorteilhafteste Verhältnis von CNR und Dosis ergab sich für die niedrigste untersuchte Röhrenspannung (40 kV) und ohne Zusatzfilterung. Im Vergleich zu Aufnahmen bei 50 kV und 60 kV konnte die mittlere absorbierte Dosis bei Beibehaltung der Bildqualität (des CNR) um 24 % bzw. 50 % gesenkt werden. Schlussfolgerung: Durch Verwendung niedrigerer Röhrenspannungen erscheint für die digitale Radiographie distaler kindlicher Extremitäten eine weitergehende Optimierung von CNR und Dosis möglich. Die Relevanz der Ergebnisse für klinische Anwendungen muss weiter überprüft werden.

Purpose: To investigate the influence of X-ray tube voltage and filtration on image quality in terms of contrast-to-noise ratio (CNR) and dose for digital radiography of distal pediatric extremities and to determine conditions that give the best balance of CNR and patient dose. Materials and Methods: In a phantom study simulating the absorption properties of distal extremities, the CNR and the related patient dose were determined as a function of tube voltage in the range 40 – 66 kV, both with and without additional filtration of 0.1 mm Cu/1 mm Al. The measured CNR was used as an indicator of image quality, while the mean absorbed dose (MAD) – determined by a combination of measurement and simulation – was used as an indicator of the patient dose. Results: The most favorable relation of CNR and dose was found for the lowest tube voltage investigated (40 kV) without additional filtration. Compared to a situation with 50 kV or 60 kV, the mean absorbed dose could be lowered by 24 % and 50 %, respectively, while keeping the image quality (CNR) at the same level. Conclusion: For digital radiography of distal pediatric extremities, further CNR and dose optimization appears to be possible using lower tube voltages. Further clinical investigation of the suggested parameters is necessary.

Introduction

graphic contrast because of the still immature bone development. On the other hand children are more sensitive to radiation than adults because of the more rapidly dividing cells and longer life expectancy [1]. As there is usually a trade-

● extremities ● digital radiography ● image quality ● patient dose ● contrast-to-noise ratio (CNR) ● mean absorbed dose (MAD) " " " " "

eingereicht 5.11.2011 akzeptiert 31.3.2012 Bibliography DOI http://dx.doi.org/ 10.1055/s-0032-1312727 Published online: 22.5.2012 Fortschr Röntgenstr 2012; 184: 643–649 © Georg Thieme Verlag KG Stuttgart · New York · ISSN 1438-9029 Correspondence Prof. Dr. Robert Hess University of Applied Science Hamburg Berliner Tor 7 20099 Hamburg Germany Tel.: ++ 49/40/4 28 75 81 70 [email protected]

!

Pediatric skeletal imaging using X-rays poses specific challenges. On the one hand the anatomy of very young patients shows only limited radio-

Hess R, Neitzel U. Optimizing Image Quality … Fortschr Röntgenstr 2012; 184: 643–649

