- Email: [email protected]
Technique charts for EC film: direct optical measurements to account for the effects of X-ray scatter
Int. J. Radiation Oncology Biol. Phys., Vol. 50, No. 3, pp. 829 – 836, 2001 Copyright © 2001 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/01/$–see front matter
PII S0360-3016(01)01564-4
PHYSICS CONTRIBUTION
TECHNIQUE CHARTS FOR EC FILM: DIRECT OPTICAL MEASUREMENTS TO ACCOUNT FOR THE EFFECTS OF X-RAY SCATTER PETER MUNRO, PH.D.,*† KEVIN JORDAN, PH.D.,*† CRAIG LEWIS, M.SC.,*†
AND
TIM HEEREMA*
*London Regional Cancer Centre, London, Ontario, Canada; †Department of Oncology, University of Western Ontario, London, Ontario, Canada Purpose: To develop a method of measuring technique charts for enhanced contrast (EC) film, to demonstrate how X-ray scatter changes the response of EC film, and to generate technique charts for general use. Methods and Materials: We have developed a “digital cassette”— consisting of a metal plate/phosphor screen, a light guide, a photodiode sensor, and an electrometer—that can be used to measure the light generated in the phosphor screen of the film cassette. In turn, these measurements can be used to generate technique charts for EC film. The digital cassette has been used to measure technique charts for 4-MV and 6-MV X-ray beams for a variety of different phantom thicknesses, field sizes, and phantom-to-cassette air gaps. Results and Discussion: We have observed that the signals generated in an ionization chamber located 9.4 cm behind a 30-cm-thick water-equivalent phantom increase by a factor of 1.9 when the field size is increased from 4 ⴛ 4 cm2 to 40 ⴛ 40 cm2 when irradiated by a 6-MV X-ray beam. However, the change in EC film response is a factor of 3.5 under the same conditions. Irradiations to optimally expose the EC film predicted by the digital cassette differ by up to 82% compared to those predicted by ion chamber measurements. Nevertheless, the technique charts measured using the digital cassette predict the response of the EC film to ⴞ 0.2 optical density. The overresponse of the EC film is most likely due to low-energy scattered photons, which interact with the high atomic number (Z ⴝ 64) phosphor screen of the enhanced contrast localization cassette. Therefore, simple solutions, such as placing a high atomic number material above the enhanced contrast localization cassette, can reduce this contribution by scattered photons to the signal generated in the cassettes. Conclusions: We have developed a digital cassette that can make more accurate measurements of the technique charts for EC films. Our measurements show that under some conditions, X-ray scatter can generate a large fraction of the signals recorded by the EC film. Technique charts have been generated at 4 MV and 6 MV, and these charts should have universal applicability. © 2001 Elsevier Science Inc. X-ray scatter, Portal film, EC film.
plate (3, 4). Finally, the film has a gamma (i.e., display contrast), which is 3.5 times larger than the conventional metal plate/film combinations (1, 5). This high film gamma, which is the result of a narrow distribution of grain sizes and metal ion doping of the microcubic film grains, improves the display of the low-contrast structures found in megavoltage radiography and is often the main advantage cited for the use of EC film (1). One complication is that the high gamma limits the latitude of the EC film, requiring that users make accurate estimates of the irradiation necessary to optimally expose the film. As a result, methods for generating technique charts (tables of monitor unit settings as a function of patient thickness and field size) have been described in the literature (6). The method of Lee and Glasgow uses ionization chamber measurements (e.g., measurements of tissue-
INTRODUCTION In 1996, Kodak introduced a new film-screen combination system for megavoltage radiography. The system consists of an enhanced contrast (EC) film, and new localization (EC-L) and verification (EC-V) cassettes. The EC-L cassette is constructed using a 1-mm copper front plate and a Lanex Fast gadolinium oxysulfide screen pair. This filmscreen system has three advantages compared to the conventional metal plate/film combinations used in the past. The film granularity for the new EC film is much lower than that for conventional portal films because of the use of a microcubic film grain structure (1, 2). Also, the use of a phosphor screen increases the quantum efficiency of the EC-L cassettes, because X-ray quanta can interact directly in the phosphor screen as well as interacting in the metal
search) and Cancer Care Ontario is greatly appreciated. Acknowledgments—We thank Matt Mulligan, Tim Van Dijk, and Howard Siu for their help with some of the measurements with the ion chamber and the digital cassette. Accepted for publication 14 March 2001.
