British Journal of Radiology 74 (2001),938-948 © 2001 The British Institute of Radiology
A computer model of an image intensifier system working under automatic brightness control
A J Reilly, MSc
and
D G Sutton, PhD
Department of Medical Physics, Ninewells Hospital and Medical School, Dundee DD1 9SY, UK
Correspondence: Dr D G Sutton
 |
Abstract
|
|---|
A computer model of a fluoroscopy unit operating under automatic brightness control has been developed. The model has been validated by simulating one particular unit but is very general in nature and can easily be applied to other fluoroscopy systems. The model was developed by breaking the operation of a fluoroscopy unit down into its constituent parts then implementing each part in a module of code. It is controlled using the input air kerma to the intensifier face. Discrepancies from the situation when the model is controlled by the energy deposited in the input phosphor have been investigated and shown to be negligible over the operating range. To calculate entrance surface dose rates (ESDRs) to water and polymethylmethacrylate (PMMA) phantoms, Monte Carlo techniques were used to generate backscatter factors for these materials using the beam geometry and range of possible tube potentials and field sizes of a typical mobile image intensifier unit. The model was validated by calculating ESDRs to different thicknesses of water and PMMA phantoms. The predictions generated by the model were in good agreement with experimental measurements. Potential uses of the model include evaluation of dose reduction techniques, investigation of the balance between patient dose and image quality, and assessment of scatter dose.
|
Introduction
|
|---|
The publication of Patient Dose Reduction in Diagnostic Radiology by the National Radiological Protection Board in 1990 [1] resulted in increased attention being paid to the general topic of dose reduction, dose measurement, dose audit and so on. The document provided evidence of the need for patient dose reduction. It also assessed priorities from amongst the many methods available for reducing medical exposures and made specific recommendations regarding the more effective methods to encourage their wide-spread adoption. One of the central recommendations of the document was that any dose saving strategy should be assessed in terms of its cost effectiveness (£ per manSv saved). In that context, changes to procedures and techniques should take precedence over changes to equipment since they are associated with very little, if any, cost. Furthermore, the dose savings associated with any particular strategy should be quantifiable [2].
Implicit in any patient dose reduction programme is the assumption that diagnostic accuracy will not be sacrificed. There is a need to ensure that any investigation yields diagnostically useful results whilst subjecting the patient to minimum risk [3]. However, it is not always easy to anticipate the impact of a strategy to reduce patient dose or its effect on image quality. For example, it might be assumed that the use of pulsed techniques in fluoroscopy may result in reduced radiation dose at the expense of image quality, but this is not necessarily the case [3, 4]. Similarly, the impact on image quality of such measures as introducing copper filtration has not been fully explored [5–7].
In fluoroscopy, investigation of the balance between patient dose and image quality, and assessment of the impact of dose reduction techniques, are both important and complex. Part of this complexity is introduced by the fact that the majority of fluoroscopic equipment operates under automatic brightness control (ABC). To obtain a good quality image on the TV screen, the level of the video signal output by the image intensifier should be within a particular range. The signal level corresponds directly to the final light output of the intensifier, which is a function of the response of the input phosphor. ABC attempts to maintain the signal level within the appropriate range by monitoring the light output resulting from a nominal area of the image intensifier face (normally a region about the centre) and adjusting the tube potential (kVp) and tube current (mA) of the unit according to a predefined algorithm. The effect of ABC is that in common diagnostic situations the air kerma rate at the image intensifier face is kept relatively constant [8].
Different manufacturers use different algorithms to control ABC. Some ABC algorithms increase the tube current rather than the tube potential with the aim of improving image quality by reducing the effect of Poisson noise, whilst others try to minimize patient dose by increasing the tube potential at the expense of the tube current. ABC curves may be specifically designed for use with contrast media, with the tube potential being rapidly increased to just beyond the K-edge then kept constant whilst the tube current is varied [9]. Most units also have a variety of intensifier input air kerma rate options, so that differing sizes of patient can be accommodated. As a result, it is not at all apparent what the impact of differing patient dose strategies will be even on a standard size patient. It is even more complex to determine the impact on image quality of any such patient dose reduction strategy.