b This is a copy of the author's personal reprint

2

644

This is a copy of the author's personal reprint

b


off between image quality and patient X-ray dose, optimization of imaging parameters must be guided by the ALARA (as low as reasonably achievable) principle, meaning that the dose to the patient should be at the lowest level that still guarantees a sufficient diagnostic image quality [2, 3]. An important aspect of optimization is the selection of the most suitable radiation beam quality, defined by the X-ray tube voltage and added beam filtration. Many authors have explored this aspect for various imaging tasks and for both conventional and digital techniques [4 – 12]. Early investigations showed that a reduction of entrance dose by 51 % can be achieved for infant pelvis AP radiographs by adding a copper filter and increasing the tube potential [5]. The situation is less clear regarding the effective dose, which is the more relevant parameter for the quantification of the stochastical radiation risk. Additional copper filtration may or may not reduce the effective dose, depending on the examination considered [13]. Increased tube voltages lead to a reduced patient dose when using a given image receptor with a specific sensitivity or receptor dose requirement. However, added filtration and higher tube voltage generally have a negative effect on the image quality since the harder beam quality leads to contrast loss in the image. While this side effect is usually accepted as unavoidable and often found to be clinically not significant, it means that dose reduction in conventional imaging is intrinsically associated with image quality reduction. While conventional imaging is characterized by the given dose requirements and the contrast limitations of the film as image receptor and display medium, the decoupling of image acquisition and display with digital X-ray systems provides a new degree of freedom for X-ray imaging optimization. Using a Monte Carlo simulation approach, Tapiovaara et al. showed in a seminal paper [14] that patient dose can be minimized for fixed image quality (expressed as detail signal-to-noise ratio) if the image receptor kerma is allowed to vary. They found that under these conditions the most dose-efficient spectrum is obtained by using a relatively low X-ray tube voltage (50 – 60 kV) together with heavy beam filtration (up to 0.5 mm Cu). Similar approaches have been used in other studies of digital imaging systems, mostly for the optimization of adult chest, pelvis, or breast examinations [12, 15 – 17]. The aim of this study was to investigate the influence of tube voltage and added copper/aluminum filtration on image quality and the patient dose of pediatric X-ray examinations of distal extremities. While the patient dose may be of comparatively minor concern for extremity imaging since these body parts do not contain highly X-ray sensitive organs and the level of radiation in general is rather low for this application, image quality is of high importance for detecting small bone fissures. A phantom setup was used to simulate the X-ray imaging situation of distal pediatric extremities. The intention was to determine which radiation condition, i. e., which tube voltage and filter configuration, maximizes the image quality for a given, fixed patient dose. Alternatively, the study protocol can also be used to minimize the patient dose for a fixed image quality level. Similar to previous studies [12, 14 – 18], the contrast-to-noise ratio (CNR) was used as an image quality metric, while the mean absorbed dose (MAD) in the phantom served as a surrogate for the patient dose. The MAD quantifies the energy imparted to the irradiated object and is a relative measure for the stochastic radiation risk. Hence, two images acquired with different tube voltages and pre-filter configurations but the same MAD will carry equal patient risks.

Fig. 1 Abb. 1

Sketch of phantom study setup. Schematischer Messaufbau für die Phantomstudie.

Materials and Methods !

Experimental setup The X-ray imaging situation was simulated using a phantom setup. An aluminum plate with a thickness of 1 mm representing the absorption properties of a bone detail was placed on the entry surface of a polymethyl methacrylate (PMMA) block with a thickness of 1 cm simulating the soft tissue absorption of distal pedia" Fig. 1). The contrast step generated by the Al tric extremities (● plate was used to determine the CNR in the image. All imaging was done using a digital radiography system (DigitalDiagnost VM, Philips Healthcare) with a large area flat-panel detector (Pixium 4600, Trixell). The detector employs a cesium iodide (CsI) scintillator as the X-ray converter and an amorphous silicon read-out array with a pixel sampling pitch of 0.143 mm. Images were acquired at various tube voltages in the range from 40 kV to 66 kV. A filter sandwich of 0.1 mm Cu/1 mm Al could be moved in and out of the beam using the standard filter wheel in the X-ray tube collimator. For each tube voltage and filter selection, four current-time-product values (mAs) were chosen to " Table 1, 2). For all image cover the MAD range from 5 to 50 μGy (● acquisitions the source-to-image-receptor distance (SID) was kept at 180 cm. While this is a larger SID than is normally used for clinical pediatric imaging, it reduces intensity variations in the image due to the Heel effect and the inverse square law. The influence of the enlarged SID on the CNR at a given fixed patient dose is assumed to be negligible. No anti-scatter grid was employed in the study reported here. The images were processed using standard image pre-processing (correction for gain and offset, pixel drop-out correction). No processing for image display (windowing, edge enhancement etc.) was applied. The resulting original data were used for the evaluation of the contrast-to-noise ratio in the image.