Reprint requests to: Peter Munro, London Regional Cancer Centre, 790 Commissioners Road East, London, Ontario N6A 4L6, Canada. Tel: (519) 685-8500, ext. 53317; Fax: (519) 6858658; E-mail: [email protected] Financial assistance from the Medical Research Council of Canada (now known as the Canadian Institutes of Health Re829
830
I. J. Radiation Oncology
●
Biology
●
Physics
Fig. 1. The “digital cassette” formed from a conventional film/ screen cassette, a 1-mm copper plate, a Lanex Fast (back) Gd2O2S:Tb phosphor screen, a liquid light guide, a photodiode, and an electrometer. The photodiode is shielded in a cylindrical lead container located beside the cassette.
maximum ratios, tissue-air ratios, field size factors, etc.) to generate values from which technique charts are calculated (6). However, our experience has shown that the technique charts generated in this way are not universally usable; small field sizes result in films that are too light, and large field sizes result in films that are too dark. The results of Keys et al. also show that optimum exposure is difficult to predict (7). Keys et al. found that adjusting the air gap between an EC-V cassette and the patient—to control the dose reaching the cassette during verification imaging— resulted in an inverse cube relationship, rather than the expected inverse square relationship (7). In this paper we describe a “digital cassette,” which consists of a metal plate/phosphor screen, a light guide, a photodiode sensor, and an electrometer. Using this cassette we can directly measure the amount of light emitted by the phosphor screen during an irradiation. Knowing the amount of light needed to optimally expose the EC film, we can use the measurements from the digital cassette to predict the irradiation that will optimally expose the EC film under a wide variety of conditions (e.g., field size, patient thickness, air gap between patient and cassette). We also present evidence that shows that low-energy scattered photons are the most likely cause of the discrepancy in response between the ion chamber measurements and the EC-L cassette.
METHODS AND MATERIALS Digital cassette The digital cassette (Fig. 1) was constructed using a standard kilovoltage film cassette from which the phosphor screens had been removed. The phosphor screens were replaced by a 1.0-mm copper plate in contact with a single Lanex Fast back screen. The gadolinium oxysulfide phosphor loading of this detector (134 mg/cm2) was only slightly less than that of the standard EC-L cassette (168 mg/cm2).
Volume 50, Number 3, 2001
A Series 800 liquid light guide, with a core diameter of 5 mm and a length of 1.5 m (Lumatec GMBH, Munich, Germany), was attached to the lid of the cassette using a mechanical coupling that had been fabricated locally. When the lid was closed, the light guide was in intimate contact with the phosphor screen. Another coupling was used to hold a series 5T photodiode, with an active area of 100 mm2 (Centronic, Newbury Park, CA) in contact with the other end of the light guide. The light guide allowed the photodiode to be positioned well outside the X-ray beam of the accelerator while measuring the light emitted from the phosphor screen. Optical filters (an infrared filter and a Wratten 15 filter) were placed between the light guide and the photodiode to limit the wavelengths of light (500 – 650 nm) reaching the photodiode. Since our measurements showed that direct interaction of X-ray radiation with the photodiode could generate large signals, the photodiode was shielded in a lead cylinder with a wall thickness of 3 cm. Additional lead blocks 5-cm thick surrounded the cylinder to ensure a high degree of shielding. A 30-m coax cable connected the photodiode to a Keithley Model 35617EBS electrometer, which was located outside the treatment room. Irradiations Irradiations used Varian 2100 C/D dual-energy linear accelerators, with either 6/18 or 4/10 X-ray energies. All measurements were made using a 6-MV X-ray beam, except for the results presented in one table, where measurements were made using a 4-MV X-ray beam. Lucite sheets 50 ⫻ 70 cm2 ⫻ ⬃1 cm thick were used to generate a phantom that varied in thickness between 8.5 and 52.5 cm. The Lucite sheets were placed on the treatment couch, and the cassette was placed either in the cassette holder of the couch (9.4 cm air gap) or on a mechanical stage that allowed the air gap to be adjusted to either 15 or 20 cm. The couch was not moved as the Lucite sheets were added, so that the source-to-exit surface of the phantom remained fixed at 135 cm. Collimator settings ranged between 4 ⫻ 4 and 40 ⫻ 40 cm2, which corresponded to field sizes between 5.4 ⫻ 5.4 and 54 ⫻ 54 cm2 when defined at the exit plane of the phantom. Care had to be taken to eliminate spurious signals generated in the digital cassette. The digital cassette was irradiated with an opaque material located between the phosphor screen and the liquid light guide. With the photodiode well outside the treatment beam and the photodiode shielded by the lead cylinder and lead blocks, a “nonphosphor” signal ⬃10% of the total optical signal was generated. Use of Wratten 15 and infrared filters reduced the “nonphosphor” signal by ⬃80% and reduced the optical signal to 60% of its original magnitude. Thus, after optical filtering, the “nonphosphor” signal was only ⬃3% of the optical signal. The Wratten 15 filter, which removes short wavelength light, and the IR filter, which removes long wavelength light, contributed equally to the reduction in the “nonphosphor” signal. Measurements showed that the “nonphosphor” signal changed from ⬃3% to ⬃4% of the total
EC film, technique charts, and X-ray scatter