Methods for measuring image quality are described in the International Commission on Radiation Units and Measurements (ICRU) Report 54 [10] and by Martin et al [11]. Experimental assessment is both time consuming and difficult. Development of a simulated environment in which the balance between patient dose and image quality could be studied would have significant advantages. The aim of the work reported here was the development of the first stage of that environment — a computer model of an image intensifier TV system operating under ABC. A variety of ABC algorithms can be employed in the model and an assortment of dose reduction measures can be simulated. The dose reduction factor R discussed by Sutton and Cranley [2] can then be derived. Consequently, the cost effectiveness of any dose reduction technique can be assessed [2]. It will also be possible, with the incorporation of Monte Carlo techniques, to use the results from the model in investigations of the balance between patient dose and image quality and to assess the distribution and quality of radiation scattered from a patient under real conditions. Although the model is completely general in nature, so that it can be applied to any fluoroscopy system operating under ABC, it has been programmed to simulate a Siemens Siremobil 2000 unit (Erlangen, Germany) for the purposes of validation. Validation has been performed by calculating predicted entrance surface dose rates to water and polymethylmethacrylate (PMMA) phantoms and comparing the results with those of direct measurement.
|
Model description
|
|---|
The geometry modelled was a C-arm image intensifier system with a fixed focus-to-intensifier distance of 90 cm. The ABC aspects of the model were based on the Siemens Siremobil 2000, although it would be a simple matter to incorporate any other ABC algorithm. On the units modelled, the user can choose between any of three input dose rates to the intensifier face (0.13 µGy s-1, 0.22 µGy s-1 and 0.44 µGy s-1) and five ABC curves: two anti-isowatt curves; two high contrast curves; and one low dose curve. When using the anti-isowatt curves, tube current is proportional to tube potential, whilst the high contrast curves aim to optimize image contrast by initially ramping up the tube current. The low dose curve is intended for use with small patients, e.g. children. The simulated unit had a constant potential generator, a 9° tungsten target, 3 mm aluminium (Al) total filtration, a circular field of view of diameter 17 cm and a grid factor of 8 (40 strips cm-1). Operation of the unit was broken down into a number of modular parts. The best way to describe the operation of the model as a whole is by illustrative example. If one considers the geometry shown in Figure 1
, which simulates a simple measurement of entrance surface dose rate (ESDR), then the way in which the simulation works is explained by the flow diagram in Figure 2. Each component of the model was simulated by a distinct software module, as will be described briefly below.
View larger version (16K):
[in this window]
[in a new window]
|
Figure 1. Geometry used for a simple example of the functionality of the model. The entrance surface dose rate is calculated for different depths of water.
|
|
View larger version (38K):
[in this window]
[in a new window]
|
Figure 2. Flowchart outlining the procedure the model would follow to calculate dose rates in an ion chamber of geometry shown in Figure 1 . ABC, automatic brightness control.
|
|
Software platform
The Java programming environment was chosen as the software platform to implement the model. Choosing such an object-orientated language enabled powerful data structures to be constructed to store and process the data required. The X-ray spectra used in the simulation were generated by incorporating a modified version of the Institute of Physics and Engineering in Medicine (IPEM) Report 78 software [12] in the code. Additional modules were written to interpret and process the data types produced by the processor.
X-ray spectra
X-ray spectra were generated using the techniques of Birch and Marshall as reported in IPEM Report 78 [13]. This report generates spectra at 1.0 keV intervals from 1 keV to the applied accelerating potential, and the data are specified at 750 mm from the target. For the purposes of this simulation these spectra were interpolated to 0.1 kVp intervals by considering the spectra at two successive tube potential (kVp) valves and finding the linearly weighted average between the data reported at each energy value. For example, the spectrum at 95.3 kVp was found by multiplying the95 kVp spectrum by 0.7 and the 96 kVp spectrum by 0.3, then summing the results.
Interpolation of IPEM Report 78 spectra was necessary to allow small steps along the ABC curves, enabling finer control of air kerma rates. The 0.1 kVp step size was found to provide sufficiently smooth transitions between ABC points (larger step sizes resulted in widely varying output doses) and had a negligible effect on processor overhead owing to the binary chop algorithm described later. In practice, fluoroscopy units do not afford such precise control over kerma rate but rather bring the image intensifier video signal to the optimum level by adjusting the signal gain once the signal lies within a pre-set range (C Lawinski, KCARE, 1999, personal communication). Since the modelling of gain controls is a complicated process, approximation of extremely fine air kerma control was used instead and should produce similar results.