Dose determination In this article the MAD was used as an indicator of the dose to the phantom. While being a relevant dose parameter to quantify the radiation risk, the MAD cannot be easily measured in practice. Instead the MAD for a given exposure was derived from the X-ray tube current-time-product (mAs) using a calibrated X-ray spectrum simulation. The radiation incident on the phantom was spectrally simulated using the Birch-Marshal model [19]. This model allows simula-



current-time-

exposure time

entrance dose

mean absorbed

(kV)

product (mAs)

(msec)

(µGy)

dose (µGy)

CNR

Table 1 Measurements without pre-filter. Tab. 1 ter.

Messungen ohne Vorfil-

40

2.4

12.5

6.7

4.8

14.1

40

5.5

28

15.5

11.0

22.8

40

12.4

62.5

34.8

24.9

34.1

40

24.9

125

69.9

49.9

44.8

44

1.7

7.8

6.6

4.8

13.9

44

3.9

17.3

15.1

11.0

22.4

44

7.9

34.7

30.6

22.4

30.4

44

17.9

78.1

69.4

50.7

41.1

50

1.1

4.3

6.2

4.7

13.1

50

2.7

10.1

15.1

11.5

20.9

50

5.5

20.2

30.8

23.5

28.8

50

12.4

45.1

69.5

53.0

37.9

57

0.7

2.4

5.5

4.3

12.0

57

1.7

5.4

13.3

10.6

18.4

57

3.9

12.1

30.5

24.2

25.7

57

7.9

24.2

61.8

49.1

32.3

66

0.5

1.5

5.5

4.5

11.7

66

1.1

3.1

12.1

10.0

16.8

66

2.4

6.6

26.3

21.7

21.9

66

5.5

14.8

60.3

49.8

28.6

CNR

Table 2 Measurements with prefilter: 1 mm Al/0.1 mm Cu. Tab. 2 Messungen mit Vorfilter: 1 mm Al/0,1 mm Cu.

tube voltage

current-time-

exposure time

entrance dose

mean absorbed

(kV)

product (mAs)

(msec)

(µGy)

dose (µGy)

40

11.1

56

6.8

5.0

14.6

40

24.9

125

15.2

11.3

22.1

40

49.9

250

30.4

22.6

30.2

40

111.9

560

68.2

50.6

40.6

44

7.0

30.8

6.8

5.2

14.5

44

15.9

69.4

15.5

11.9

21.6

44

31.4

136.7

30.5

23.5

29.3

44

70.9

308.1

69.0

53.1

38.2

50

3.9

14.4

6.5

5.3

13.5

50

7.9

28.8

13.1

10.7

19.1

50

17.9

64.9

29.7

24.2

26.3

50

39.9

144.4

66.3

54.0

33.8

57

2.1

6.6

5.6

4.8

11.7

57

4.9

15.1

13.1

11.2

17.6

57

9.9

30.2

26.5

22.6

23.2

57

22.3

67.7

59.8

50.9

29.5

66

1.3

3.7

5.6

5.0

10.9

66

3.0

8.2

13.0

11.4

15.9

66

6.2

16.6

26.9

23.6

20.4

66

13.9

37

60.2

53.0

25.9

tion of the X-ray tube output spectrum as a function of tube voltage, tube loading, anode angle, and filtering material in the beam (tube window, permanent and added filtration). In addition to the X-ray spectrum, the simulation also delivers integrated quantities like the overall energy density in the beam and the corresponding air kerma. Two model parameters were used to adapt the simulated tube output (in terms of air kerma) to the measured kerma free-inair: the thickness d of the tungsten or copper pre-filter to adapt the trend over the tube voltage and a constant scale parameter c to adjust the overall output. The values for d and c were separately adjusted for the situations with and without an added pre-filter in an empirical way. Within one setup the parameters were kept constant for all tube voltages.