signal as the field size changed. Thus, to simplify data acquisition and data correction, the small, almost constant, contribution of the “nonphosphor” signal was ignored. Another problem was that the measured signal decreased with increasing accumulated dose received by the digital cassette. The radiation created yellow products in the liquid light guide, thereby changing its transmission properties. To account for these changes, which could be as large as 7% for a 3000 monitor unit irradiation, measurements were repeated at standard field sizes and phantom thicknesses after every 5–7 irradiations. The raw readings were then corrected for any changes between the two standard measurements, assuming that the optical transmission of the light guide changed linearly with radiation dose. To minimize the total dose delivered to the digital cassette, all readings were acquired using 25 monitor unit irradiations. Thus, a technique chart (for one phantom to cassette air gap) could be measured using a total irradiation of only 3000 monitor units. Despite the small irradiations, repeatability was ⫾0.3%, and the signals were quite large, ranging between 4 and 65 nC. Receptor response and scatter measurements The response of an ion chamber, the digital cassette, and the EC-L film/cassette to a 6-MV radiation beam was measured. The ion chamber, a Capintec PR06C air-equivalent ion chamber connected to a Capintec 292 electrometer, was placed behind a 27.5-cm-thick Lucite phantom, which was sitting on the treatment couch with the couch top at 135 cm. The ion chamber was placed at a depth of 1.5 cm inside a 3-cm-thick Lucite plate to ensure that the chamber was at a depth of full buildup. With the air gap set to 9.4 cm, the signals from the ion chamber were recorded as the collimator settings varied between 4 ⫻ 4 cm2 and 40 ⫻ 40 cm2. The digital cassette was placed in the cassette holder of the treatment couch (air gap 9.4 cm), and the measurements for the same field sizes were repeated. Finally, the EC-L film/ cassette was placed in the cassette holder of the couch, and the monitor unit settings were adjusted with field size so that the films were all exposed to the same net optical density (OD) of ⬃1.5 OD. The ratio of monitor unit settings for each field size was taken to represent the relative response of the EC-L film/cassette. If the optical densities on the films were not exactly the same, the above ratio was corrected assuming a linear film response. This was a reasonable assumption because of the small differences between the two optical densities and because the optical densities were in the linear region of the film response curve. A 1.5-mm tungsten plate was used to measure the fraction of the signals generated in the film cassettes by low-energy photons. This plate was placed above the film cassettes and the ion chamber to absorb (i.e., filter) the low-energy components from the beam. The experiments described in the previous paragraph were then repeated to measure how the presence of the tungsten plate changed the response of the image receptors. The D vs. log (E) (H&D) curves of the EC film were also
● P. MUNRO et al.
831
measured using the phantom setup described above. The dose at the plane of the cassette for 4 ⫻ 4 cm2, 10 ⫻ 10 cm2, and 40 ⫻ 40 cm2 collimator settings and various monitor unit settings was measured using the ion chamber. Films were exposed in the EC-L cassette using the same irradiations and were processed in a Kodak M35 processor using Kodak RP chemicals. The resulting optical density, which was measured using an X-rite Model 301 densitometer, was then plotted vs. log (dose) to yield the desired curves. Effective source location The measurements using the digital cassette were made with the phantom at extended distances, whereas most patients will not be treated at extended distances. Therefore, the effect on the measured signal of changing the position of the phantom, while keeping the air gap constant, needed to be determined. The phantom was moved over a range of 30 cm, with the position of the couch top changing between 135 and 105 cm SSD. Measurements were made for “nominal” collimator settings of 4 ⫻ 4 cm2 and 40 ⫻ 40 cm2 and Lucite thicknesses of 18.5 and 42 cm. The collimator settings were only “nominal” because, as the position of the phantom changed, the collimator setting was adjusted so that the same area, at the position of the treatment couch, was irradiated. The effective source location was then estimated from a plot of the square root of normalized dose vs. distance. A source offset was calculated from these measurements, which was used in the calculations of the technique charts (See Appendix A). Tests of the technique charts The technique charts were calculated using the approach described in Appendix A. The technique charts were tested by exposing EC-L film/cassettes under a variety of different fields sizes (square and rectangular) and patient thicknesses. If the predicted irradiation was not an integer value of monitor units, integer monitor unit irradiations greater than and less than the predicted value were used to expose two films. The optical density from both films was then used to estimate the optical density that would have resulted had a fractional monitor unit irradiation been possible, assuming a linear response between the two exposed films. RESULTS The D vs. log (E) curves are shown in Fig. 2 for a 27.5-cm-thick Lucite phantom (⬃30 cm water equivalent) with a 9.4-cm air gap between the phantom and the cassette irradiated by a 6-MV X-ray beam. The figure shows the effect of changing the collimator setting on the response of the EC film. The abscissa represents the dose reaching the plane of the image receptor as measured using the PR-06 ion chamber. The dose as measured by the ion chamber does not predict the optical density generated on the EC film. The number of developable film grains depends on the number of optical quanta reaching the film (8, 9), which in turn depends on the dose (energy) deposited in the EC-L cas-
832
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
Table 1. Effective source location Patient separation (cm)
FS (cm)
Source displacement (cm)
20 20 47 47
4⫻4 40 ⫻ 40 4⫻4 40 ⫻ 40
8 ⫺1 23 46
The apparent position of the X-ray source (SSDoff) as a function of field size and patient thickness. To simplify our calculations, SSDoff was set to 15 cm for all patients thicknesses and field sizes.