The X-ray beam was assumed to evolve conically from the target, with the diameter of the field of view at the image intensifier being equal to the diameter of the intensifier. This represents the maximum field size of the set, although the field of view can readily be reduced, to simulate collimation or electronic magnification, by varying parameters within the model. Since the beam originates from a point source, fluence rate is inversely proportional to the square of the distance from the target. The ith value in the fluence rate spectrum at any distance from the tube was therefore determined by
where 
i–R78 is the interpolated IPEM Report 78 spectrum and r is the distance (mm) from the tube.
Efficiency
IPEM Report 78 presents X-ray spectra for ideal X-ray tubes. However, real tubes are not 100% efficient and produce spectra that have a lower output than might be expected from the report. For an X-ray tube of efficiency 
, the fluence rate spectrum is given by
where can be determined by comparing the actual output in air with that predicted for a theoretical tube with the same operating parameters. The efficiency of the tube used in the validation experiments was measured as 79±2%.
Filtration and attenuation
The majority of attenuators in the X-ray beam, such as Al filtration and patient analogue, were accounted for using the linear attenuation coefficients of Hubbell [14]. The fluence rate spectrum of the beam after attenuation by x mm of any material was calculated using
where i–before and i–after are the spectra before and after attenuation, respectively and µi is the linear attenuation coefficient obtained either from IPEM Report 78 or by interpolation of Hubbell's data.
Dose rate measurement
The mass-energy transfer coefficient µtr/
and the mass-energy absorption coefficient µen/ are related by [14, 15]
where g is the fraction of energy transferred to secondary charged particles, which is lost as radiative (photon) emissions through the slowing down of the particles in the medium. In materials of low atomic number such as air, soft tissue, water and PMMA, g is very small for photons at diagnostic energies, so that absorbed dose is a very good approximation of kerma [15]. For the purpose of the simulation, kerma rate was therefore calculated using
where Ei is the energy of the ith point in the photon fluence rate spectrum. Either linear energy absorption coefficients µen-i were interpolated from those presented in the National Institute of Standards and Technology Physical Reference database [14], or photon fluence to kerma conversion factors (=µen–i x Ei) were taken from IPEM Report 78 [13].
Image intensifier
ABC works by monitoring either the luminance or the video signal arising from a region of interest on the image intensifier output phosphor, and attempting to maintain this at a constant level. Since both output luminance and video signal are proportional to the rate at which energy is deposited in the caesium iodide (CsI) input phosphor, ABC can be considered as a method to keep the energy deposition rate in the input phosphor constant [8, 16]. In practice, however, the operation of an ABC unit is specified by the input air kerma at the image intensifier face, which is a measurable as opposed to a calculated quantity.
As shown in Figure 3, the K absorption properties of CsI are observed between 33 keV and 36 keV (the K-edge for I is 33.2 keV and that for Cs is 36 keV [14]) whereas air has no significant K-edge. This means that the ratio between kerma in air and energy deposited in CsI will vary depending upon the mean energy of the fluence spectrum, with the CsI being relatively more sensitive to higher energy photons. However, it has been reported that hardening the beam by about 1 mm of copper (Cu) sufficiently removes photons below the K-edge so that this ratio becomes substantially constant, and employing ABC under these conditions results in a constant air kerma rate at the face of the intensifier [17].
View larger version (10K):
[in this window]
[in a new window]
|
Figure 3. Comparison between the mass-energy absorption coefficients of air and caesium iodide (CsI). It can be seen that the K absorption properties of CsI (—) are observed between 33 keV and 36 keV whilst air (– – –) has no significant K-edge. (Data from [14].)
|
|
Using IPEM Report 78 [13] it was found that 10 cm of water or PMMA provides similar attenuation to 1 mm Cu and significantly reduces the relative fluence of photons below the K-edge. It is therefore reasonable to assume that, for water or PMMA phantoms thicker than 10 cm, the ABC system is governed by air kerma rate, as described above. In typical clinical situations, attenuation due to the patient will be sufficient to ensure that this is the case, hence it was decided to model the image intensifier in terms of air kerma rates at its entrance surface. An investigation undertaken to verify the validity of this approach showed it to be acceptable and is presented in the Appendix.
Grid
The Siremobil 2000 comprises a Lysholm anti-scatter grid (Gridline AB, Smidesvagen, Sweden) with a focal length of 100 cm. Since the Siremobil 2000 focus-to-grid distance is fixed at 90 cm, lying within the operational limit of the grid, transmission of the grid can be considered uniform across its surface [18]. The grid is composed of vertical lead strips separated by paper (M Lothert, Siemens Erlangen, 1999, personal communication) and was specified as having 40 strips cm-1 and a ratio of r=8. The interspace thickness D was 0.178 mm and the width of the lead strips d was 0.072 mm. Using r=h/D, the height h of a layer was calculated as being 1.424 mm.