b

The specific X-ray tube output for the experimental setup was measured with a calibrated dosimeter (Unfors Xi). The measuring probe including back scatter shielding was placed free-in-air at a distance of 1 m from the focal spot approximately 30 cm above the X-ray table. To minimize scatter effects, the X-ray beam was narrowly collimated around the measuring probe. For tube voltages from 40 to 66 kV and a setting of 10 mAs, the kerma free-inair was measured and divided by the actual tube current-timeproduct displayed on the system. The calibrated simulated X-ray spectrum served as input to a Monte Carlo-based simulation tool to determine the backscattered, absorbed and transmitted X-ray energy fraction within the phantom. The absorbed energy Eabs was required to calculate the mean absorbed dose (MAD):



tube voltage

645

646


Fig. 2

Regions of interest (ROI) to measure the contrast-to-noise ratio.

Abb. 2 Regions of interest (ROI) zur Messung des Kontrast-RauschVerhältnisses.

Eabs MAD = m PMMA where m_PMMA denotes the irradiated mass.


b

Fig. 3 Specific X-ray tube radiation output at 1 m distance. Circles and crosses indicate measured values; lines indicate results from the calibrated spectrum simulation.

(1)

The MAD is independent of the irradiated area since the absorbed energy and the irradiated mass are both proportional to the exposed area. The MAD is a valid measure for patient risk only if the exposed area is kept constant.

Abb. 3 Spezifische Strahlenausbeute in 1 m Abstand. Kreise und Kreuze bezeichnen die gemessenen Werte, die Linien zeigen die Ergebnisse der Spektrensimulation.

CNR determination The contrast-to-noise ratio (CNR) was used as the indicator of image quality. The CNR was derived from the pixel values in the original (not post-processed) X-ray image of the phantom. Two 100 × 100 pixel-sized regions of interest (ROI) were placed in the image, one at the projected aluminum contrast object and one " Fig. 2). beside it (● The CNR was then determined as CNR =

√

PV1 – PV2 1 (SD 1 2 + SD 2 2 ) 2

(2)

Where PV1 and PV2 are the mean pixel values and SD1 and SD2 are the standard deviations of pixel values in ROI1 and ROI2, respectively. A background trend subtraction using a first order two-dimensional polynomial fit was applied to the image signal in ROI1 and ROI2 before evaluating the standard deviations in order to minimize low frequency intensity variations like the heel effect or the inverse square law effect. The four CNR/MAD pairs for each combination of tube voltage and filtration were fitted by a second-order polynomial on a loglog scale. This made it possible to derive the CNR value for any arbitrary MAD level in the range of 5 to 50 μGy by interpolation.

Fig. 4 Mean absorbed dose in the phantom (1 cm PMMA) per X-ray tube loading (mAs), scaled to 1 m focus-object distance. Abb. 4 Mittlere absorbierte Dosis im Phantom (1 cm PMMA) pro StromZeit-Produkt (mAs), normiert auf 1 m Fokus-Objekt-Abstand.

● Fig. 4

shows the tube voltage dependence of the phantom MAD per 1 mAs tube loading for a focus-object distance of 1 m as derived from the calibrated spectrum simulation. The shown data allow determination of the MAD for any arbitrary situation of tube voltage and pre-filter from the applied mAs value. The relationship between phantom entrance dose and MAD is " Fig. 5. The ratio of entrance dose to MAD increases shown in ● slightly with decreasing tube voltage, meaning that the entrance dose will be higher at a lower tube voltage for a fixed MAD. The additional pre-filter of 0.1 mm Cu/1 mm Al reduces the entrance dose per MAD slightly by 3 – 7 %, depending on the tube voltage. Keeping the MAD fixed while changing the radiation beam quality leads to a varying detector dose level and to changes in the de" Fig. 6 shows the measured mean pixel value per tector signal. ● MAD for 1 cm PMMA as a function of tube voltage with and with"

Results !