Fig. 2. The optical density of the EC film as a function of dose, where the dose is measured using an ion chamber. The EC film appears to need less dose to reach the same optical density as the collimator setting increases.
sette. Clearly, the dose measured by the ion chamber must not represent the actual dose deposited in the screen of the EC-L film/cassette. An explanation for the discrepancy between the response of the EC-L film/cassette and the dose measured by the ion chamber is shown in Fig. 3. Figure 3 (a) shows the signals generated in the ion chamber and the EC-L film/cassette as a function of collimator setting, with and without the presence of the 1.5-mm tungsten plate, when a 27.5-cm Lucite phantom is irradiated by a 6-MV beam. The presence of the tungsten plate does not change the response of the PR06 ion
chamber greatly; however, there is a large change in the response of the EC-L film/cassette. Indeed, after tungsten filtration, the EC-L film/cassette responds to the X-ray beam in a similar fashion as the ion chamber. Clearly, the tungsten plate is able to remove from the X-ray beam those components that are causing the overresponse of the EC-L film/ cassette, presumably low-energy scattered photons generated in the phantom. Figure 3 (b) shows the response of the digital cassette to the 6-MV beam. The response is very similar to that of the EC-L film/cassette, showing that the digital cassette is an excellent alternative for measuring technique charts for the EC-L film/cassette. The results for the measurement of the effective source location are shown in Table 1. In general, the change in signal as the Lucite phantom and image receptor are moved closer to the X-ray source is not as great as would be predicted by the inverse square relationship. The source appears to be farther away than the X-ray target. To calculate the correct change in signal using the inverse square formalism (See Appendix A), a source offset
Fig. 3. Panel (a) shows the effect of a 1.5-mm tungsten plate on the signals generated in the EC film and the ion chamber. The dotted and solid curves show the signals measured by the ion chamber with and without the tungsten plate, respectively. The solid circles and solid squares show the signals measured by the EC film with and without the tungsten plate, respectively. The tungsten plate appears to absorb a component of the beam that contributes substantially to the signals recorded by the EC film but that contributes relatively little to the signals generated in the ion chamber. Panel (b) shows the response of the “digital cassette” compared to that of the EC film and the ion chamber (without a tungsten plate). Irradiations used a 6-MV X-ray beam and a 27.5-cm Lucite phantom.
EC film, technique charts, and X-ray scatter
● P. MUNRO et al.
833
Table 3. Technique charts— 4 MV
Table 2. Technique charts— 6 MV 9.4-cm air gap
9.4-cm air gap
Sep\FS
5.4
13.5
27
40.5
54
Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.512 0.844 1.053 1.347 1.671 2.077 2.723 3.410
0.407 0.616 0.752 1.000 1.209 1.431 1.886 2.313
0.311 0.419 0.494 0.626 0.749 0.894 1.160 1.433
0.260 0.324 0.373 0.462 0.545 0.651 0.843 1.035
0.230 0.278 0.315 0.385 0.451 0.536 0.687 0.846
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.449 0.874 1.104 1.593 2.059 2.683 3.745 4.841
0.332 0.574 0.730 1.000 1.276 1.647 2.266 2.892
0.231 0.342 0.415 0.552 0.696 0.882 1.218 1.524
0.184 0.250 0.297 0.384 0.481 0.601 0.822 1.035
0.163 0.211 0.248 0.315 0.387 0.487 0.656 0.830
15-cm air gap
15-cm air gap
Sep\FS
5.4
13.5
27
40.5
54
Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.549 0.915 1.131 1.456 1.797 2.183 2.845 3.446
0.440 0.686 0.831 1.053 1.285 1.547 2.006 2.418
0.332 0.461 0.547 0.676 0.815 0.974 1.250 1.512
0.271 0.347 0.406 0.494 0.592 0.705 0.901 1.085
0.236 0.289 0.336 0.405 0.484 0.573 0.731 0.881
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.496 0.965 1.219 1.759 2.274 2.962 4.135 5.346
0.367 0.634 0.806 1.104 1.409 1.819 2.502 3.194
0.255 0.378 0.458 0.610 0.769 0.974 1.345 1.683
0.203 0.276 0.328 0.424 0.531 0.664 0.907 1.143
0.180 0.233 0.273 0.348 0.427 0.537 0.724 0.917
20-cm air gap
20-cm air gap
Sep\FS
5.4
13.5
27
40.5
54
Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.594 1.002 1.253 1.624 1.998 2.418 3.192 3.853
0.483 0.777 0.951 1.205 1.472 1.764 2.305 2.778
0.369 0.538 0.640 0.796 0.955 1.136 1.472 1.760
0.299 0.407 0.473 0.582 0.692 0.822 1.062 1.270
0.255 0.337 0.388 0.473 0.559 0.664 0.855 1.021
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.577 1.123 1.419 2.048 2.647 3.449 4.814 6.224
0.427 0.738 0.939 1.286 1.641 2.117 2.913 3.718
0.296 0.440 0.534 0.710 0.895 1.134 1.566 1.959
0.237 0.321 0.382 0.494 0.619 0.772 1.056 1.330
0.209 0.271 0.318 0.405 0.497 0.626 0.843 1.067
The relative response of the digital cassette as a function of field size (FS), patient thickness (SEP), and air gap for a 6-MV X-ray beam. The values in the table represent the irradiation necessary to give the same optical density on an EC film irradiated in an EC-L cassette. The field size, which is given as the side of the square field, is defined at the exit surface of the phantom at 135 cm from the source. All values have been normalized to a 13.5 ⫻ 13.5 field (10 ⫻ 10 collimator setting), patient thickness of 30.5 cm (water equivalent), and an air gap of 9.4 cm.