The transmission T of primary radiation through the grid at any given accelerating potential is given by[19]
Thus, using the values above, T=71.2%.
The manufacturer specifies a dose reduction factor for the grid that is equal to the ratio between the air kerma at the surface of the grid and that at the input face of the image intensifier. This factor is therefore equal to the reciprocal of T and the quoted value of 1.4 agrees well with the 71.2% transmission calculated above.
Automatic brightness control
Points on the five Siremobil 2000 ABC curves were obtained by stepping through each of the available curves on a unit and recording the tube potential/tube current points permitted. It was found that the experimentally obtained curves (Figure 4) on individual units differed slightly from those specified by the manufacturer. Equations were fitted to each ofthe experimentally obtained curves and tube potential/tube current look-up tables were generated at 0.1 kVp intervals. Since the modelling equations have no direct physical significance they have not been presented here.
ABC is modelled by attempting to maintain a constant air kerma rate at the image intensifier face. Nominally, this rate can be selected by the user to be 0.13 µGy s-1, 0.22 µGy s-1 or 0.44 µGy s-1. The easiest way in which to determine the ABC operating point for any attenuator is to calculate the air kerma rate to the image intensifier face at all the points on the chosen ABC curve, then choose the point that best matches the specified rate. Such a technique is extremely robust, since it works for all ABC curves, but it is also intensely time consuming, since a calculation is required at every 0.1 kVp interval. However, when the air kerma rate increases monotonically, a more efficient, iterative binary chop algorithm can be employed. Since this is true for all the Siremobil 2000 curves, this algorithm was adopted by default and is illustrated in Figure 5. In this figure, the x-axis represents successive points along the entire ABC curve data set whilst the y-axis shows the image intensifier air kerma rate for any tube potential/tube current combination, given an arbitrary assortment of attenuators in the X-ray beam. The spot represents the tube potential/tube current point that results in the air kerma rate being closest to the desired value. The algorithm begins by selecting the point exactly halfway through the entire curve data set and then calculates the input air kerma rate at either extreme of the data set. Depending on which of these kerma rates is closer to that desired, it chooses one-half of the data set and finds a new midpoint. It then calculates the input air kerma rates to the image intensifier at either end of this new data set and splits the data in half again. The algorithm keeps halving data sets in this way until the operating point that produces an image intensifier input closest to that desired is obtained.
View larger version (18K):
[in this window]
[in a new window]
|
Figure 5. Illustration of the binary chop algorithm finding the best operating point for an arbitrary assortment of attenuators in the beam. The x-axis represents successive points along the entire automatic brightness control (ABC) curve data set whilst the y-axis shows the image intensifier air kerma rate for the tube potential/tube current combination of the ith point on the curve. Each arrow shows the operating point chosen by successive iterations of the algorithm and the spot represents the tube potential/tube current point that results in the air kerma rate being closest to the desired value.
|
|
Backscatter
When calculating phantom ESDRs, it is necessary to consider radiation scattered back from within the phantom. Although Petoussi-Henss et al [20] give backscatter factors for rectangular fields, no such factors have been published for circular fields such as those modelled here. Factors were therefore generated using the MCNP transport code [21] for the tube potential and field size combinations that can be encountered in the Siremobil 2000. The X-ray spectra used in the Monte Carlo calculations were again obtained using IPEM Report 78 [12, 13] and the calculations were performed using 11 Sun Solaris (UNIX) workstations linked via parallel virtual machine technology [22].
Each phantom was modelled as a solid cylinder of radius 50 cm and variable height. The focus-to-surface distance was set to 100 cm with the X-ray beam incident on one of the flat surfaces of the cylinder, as shown in Figure 6. An ion chamber of radius 1 cm (capacity 3.1 cm3) was modelled on the surface of the phantom in the centre of the X-ray beam and the absorbed dose rate in the ion chamber was calculated using an MCNP energy deposition tally. The phantom was then removed with the ion chamber being kept in position and the dose rate calculated again. The backscatter factor was equal to the ratio between the dose rates with and without the phantom present. In this simulation, doses should ideally have been calculated at a point on the surface of the phantom. Although it would have been possible to do this in MCNP, it was decided instead to model a finite ion chamber to increase the efficiency of the calculations. Simulations were run using different sizes of ion chamber and it was found that reducing the radius below 1 cm produced negligible change in the calculated dose rates. This size of chamber was therefore chosen for use in the simulations.