Spectrum simulation and dose determination

● Fig. 3 shows the specific X-ray tube output, scaled to a focal "

spot distance of 1 m, as measured values (points) in comparison with the spectral simulation (lines), both with and without a pre-filter. For the simulated spectra without a pre-filter, an (inverse) filter of –3 μm tungsten and a scale parameter c = 0.89 were applied to calibrate the simulation results to the measured values. For the simulation of the spectra with a pre-filter, the nominal copper filter thickness of 0.1 mm was reduced by 30 μm and a scale parameter of c = 0.8 was applied. Using these parameters, the results for the simulated X-ray spectra deviated by less than 2 % from the measured air kerma values.


Fig. 7 Measured contrast-to-noise ratio as a function of the mean absorbed dose (MAD) in the phantom with X-ray tube voltage as parameter and with no added filter. The lines show the polynomial fit to the measured points.

Fig. 5 Ratio of entrance dose and MAD for 1 cm PMMA as a function of tube voltage. Abb. 5 Verhältnis der Eintrittsdosis zur mittleren absorbierten Dosis für 1 cm PMMA als Funktion der Röhrenspannung.

Abb. 7 Gemessenes Kontrast-Rausch-Verhältnis (CNR) als Funktion der mittleren absorbierten Dosis (MAD) im Phantom für verschiedene Röhrenspannungen ohne Vorfilter. Die Linien zeigen die Polynom-Fits zu den Messungen.

Fig. 6 Mean pixel value per mean absorbed dose (MAD) as a function of tube voltage, with and without a pre-filter. Abb. 6 Mittlerer Pixelwert pro mittlerer absorbierter Dosis (MAD) als Funktion der Röhrenspannung mit und ohne Vorfilter (0,1 mm Cu/1 mm Al).

out an additional pre-filter. For equal MAD, the pixel values for the situations 60 kV/with a pre-filter and 40 kV/without a pre-filter differ by a factor of approximately three. This difference in signal level is covered by the dynamic range of the digital detector and needs to be compensated for by appropriate processing (windowing) when displaying the images on a reading workstation.

Influence of tube voltage and filtration on CNR and dose

● Table 1, 2 list the tube voltage, the tube current-time-product, "

the exposure time, the phantom mean absorbed dose, as well as the CNR for all measurements with and without the added prefilter of 1 mm Al/0.1 mm Cu, respectively. These data were used " Fig. 7 – 10. as the basis for the graphical displays in ● ●" Fig. 7, 8 show the measured CNR versus the MAD on a log-log scale for the five tube voltages investigated with and without added filtration, respectively. For each configuration the CNR in-

Fig. 8 Measured contrast-to-noise ratio as a function of the mean absorbed dose (MAD) in the phantom with X-ray tube voltage as parameter and with an added pre-filter of 0.1 mm Cu/1 mm Al. The lines show the polynomial fit to the measured points. Abb. 8 Gemessenes Kontrast-Rausch-Verhältnis (CNR) als Funktion der mittleren absorbierten Dosis (MAD) im Phantom für verschiedene Röhrenspannungen mit 0,1 mm Cu/1 mm Al Vorfilter. Die Linien zeigen die Polynom-Fits zu den Messungen.

creases with increasing MAD as is expected for a system mainly limited by quantum noise. However, while the CNR is expected to be proportional to the square root of the image receptor dose (which in turn is directly proportional to the MAD for a fixed radiation quality) for a purely quantum noise limited system, the CNR is found to deviate from this behavior, indicating that other noise sources may also be present. Considering the results for different tube voltages for a given, fixed MAD, it can be seen that the CNR is generally higher at low" Fig. 9 shows this for a fixed MAD = 25 μGy for er tube voltages. ●


647



648


Discussion !

Fig. 9 Contrast-to-noise ratio (CNR) as function of tube voltage for a fixed mean absorbed dose (MAD) in the phantom of 25 µGy.


b

Abb. 9 Kontrast-Rausch-Verhältnis (CNR) als Funktion der Röhrenspannung bei konstanter mittlerer Dosis (MAD) im Phantom von 25 µGy.