The relative response of the digital cassette as a function of field size (FS), patient thickness (SEP), and air gap for a 4-MV X-ray beam. The values in the table represent the irradiation necessary to give the same optical density on an EC film irradiated in an E C-L cassette. The field size, which is given as the side of the square field, is defined at the exit surface of the phantom at 135 cm from the source. All values have been normalized to a 13.5 ⫻ 13.5 field (10 ⫻ 10 collimator setting), patient thickness of 30.5 cm (water equivalent), and an air gap of 9.4 cm.
has been introduced into the inverse square calculation. For simplicity, we chose a single value of 15 cm, even though the source offset ranged from 0 to 45 cm, depending on field size and phantom thickness. This value was obtained by averaging all the numbers in Table 1 and rounding off the average value to the lower multiple of 5 cm. Tables 2 and 3 give the technique charts for the EC-L film/cassette for 4-MV and 6-MV X-ray beams. All the measurements have been normalized to the 13.5 ⫻ 13.5 cm2 field size (10 ⫻ 10 cm2 collimator setting), the 30.5 cm (27.5 cm Lucite) phantom thickness, and an air gap of 9.4 cm. The values in the tables are the relative exposures required to generate the same optical density on the film. Therefore, to use these tables in his own clinical practice, the user needs only to determine the irradiation (in monitor
units) to properly expose the EC-L film/cassette at the reference conditions. Table 4 gives the ratio of exposures predicted by ion chamber measurements divided by the exposures predicted by the digital cassette. (Note that the ion chamber measurements have been made under identical conditions to those used for the digital cassette [i.e., 6-MV X-ray beam, 9.4-cm air gap, the same field sizes, and phantom thicknesses]. Also note that both sets of data have been normalized to 13.5 ⫻ 13.5 cm2 field size [10 ⫻ 10 cm2 collimator setting] and the 30.5-cm phantom thickness before the ratios have been calculated.) At large field sizes, the ion chamber measurements predict that the irradiation required for optimal exposure is up to 82% larger (54 ⫻ 54 cm2; 35-cm separation) than exposures predicted by the digital cassette. At small field sizes, the ion chamber measurements predict an irra-
834
I. J. Radiation Oncology
●
Biology
●
Physics
Table 4. Ion chamber over- and under-response
Volume 50, Number 3, 2001
thickness used for the measurements. Figure 4 shows that we are able to predict the optical density of the exposed films to ⫾0.2 OD.
9.4-cm air gap Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.95 0.95 0.94 0.94 0.97 0.91 0.92 0.85
1.05 1.06 1.05 1.00 1.04 1.02 1.02 0.97
1.22 1.30 1.32 1.28 1.33 1.28 1.28 1.20
1.38 1.54 1.57 1.55 1.61 1.55 1.53 1.44
1.51 1.71 1.76 1.75 1.82 1.75 1.73 1.62
The relative response of the ion chamber divided by that of the digital cassette, when both sets of data have been normalized to the reference conditions. When the numbers in the table are greater than 1.0, they indicate that the ion chamber predicts a larger irradiation than is necessary to properly expose the EC film.
diation that is 15% smaller (5.4 ⫻ 5.4 cm2; 52 cm separation). These results are consistent with our clinical observations when we use technique charts generated by ion chambers: small field sizes result in light films, and large field sizes result in dark films. One can generate the results in Table 4 by taking the ratio of the ion chamber and digital cassette curves in Fig. 3. However, the normalization is for a 13.5 ⫻ 13.5 cm2 field size (10 ⫻ 10 cm2 collimator setting) in Table 4, and this is why some values are less than 1.0. The accuracy of our technique charts in predicting the optical density on irradiated EC films is shown in Fig. 4. All films were exposed with an air gap of 9.4 cm. The height of the bars represent the optical density on the film, while the labels for each bar indicate the field size and phantom
Fig. 4. The accuracy of the technique charts when used to expose EC film. The goal was to obtain an optical density of 1.5.