Backscatter factors were calculated at an accelerating potential of 110 kVp and a field of diameter 20 cm for water and PMMA phantoms of thickness 5–35 cm, in 5 cm intervals. The results, plotted in Figure 7, show that backscatter is independent of phantom thickness for phantoms thicker than 10 cm. Since a typical patient cross-section exceeds this value, backscatter was taken as independent of phantom thickness in the model.
Using a constant phantom thickness of 40 cm, backscatter factors were calculated at field diameters of 5 cm, 10 cm, 15 cm, 20 cm and 25 cm for accelerating potentials of 40–110 kVp, in 10 kVp intervals. All factors were calculated to better than 2% accuracy, as quoted by MCNP, and the total execution time was approximately 48 h. The results are presented in Tables 1 and 2. The backscatter factors generated agree with the rectangular field data of Petoussi-Henss et al [20] to within 5% when equivalent diameter [23] is taken into consideration. (It should be noted that the data of Petoussi-Henss et al [20] would appear to have been generated using a 14° target angle.)
View this table:
[in this window]
[in a new window]
|
Table 1. Backscatter factors for water phantoms of depth greater than 10 cm at various field diameters, calculated using MCNP to simulate the geometry in Figure 6 (9° target angle, 3 mm Al total filtration)
|
|
View this table:
[in this window]
[in a new window]
|
Table 2. Polymethylmethacrylate backscatter factors generated using MCNP to simulate the geometry in Figure 6, for phantoms thicker than 10 cm at various field diameters (9° target angle, 3 mm Al total filtration)
|
|
Equations of surfaces were fitted to the results of the simulations and these were used to calculate the appropriate backscatter factor for any combination of tube potential and field diameter. Since these equations are two-dimensional polynomials with no physical significance they have not been reproduced here.
|
Model validation
|
|---|
To determine whether the model produces realistic results, the arrangement demonstrated in Figure 1 was simulated. The ABC function was set to provide the image intensifier with a nominal input dose rate of 0.13 µGy s-1 using the 5 mA anti-isowatt curve. A 15 cm3 ion chamber and Keithley 35050A dosemeter (Keithley Instruments Inc, Cleveland, OH) were used to measure ESDRs. The actual input dose rate was measured as 0.16 µGy s-1, and this value was applied to the model. Grid transmission and tube efficiency were taken as 71.2% and 79%, respectively, as described above, and the equivalence of the table was measured as 0.7 mm Al.
With the distance from the phantom exit surface to the image intensifier set to 20 cm, ESDRs were measured for PMMA phantoms of thickness 10 cm, 15 cm, 20 cm, 25.5 cm, 29 cm and 35 cm. Water phantoms were constructed using sealed plastic tubs; three of depth 9.8 cm and one of depth 4.0 cm. The distance from the phantom exit surface to the image intensifier was increased to 30 cm and ESDRs were measured to water phantoms of depths 9.8 cm, 13.8 cm, 19.6 cm, 23.6 cm, 29.4 cm and 33.4 cm. Each of these simulations took less than 5 s to run to completion on a Pentium II 266 MHz PC.
|
Results and discussion
|
|---|
For each experiment, ESDRs were calculated by the model over the whole range of phantom thicknesses and these are plotted superimposed upon the experimental values in Figures 8a,b. The error bars represent the estimated uncertainty in the ESDR measurements (±5% for PMMA phantoms, ±10% for water phantoms). It is evident from these figures that the predictions of the computer model agree with the experimental results to within the limits of experimental uncertainty. The model can therefore be considered validated for water and PMMA phantoms.
View larger version (14K):
[in this window]
[in a new window]
|
Figure 8. Results of validation experiments, using (a) PMMA and (b) water phantoms. Points indicate experimental measurements and lines represent predictions of model.
|
|
For each thickness of phantom, the air kerma rate actually achieved at the input surface of the intensifier was monitored. The use of air kerma to control the unit is explored in Appendix A. It is clear from the results plotted in Figure 9 that for each material there is an optimal thickness range for employing ABC. These are 5–33 cm for PMMA and 5–37 cm for water, and it is only over these ranges that ABC functions correctly. If the phantom is too thin the unit is unable to select a sufficiently low operating point to reduce the dose rate at the image intensifier to the desired level. Conversely, if the phantom is too thick the unit cannot go to a high enough operating point to overcome the reduction in dose rate owing to the increased attenuation of the beam.