Fig. 10 Mean absorbed dose (MAD) as a function of the tube voltage for a fixed contrast-to-noise ratio of CNR = 26. Abb. 10 Mittlere absorbierte Dosis (MAD) als Funktion der Röhrenspannung bei konstantem Kontrast-Rausch-Verhältnis von CNR = 26.

situations with and without the pre-filter. For any given tube voltage, the CNR at a fixed MAD is higher without than with the prefilter, the difference being 11 % at 66 kV and 9 % at 40 kV. Another view on the data is to keep the image quality (i. e., the CNR) fixed and to investigate the potential to reduce the patient " Fig. 10 shows the MAD versus tube voldose in terms of MAD. ● tage for a fixed CNR of 26. The MAD decreases with decreasing tube voltage and reaches a (relative) minimum at the lowest investigated tube voltage of 40 kV. For example, reducing the tube voltage from 50 kV or 60 kV (the lower and upper endpoints of the range suggested in [20] for distal extremities) to 40 kV leads to MAD reductions of 24 % and 50 %, respectively. For all tube voltages the MAD is lower without the additional pre-filter. The guidelines do not suggest the use of any pre-filtration for the radiography of distal extremities [20].

This study shows that for a phantom representing the situation of a distal pediatric extremity the CNR is greatest at the lowest investigated tube voltage (40 kV) and without an added pre-filter when the mean absorbed dose is kept fixed, i. e., at an equal radiation risk. Conversely, for a given fixed CNR, the patient dose in terms of MAD (and entrance dose) is also lowest at this radiation quality. This is contrary to the widespread belief that the beam quality should be hardened by increasing the tube voltage and adding filtration to reduce the dose. The reason for this apparent discrepancy can be found in the different boundary conditions for digital imaging compared to conventional (screen-film) imaging, on which most of commonly accepted practice rules are still founded. Conventional X-ray imaging is characterized by the given sensitivity of the screenfilm image receptor used. To reach a proper film density, a certain image receptor dose is necessary, e. g., approximately 2.5 μGy for a 400-speed screen-film system. Optimizing the imaging condition then means reducing the patient dose while maintaining the required receptor dose. This can only be achieved by using radiation with a higher penetration that is less absorbed in the patient. Typically, the choice is to increase the tube voltage and to insert an additional filter, e. g., 0.1 or 0.2 mm copper. This approach has been codified for most examinations (though not for the distal extremities) in national and international guidelines for pediatric radiographic imaging [20, 21]. However, the harder beam quality also leads to less contrast in the image and, consequently, to a reduced CNR, meaning that dose reduction in conventional imaging is intrinsically associated with image quality reduction. In contrast, digital detectors have a wide dynamic range that allows the acquisition of images at different detector dose levels. The resulting differences in image pixel value can be compensated for by appropriate image processing (sometimes called “ranging” or “exposure recognition”), which ensures that the displayed image has the correct density or brightness levels. It is therefore possible to accept a lower detector dose with an intrinsically higher noise level at a lower tube voltage if the increased image contrast for this condition over-compensates for the increase of noise. In this case the CNR will increase. In this study the CNR values increased towards lower tube voltages, but did not reach an absolute maximum. Other studies using a similar approach for X-ray examinations of thicker body parts showed a maximum CNR/dose performance in the tube voltage range of approximately 50 to 55 kV [12] or 50 to 60 kV [14], depending on the examination. It is expected that the CNR maximum shifts towards lower tube voltages for thinner objects. For the very thin objects studied here, a maximum CNR is expected for tube voltages below 40 kV. However, these voltages are usually not accessible with a general purpose radiography X-ray generator. A pre-filter absorbs more low-energy photons than high-energy photons. Hence, the mean photon energy of the beam increases, i. e., the radiation gets harder. Therefore, a pre-filter has a similar effect as a higher tube voltage. In the case investigated here the absence of a pre-filter with the tube-current-time-product (mAs) adjusted for equal MAD increases the contrast-to-noise ratio. For thicker objects it was found that an aluminum and/or copper filter will improve the image quality for a given patient risk [12, 14]. This is true if the system is operated at or below the



Conclusion !