DISCUSSION AND CONCLUSIONS Our results suggest that X-ray scatter from the patient can generate a large fraction of the signal recorded on EC film, depending upon field size, patient thickness, and air gap. This X-ray scatter complicates the measurements of technique charts for EC film, because the X-ray scatter does not generate as large a signal in an ion chamber as in the film/cassette. As a result, we have had to develop a method to directly measure the light generated by a Gd2O2S screen when irradiated by megavoltage X-ray beams. We found that the largest differences between the response of the ion chamber and that of the EC film occur as a function of field size rather than as a function of depth. A simple method to account for the effects of X-ray scatter would be to make measurements using ion chambers, but use different film response curves (e.g., Fig. 2) for each field size. In other words, use different “doses” (as measured by the ion chamber) at different field sizes to optimally expose the EC film. Whereas such an approach would not be as accurate as our direct measurements, the approach might generate clinically usable technique charts with equipment widely available in radiation therapy departments. Certainly, use of different film response curves would be considerably more accurate than using one response curve. Although our measurement approach worked well, there are some possible improvements. One of the difficulties with our experimental procedure was the change in transmission of the liquid light guide. Over the course of many measurements, the transmission dropped by a factor of five, making the later measurements more difficult because of the smaller signals generated in the photodiode. An improvement would be to use a quartz rod, or fiber optic materials (such as those used in scintillation detectors), to transport the light outside of the radiation beam (10, 11). These materials are less susceptible to radiation coloring than the liquid light guide and thus should result in a more stable measuring tool. Another difficulty was the “nonphosphor” signals (probably mostly Cerenkov radiation [12, 13]) generated in the liquid light guide. Our unpublished calculations suggest that given the 1/3 dependence of Cerenkov radiation intensity, the transmission of the light guide, and the response of the silicon photodiode, the “nonphosphor” signal is likely to have a uniform value at all wavelengths (12, 13). (Silicon photodiodes have very high optical sensitivities at red and infrared wavelengths, so even a small number of optical photons at these wavelengths can result in large signals being generated in the photodiode.) We used readily available optical filters to create an optical passband of 500 to 650 nm to minimize the spurious optical signals reaching our photodiode. An improvement would be to use an optical interference filter with a narrower passband centered on the main emission peak of Gd2O2S:Tb located at
EC film, technique charts, and X-ray scatter
545 nm. The main emission peak generates 54% of the optical signal from the Gd2O2S:Tb phosphor, and it can be completely transmitted by a filter with a 20-nm passband. In principle, such a filter would reduce our signal to ⬃54% of its original value (the current filters currently remove some of the signal from the main emission peak) but might reduce the “nonphosphor” signals by the ratio of the passband widths (150/20 ⫽ 7.5). This would make the “nonphosphor” signals even less of a concern. The effective source location measurements in Table 1 appear to have some uncertainties, since we would expect the source location displacement to be larger for larger field sizes where X-ray scatter is more important. However, for the 20-cm-thick phantom, this trend does not occur. The measurements were made when the transmission of the light pipe was quite low. So we believe that experimental uncertainty is the cause of the unpredictable trends. This is why we decided on one value for the source offset. Despite the assumption of a single source offset our table performs well. Let us assume that we are in error in our estimate of the source offset by 10 cm. This would make an error in our estimates of dose of 1052/1152 ⫽ 0.834. Using Fig. 2, such a change in dose for the high-gradient region of the film would result in a change in optical density of 0.55. However, our clinical experiences suggest that the error in our estimate of the source offset is not this large. During the calculation of the technique charts, we used a number of models to predict the equivalent square field for rectangular fields. The most accurate model related the side of the equivalent square field with the square root of the area of the rectangular field (See Appendix A). This model predicts a larger change in equivalent square field, as the aspect ratio of the rectangular fields become greater, than other models (e.g., BJR 17) typically used to predict equivalent square fields for megavoltage beams. This suggests that only a component of the X-ray scatter generated by the patient contributes to the spurious signals generated in the EC film and that this component is more widely scattered than the total X-ray scatter. This results in larger scatter contributions from the edges of the field. We plan to perform a Monte Carlo study to examine this phenomenon more thoroughly. The results in Table 4 also suggest that the scatter is a complex phenomenon. The largest value in Table 4 does not occur for the largest phantom thickness but for a patient thickness of 35 cm. Therefore, X-ray scatter components generating signals in the digital cassette decrease more rapidly (or increase less rapidly) with increasing phantom thickness beyond 35 cm than the X-ray scatter components that generate signals in the ion chamber. Presumably this is because some of the low-energy scatter components are absorbed in the patient as the patient thickness increases. Thus, the volumes of the patient contributing to the highand low-energy scatter components differ. We can predict the irradiation required to optimally expose the EC film with a high degree of confidence for a large range of field sizes, patient thicknesses, and air gaps.
● P. MUNRO et al.