View larger version (12K):
[in this window]
[in a new window]
|
Figure 9. Calculated air kerma rate at surface of image intensifier for different thicknesses of PMMA (....) and water (—) phantoms. The 5 mA anti-isowatt curve was used in the automatic brightness control algorithm with the target intensifier dose rate set nominally at 0.13 µGy s-1.
|
|
It is interesting to note the discontinuity in ESDR that occurs at 33 cm for PMMA and 36.5 cm for water (Figures 8a,b). As phantom thickness is increased up to these values, the unit is able to maintain the desired input air kerma rate to the image intensifier by moving along the selected ABC curve. Phantom ESDR therefore increases due both to the change in operating point and to the phantom surface becoming closer to the X-ray tube. However, the discontinuity in ESDR represents the final tube potential/tube current operating point on the ABC curve being reached such that the desired input level to the image intensifier cannot be maintained if phantom thickness is increased further. From this point onwards, phantom ESDR increases only in proportion to 1/r2 (+ backscatter) as the entrance surface moves closer to the tube but the operating point remains constant.
The presence of the discontinuities has important implications for clinical practice. To obtain a good quality image it is essential that the input air kerma rate to the image intensifier is at the expected level so that the luminance of the output phosphor is within the appropriate range. However, for very large patients the maximum ABC operating point of the unit may be insufficient to achieve this. A large system gain will then be necessary to produce a large enough video signal from the low output phosphor luminance and an image with substantial quantum noise will result. By using the model to generate graphs of ESDR vs patient thickness for a particular clinical situation, the maximum thickness of patient that could successfully be imaged at each image intensifier input setting would be revealed by the discontinuity in the appropriate graph. The model can therefore be used to indicate the intensifier input necessary to produce a good quality image.
The model can be considered as the first stage in the development of a simulation environment that can be used to look at both the efficacy of dose reduction methods and the effect on image quality introduced by the adoption of these methods. It represents an accurate simulation of the X-ray elements of an image intensifier system working under ABC. As such, it can be used as the input to further modules that can be utilized to simulate image degradation by, for example, patient scatter, scatter in the phosphor itself, noise and pixelation [24]. An investigation of these aspects will be the subject of future work.
The model can also be used as the input stage of a system designed to assess variation of scattered radiation with angle around a simulated patient. This would make it possible to model exposure and, therefore, estimate risk to staff at different physical positions. The calculation of scatter dose and spectra can be performed using Monte Carlo methods as outlined by Sutton and Williams [25].
|
Conclusions
|
|---|
A computer model of a fluoroscopy unit operating under ABC has been developed. Although the model specifically simulates a Siemens Siremobil 2000 unit, it is very general in nature and can easily be applied to other image intensifier systems. In order to develop the model, backscatter factors have been generated using Monte Carlo techniques. Predictions of ESDRs generated by the model for water and PMMA phantoms are in very good agreement with experimental measurements.
The model can be used for a variety of applications, including examining the effectiveness of different dose reduction measures. Examples of investigations that can be performed include determining the effect of changing the ABC curve and input dose rate, changing geometries and employing different degrees and types of filtration. The model can also be used to provide data that can be input to Monte Carlo problems reflecting a variety of situations encountered in practice. One example is the study of the distribution and quality of scattered air kerma in the environment of the unit during specific procedures. Another is the investigation of image quality parameters and their relationship with patient dose.
|
Appendix: Justification of use of air kerma in the automatic brightness control algorithm
|
|---|
Although automatic brightness control (ABC) works to maintain a constant rate of energy deposition in the caesium iodide (CsI) input phosphor, the ultimate goal of the ABC algorithm in the model is simply to choose the correct tube potential/tube current operating point for the unit. Therefore, by showing that the same operating point is selected when air kerma rate at the phosphor is considered instead, then the use of air kerma in the model is justified. To do this, two models were constructed: the first with the ABC algorithm employing air kerma rates (as described in the main text); the second with the algorithm governed by absorbed dose rate in CsI.
Before an ABC system that aims to maintain a target dose rate in the phosphor can be simulated, it is first necessary to determine what the target dose rate is. This was achieved by monitoring the dose rate to the CsI phosphor at the intensifier face whilst running a simulation with the air kerma based system. In the investigation the same geometry as in Figure 1 was modelled, with the table transmission equivalent to 0.7 mm aluminum at 110 kVp. The distance between the base of the phantom and the surface of the image intensifier was 10 cm and the ABC algorithm was set to deliver 0.13 µGy s-1 (in air) to the intensifier by following the 5 mA anti-isowatt curve.