The results of this phantom study suggest using a tube voltage not higher than 40 kV and avoiding additional beam filtration for digital radiography of distal pediatric extremities. This is partly in contradiction to widely accepted clinical guidelines, which have been developed for conventional screen-film imaging with its limitation of a given film sensitivity. Digital detectors with their wide dynamic range provide an extra degree of freedom for optimization which has not been fully exploited to date. Further clinical evaluation of the suggested parameters is necessary before the changes can be adopted for general clinical use and eventually be incorporated into guidelines for pediatric digital radiography.

References 01 Stephan G, Schneider K, Panzer W et al. Enhanced yield of chromosome aberrations after CT examinations in paediatric patients. Int J Rad Biol 2007; 83: 281 – 287 02 ICRP Publication 73: Radiological protection and safety in medicine. Ann ICRP 1996; 26: 03 ICRP Publication 103: The 2007 Recommendations of the International Commission on Radiological Protection. Ann ICRP 2007; 37: 04 Nicholson RA, Thornton A, Akpan M. Radiation dose reduction in paediatric fluoroscopy using added filtration. Br J Radiol 1995; 68: 296 – 300 05 Mooney R, Thomas PS. Dose reduction in a paediatric x-ray department following optimization of radiographic technique. Br J Radiol 1998; 71: 852 – 860 06 Hansson B, Finnbogason T, Schuwert P et al. Added copper filtration in digital paediatric double-contrast colon examinations: effects on radiation dose and image quality. Eur Radiol 1997; 7: 1117 – 1122 07 Seifert H, Jesberger HJ, Schneider G et al. Dose reduction in thorax radiography in simulated neonates with additional filtration and digital luminescence radiography. Acta Radiol 1998; 39: 514 – 519 08 Fenner JW, Morrison GD, Kerry J et al. A practical demonstration of improved technique factors in paediatric fluoroscopy. Br J Radiol 2002; 75: 596 – 602 09 Nickoloff EL, Berdon WE, Lu ZF et al. Pediatric high KV/filtered airway radiographs: comparison of CR and film-screen systems. Pediatr Radiol 2002; 32: 476 – 484 10 Onnasch DGW, Schemm A, Kramer HH. Optimization of radiographic parameters for paediatric cardiac angiography. Br J Radiol 2004; 77: 479 – 487 11 Wiltz HJ, Petersen U, Axelsson B. Reduction of absorbed dose in storage phosphor urography by significant lowering of tube voltage and adjustment of image display parameters. Acta Radiol 2005; 46: 391 – 395 12 Gislason AJ, Davies AG, Cowen AR. Dose optimization in pediatric cardiac x-ray imaging. Med Phys 2010; 37: 5258 – 5269 13 Brosi P, Stuessi A, Verdun FR et al. Copper filtration in pediatric digital X-ray imaging: its impact on image quality and dose. Radiol Phys Technol 2011; 4: 148 – 155 14 Tapiovaara MJ, Sandborg M, Dance DR. A search for improved technique factors in paediatric fluoroscopy. Phys Med Biol 1999; 44: 537 – 559 15 Tingberg A, Sjöström D. Search for optimal tube voltage for image plate radiography. Proc SPIE 2003; 5034: 187 – 196 16 Samei E, Dobbins JT 3rd, Lo JY et al. A framework for optimising the radiographic technique in digital x-ray imaging. Radiat Prot Dosimetry 2005; 114: 220 – 229 17 Doyle P, Martin CJ, Gentle D. Application of contrast-to-noise ratio in optimizing beam quality for digital chest radiography: comparison of experimental measurements and theoretical simulations. Phys Med Biol 2006; 51: 2953 – 2970 18 Geijer H, Norrman E, Persliden J. Optimizing the tube potential for lumbar spine radiography with a flat-panel detector. Br J Radiol 2009; 82: 62 – 68 19 Birch R, Marshall M. Computation of Bremsstrahlung x-ray spectra and comparison with spectra measured with a Ge(Li) detector. Phys Med Biol 1979; 24: 505 – 517 20 Leitlinie der Bundesärztekammer zur Qualitätssicherung in der Röntgendiagnostik – Qualitätskriterien röntgendiagnostischer Untersuchungen. Gemäß Beschluss des Vorstandes der Bundesärztekammer vom 23. November 2007. http://www.bundesaerztekammer.de/downloads/Leit Roentgen2008Korr2.pdf (abgerufen am 15.9.2011) 21 European Guidelines on quality criteria for diagnostic radiographic images in pediatrics – EUR 16261, Office for official publications of the European Communities. Luxembourg: 1996 22 Granfors PR, Aufrichtig R. Performance of a 41×41 cm² amorphous silicon flat panel x-ray detector designed for angiographic and R&F imaging applications. Med Phys 2003; 30: 2715 – 2726 23 Huda W. Assessment of the problem: Pediatric doses in screen-film and digital radiography. Pediatr Radiol 2004; 34: 173 – S182 24 Tapiovaara M, Lakkisto M, Servomaa A. PCXMC: A PC-based Monte Carlo program for calculating patient doses in medical x-ray examinations. Report STUK-A139. Helsinki: Finnish Centre for Radiation and Nuclear Safety 1997, http://www.stuk.fi/sateilyn_kaytto/ohjel mat/PCXMC/en_GB/pcxmc/