835
The accuracy is ⫾0.2 OD. This accuracy is significantly better than what is required clinically and is better than what can actually be achieved in practice, because of the inability of linear accelerators to deliver irradiations of fractional monitor units. A random check of 25 EC radiographs from our center for treatments in the pelvis and head and neck regions has shown a range of optical densities of 0.84 to 1.93. In our experience, EC film generates extremely highquality portal films—if exposed properly. Our ability to generate optimal exposures has improved the clinical utility of EC film. And although our results have been measured on Varian linear accelerators, there is every reason to expect that Tables 2 and 3 would apply to other accelerators. The large change in signals generated in the EC-L cassette/film is due to X-ray scatter from the patient. Apart from minor differences in X-ray spectrum between accelerators, which would change the penetration of the primary beam slightly and which would change X-ray scatter spectrum slightly, we would expect our results to apply to all accelerators. APPENDIX A: CALCULATION OF TECHNIQUE CHARTS The calculation of the technique charts uses the following equation: MUi ⫽ MUref ⴱ REC-L共FSexit, S, G兲/INV
(1)
where MUi is the irradiation in monitor units required to optimally expose the EC film; MUref is the irradiation in monitor units required to optimally expose the EC film at the reference conditions (field size ⫽ 13.5 ⫻ 13.5 [10 ⫻ 10 collimator setting], patient separation ⫽ 30.5 cm of water, and air gap ⫽ 9.4 cm); REC-L(FSexit, S, G) is the measured response of the EC-L film/cassette given in Tables 2 and 3; FSexit is the field size defined at the exit surface of the patient; S is the patient separation (thickness); G is the air gap between the back of the patient and the film cassette; and INV is the inverse square law correction that corrects for the difference in distance from the X-ray source between the current location of the patient and the location of the phantom that was used to generate Tables 2 and 3. The reader needs only to determine the three quantities in Eq. 1 to calculate the irradiation required to optimally expose the EC film. The quantity MUref can be measured by irradiating a 30.5-cm-thick water-equivalent phantom and an EC-L film/ cassette to different exposures while at the reference conditions, plotting the film/cassette response in a format similar to Fig. 2, selecting the desired optical density on the plot, and determining MUref from the plot. Calculation of the correct field size is necessary to select the correct value of REC-L (FSexit, S, G) from Tables 2 or 3. We match the field size defined at the exit surface of the patient with the field sizes in Tables 2 and 3, which represent the field sizes, defined at the exit surface of the phan-
836
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
冉
冊
SFDm ⫺ SSDoff SFDT ⫺ SSDoff
2
tom, that were used to measure the tables. Generally the patient will be positioned at a different source–axis distance than was used for the phantom measurements, so the divergence of the X-ray beams and hence the volume irradiated will be slightly different. However, the differences in divergence are quite small. Matching the field sizes at the exit surface, where the largest amount of scatter would be most likely to originate (14), seemed a priori the best approach to estimate the film response from the tables. The field size at the exit surface of the patient is calculated as follows:
where SSDoff is an offset to correct for the (empirically measured) apparent position of the X-ray source, SFDm is the source-to-film distance used to measure the tables with the source-to-exit distance fixed at 135 cm;
FSexit ⫽ FScoll共SAD ⫺ d ⫹ S兲
and SFDT is the treatment source-to-film distance, which is given as follows:
(2)
where FSexit is the field size used to search Tables 2 and 3, FScoll is the equivalent square for the collimator settings of the linear accelerator for the treatment field, SAD is the source–axis distance, d is the treatment depth, and S is the patient separation. Our empiric studies show that the most accurate method of estimating the length of the equivalent square for a rectangular field is to take the square root of the area of the field. FScoll ⫽ 共X ⴱ Y兲1/ 2
(3)
where X and Y are the field length and field width collimator setting on the treatment machine. We can calculate INV, the inverse square term, using the following formula:
INV ⫽
SFDm ⫽ 135 ⫹ G
SFDT ⫽ SAD ⫺ d ⫹ S ⫹ G
(4)
(5)
(6)
In summary, to calculate the irradiation that will optimally expose an EC film for their own clinical situations, readers need only determine MUref and SSDoff. The equivalent square field can be calculated using Eqs. 2 and 3. Then, knowing patient thickness, the correct value of REC-L (FSexit, S, G) can be interpolated from Table 2 or 3. The inverse square factor, INV, can be calculated using Eq. 4 if SSDoff is known. Then Eq. 1 is used to calculate the irradiation, MUi. The above equations have been added to our clinical monitor unit calculation software. Since most of the information needed is already entered for the monitor unit calculations, the only remaining parameters are S (patient separation) and G (air gap). Default values of S ⫽ 2 * d and G ⫽ 9.4 cm work well in our clinical environment.
REFERENCES 1. Dickerson RE, Haus AG, Huff KE. Development of a novel, high contrast, cassette/film/screen system for radiation therapy portal localization imaging. Proc SPIE 1997;3032:520 –529. 2. Munro P, Rawlinson JA, Fenster A. Therapy imaging: A signal-to-noise analysis of metal plate/film detectors. Med Phys 1987;14:975–984. 3. Radcliffe T, Barnea G, Wowk B, et al. Monte Carlo optimization of metal/phosphor screens at megavoltage energies. Med Phys 1993;20:1161–1169. 4. Munro P. Megavoltage radiography for treatment verification. In: Van Dyk J, editor. The modern technology of radiation oncology: A compendium for medical physicists and radiation oncologists. Madison, WI: Medical Physics Publishing; 1999. p. 481–508. 5. Haus AG, Dickerson RE, Huff KE, et al. Evaluation of a cassette-screen-film combination for radiation therapy portal localization imaging with improved contrast. Med Phys 1997; 24:1605–1608. 6. Lee PC, Glasgow GP. Technique charts for Kodak’s new film-screen systems for portal localization. Med Dosim 1998; 23:113–116. 7. Keys R, Marks JE, Haus A. Evaluation of a cassette-screenfilm combination with improved contrast for verification por-
8. 9. 10.