Different thicknesses of polymethylmethacrylate (PMMA) phantom were inserted into the X-ray beam and the tube potential/tube current operating point calculated for each. At each thickness, the operating point determined was used to calculate the dose rate in CsI at the input surface of the image intensifier. Owing to the small thickness of the phosphor (550 µm) (Zimmermann, Siemens Erlangen, 2000, personal communication), it was unnecessary to account for scatter build-up. The results are plotted in Figure 10 and clearly illustrate the ABC operating range (5–33 cm PMMA) discussed in the main body of the text. Over this range the average dose rate in CsI is 16±3 µGy s-1.
View larger version (10K):
[in this window]
[in a new window]
|
Figure 10. Calculated absorbed dose rate in the caesium iodide (CsI) input phosphor for different thicknesses of PMMA phantom. The automatic brightness control algorithm was set to provide the input phosphor with an air kerma rate of 0.13 µGy s-1 using the 5 mA anti-isowatt curve.
|
|
A model was constructed with an ABC algorithm that considers dose rate in CsI rather than kerma rate in air by exchanging the photon fluence to air kerma conversion factors in the original model with those appropriate for CsI. As an initial estimate, the target dose rate was set to16 µGy s-1, the mean value found over the operating range.
Using both the original (air kerma rate based) and the new (CsI dose rate based) models, the operating points were determined for different thicknesses of PMMA phantom. The simulation was repeated for all the available Siremobil 2000 ABC curves and a comparison was made between the operating points calculated by each model under the same conditions. It was found iteratively that the best agreement was achieved between the two models when the target CsI dose rate was set at 17 µGy s-1. A comparison between the points selected by each model for the various ABC curves is presented in Figure 11.
View larger version (16K):
[in this window]
[in a new window]
|
Figure 11. Comparison between the operating points chosen to maintain a constant air kerma rate (0.13 µGy s-1) or caesium iodide (CsI) dose rate (17 µGy s-1) at the surface of the image intensifier for different thicknesses of PMMA phantom. ---, model based on air kerma rates; —, model governed by CsI dose rates. (Each operating point is a unique combination of tube potential and tube current, and automatic brightness control (ABC) curves are designed so that the dose rate at the intensifier, in the absence of attenuating materials, monotonically increases from one operating point to the next. The points are numbered incrementally with increasing dose.) ABC curves: (a) 5 mA anti-isowatt; (b) 3 mA anti-isowatt; (c) 5 mA high contrast; (d) 3 mA high contrast; (e) low dose/paediatric.
|
|
For all ABC curves the agreement between the two models was better than 5% for thicknesses greater than 10 cm. Since each operating point corresponds to a unique combination of tube potential and tube current, demonstrating that the correct operating point is chosen also indicates that the correct tube potential and tube current were selected. The use of air kerma rate in the ABC algorithm is therefore justified.
It is interesting to note that as the thickness ofthe phantom decreases below 10 cm, the air kerma model increasingly underestimates the correct operating point. This is because photons with energy below that of the CsI K-edge begin to account for a significant portion of the fluence spectrum and hence the greater relative sensitivity of air to these photons causes the ratio between air kerma and dose in CsI to increase.
Received for publication August 31, 1999.
Accepted for publication April 18, 2001.
|
References
|
|---|
-
National Radiological Protection Board. Patient dose reduction in diagnostic radiology. Documents of the NRPB 1990;Vol. 1, No. 3
-
Sutton D, Cranley K. The potential for dose reduction in diagnostic radiology using low attenuation materials in table tops. Br J Radiol 1996;69:249–55.[Abstract/Free Full Text]
-
Martin CJ, Sutton DG, Sharp PF. Balancing patient dose and image quality. Appl Radiat Isotopes 1999;50:1–19.[Medline]
-
Aufrichtig R, Xue P, Thomas CW, Gilmore GC, Wilson L. Perceptual comparisons of pulsed and continuous fluoroscopy. Med Phys 1994;21:245–56.[Medline]
-
Nicholson RA, Thornton A, Akpan M. Radiation dose reduction in paediatric fluoroscopy using added filtration. Br J Radiol 1995;68:296–300.[Abstract/Free Full Text]
-
Hoornaert B, Kroon JN. Publication No. 4522 984 41761/722: additional beam filtering in cardiac fluoroscopy. Eindhoven, The Netherlands: Philips Medical Systems, 1993.