tube voltage for maximal CNR/dose performance. An added prefilter shifts and narrows the X-ray spectrum towards the optimal photon energies. However, since the optimal tube voltage for the phantom study performed here is below 40 kV and a pre-filter further hardens the X-ray beam, the pre-filter leads to a degradation of the CNR. The study reported here has a number of limitations. The PMMAAl combination used as the phantom is only a very rough approximation of the clinical imaging situation. The intention was to model the contrast-generating properties of bone and soft tissue, for which these materials are a common choice. The phantom does not contain fine structure elements for testing the spatial resolution properties of the imaging setup. However, the modulation transfer function (MTF) of the type of flat detector (CsI/a-Si) used in this study is known to be rather weakly dependent on beam quality [22]. It is therefore expected that the system spatial resolution is not appreciably changed by the changes in tube voltage. Lowering the tube voltage at a fixed MAD will increase the expo" Table 1, 2). This is due to the lower specific radiasure times (● tion output of the X-ray tube and a lower maximum tube current at lower voltages. Both effects lead to a prolongation of the exposure time. Some of this may be regained by applying a lower dose at equal CNR as shown in this study. For the transfer of the study results to clinical application, it will be necessary to investigate which lowest tube voltage is possible for a given system configuration in order not to exceed clinically acceptable exposure times. In this study the mean absorbed dose (MAD) was used to quantify the radiation risk to the patient. Another widely used quantity for this purpose is the effective dose [23]. The MAD and effective dose are influenced in the same manner by changes in beam quality, so both quantities can be used similarly for optimization. However, the determination of the effective dose for pediatric studies is non-trivial and there is currently no appropriate model for extremity examinations. The standard tool for effective dose calculation therefore does not include pediatric extremities as target organs [24]. The MAD is difficult to determine in clinical practice, but the easily measurable entrance dose can be used as a surrogate for the " Fig. 5, the ratio of thin objects considered here. As shown in ● both dose quantities remains rather constant over a wide range of tube voltages. Therefore, keeping the entrance dose constant while reducing the tube voltage will result in almost the same CNR gain as for a constant MAD. In this case the MAD will be slightly reduced.

649