-
Den Boer A, De Feyter P, Hummel WA, Keane D,Roeland JRTC. Reduction of radiation exposure while maintaining high quality fluoroscopic images during interventional cardiology using novel X-ray tube technology with extra beam filtering. Circulation 1994;98:2710–4.
-
Dowsett DJ, Kenny PA, Johnston RE. The physics of diagnostic imaging. London: Chapman & Hall Medical, 1998:228–53.
-
Kopp J, Maier W, Losereit C. Radiation exposure of patients by using modern digital fluoroscopy systems. Radiat Prot Dosim 1995;57:487–8.[Abstract]
-
International Commission on Radiation Units and Measurements (ICRU). ICRU Report 54: Medical imaging—the assessment of image quality. Bethesda MD: ICRU, 1996.
-
Martin CJ, Sharp PF, Sutton DG. Measurement of image quality in diagnostic radiology. Appl Radiat Isotopes 1999;50:21–38.[Medline]
-
Sutton DG, Reilly AJ. Report 78 Spectrum Processor. York: IPEM, 1997.
-
Cranley K, Gilmore BJ, Fogarty GWA, Desponds L, Sutton DG. Catalogue of diagnostic X-ray spectra & other data, IPEM Report 78 (CD-ROM). York: IPEM, 1997.
-
Hubbell JH, Seltzer SM. Tables of X-ray mass attenuation coefficients and mass-energy absorption coefficients 1 keV to 20 MeV for elements Z=1 to 92 and 48 additional substances of dosimetric interest. NISTIR 5632. Gaithersburg, MD: National Institute of Standards and Technology, 1995. http://physics.nist.gov/PhysRefData/XrayMassCoef/cover.html
-
Greening JR. Fundamentals of radiation dosimetry (2nd edn). Bristol, UK: Adam Hilger, 1985:66.
-
Gagne RM, Quinn PW. X-ray spectral considerations in fluoroscopy. Physical and technical aspects of angiography and interventional radiology. Oak Brook, IL: RSNA, 1995.
-
Boone JM, Pfeiffer DE, Strauss KJ, Rossi RP, Lin PP, Shepard JS, et al. A survey of fluoroscopic exposure rates: AAPM Task Group No. 11 Report. Med Phys 1993;20:789–94.[Medline]
-
Schonander NG. Patent specification UK569231: improvements in X-ray diaphragms for intercepting secondary radiators. Leamington Spa, UK: HMSO, 1945.
-
Dowsett DJ, Kenny PA, Johnston RE. The physics of diagnostic imaging. London: Chapman & Hall Medical, 1998:193.
-
Petoussi-Henss N, Zankl M, Panzer W, Regulla D. Calculation of backscatter factors for diagnostic radiology using Monte Carlo methods. Phys Med Biol 1998;43:2237–50.[Medline]
-
Briesmeister JF, editor. MCNPTM — a general Monte Carlo N-Particle transport code. Los Alamos National Laboratory report LA-13709-M. Los Alamos, NM: Los Alamos National Laboratory, 2000. http://www.lanl.gov/XCI/PROJECTS/MCNP/manual.html
-
Geist A, Beguelin A, Dongarra J, Jiang W, Manchek R, Sunderam V. PVM: parallel virtual machine—a users' guide and tutorial for networked parallel computing. Cambridge, MA: MIT Press, 1996.
-
Day MJ, Aird EG. The equivalent field method for dose determinations in rectangular fields. Br J Radiol 1996;69(Suppl. 25):138–51.
-
Reilly AJ. A computer model of fluoroscopy. MSc thesis, University of Aberdeen, 2000.
-
Sutton DG, Williams JR. Radiation shielding for diagnostic X-rays. London: British Institute of Radiology, 2000:4–6.
This article has been cited by other articles:
|
|
|
|
 
D J Gallacher, A Mackenzie, S Batchelor, J Lynch, and J E Saunders
Use of a quality index in threshold contrast detail detection measurements in television fluoroscopy
Br. J. Radiol.,
July 1, 2003;
76(907):
464 - 472.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
, 5 mA anti-isowatt; 
, 3 mA anti-isowatt; 
, 3 mA high contrast; +, 5 mA high contrast; x , 3 mA low dose.