11.
12. 13.
14.
tal imaging during radiation therapy treatments (Abstr.). Med Phys 1998;25:A198. Marriage A. How many quanta? J Photogr Sci 1961;9:93–100. Farnell GC, Chanter JB. The quantum sensitivity of photographic emulsion grains. J Photogr Sci 1961;9:73– 83. Beddar AS, Mackie TR, Attix FH. Water-equivalent plastic scintillation detectors for high-energy beam dosimetry. II. Properties and measurements. Phys Med Biol 1992;37:1901– 1913. Beddar AS, Mackie TR, Attix FH. Water-equivalent plastic scintillation detectors for high-energy beam dosimetry. I. Physical characteristics and theoretical consideration. Phys Med Biol 1992;37:1883–1900. de Boer SF, Beddar AS, Rawlinson JA. Optical filtering and spectral measurements of radiation-induced light in plastic scintillation dosimetry. Phys Med Biol 1993;38:945–958. Arnfield MR, Gaballa HE, Zwicker RD, et al. Radiationinduced light in optical fibers and plastic scintillators: Application to brachytherapy dosimetry. IEEE Trans Nuc Sci 1996; 43:2077–2084. Sheikh-Bagheri D, Munro P. A Monte Carlo study of verification imaging in HDR brachytherapy. Med Phys 1997;25: 404 – 414.
PII S0360-3016(01)01564-4
PHYSICS CONTRIBUTION
TECHNIQUE CHARTS FOR EC FILM: DIRECT OPTICAL MEASUREMENTS TO ACCOUNT FOR THE EFFECTS OF X-RAY SCATTER PETER MUNRO, PH.D.,*† KEVIN JORDAN, PH.D.,*† CRAIG LEWIS, M.SC.,*†
AND
TIM HEEREMA*
*London Regional Cancer Centre, London, Ontario, Canada; †Department of Oncology, University of Western Ontario, London, Ontario, Canada Purpose: To develop a method of measuring technique charts for enhanced contrast (EC) film, to demonstrate how X-ray scatter changes the response of EC film, and to generate technique charts for general use. Methods and Materials: We have developed a “digital cassette”— consisting of a metal plate/phosphor screen, a light guide, a photodiode sensor, and an electrometer—that can be used to measure the light generated in the phosphor screen of the film cassette. In turn, these measurements can be used to generate technique charts for EC film. The digital cassette has been used to measure technique charts for 4-MV and 6-MV X-ray beams for a variety of different phantom thicknesses, field sizes, and phantom-to-cassette air gaps. Results and Discussion: We have observed that the signals generated in an ionization chamber located 9.4 cm behind a 30-cm-thick water-equivalent phantom increase by a factor of 1.9 when the field size is increased from 4 ⴛ 4 cm2 to 40 ⴛ 40 cm2 when irradiated by a 6-MV X-ray beam. However, the change in EC film response is a factor of 3.5 under the same conditions. Irradiations to optimally expose the EC film predicted by the digital cassette differ by up to 82% compared to those predicted by ion chamber measurements. Nevertheless, the technique charts measured using the digital cassette predict the response of the EC film to ⴞ 0.2 optical density. The overresponse of the EC film is most likely due to low-energy scattered photons, which interact with the high atomic number (Z ⴝ 64) phosphor screen of the enhanced contrast localization cassette. Therefore, simple solutions, such as placing a high atomic number material above the enhanced contrast localization cassette, can reduce this contribution by scattered photons to the signal generated in the cassettes. Conclusions: We have developed a digital cassette that can make more accurate measurements of the technique charts for EC films. Our measurements show that under some conditions, X-ray scatter can generate a large fraction of the signals recorded by the EC film. Technique charts have been generated at 4 MV and 6 MV, and these charts should have universal applicability. © 2001 Elsevier Science Inc. X-ray scatter, Portal film, EC film.
plate (3, 4). Finally, the film has a gamma (i.e., display contrast), which is 3.5 times larger than the conventional metal plate/film combinations (1, 5). This high film gamma, which is the result of a narrow distribution of grain sizes and metal ion doping of the microcubic film grains, improves the display of the low-contrast structures found in megavoltage radiography and is often the main advantage cited for the use of EC film (1). One complication is that the high gamma limits the latitude of the EC film, requiring that users make accurate estimates of the irradiation necessary to optimally expose the film. As a result, methods for generating technique charts (tables of monitor unit settings as a function of patient thickness and field size) have been described in the literature (6). The method of Lee and Glasgow uses ionization chamber measurements (e.g., measurements of tissue-
INTRODUCTION In 1996, Kodak introduced a new film-screen combination system for megavoltage radiography. The system consists of an enhanced contrast (EC) film, and new localization (EC-L) and verification (EC-V) cassettes. The EC-L cassette is constructed using a 1-mm copper front plate and a Lanex Fast gadolinium oxysulfide screen pair. This filmscreen system has three advantages compared to the conventional metal plate/film combinations used in the past. The film granularity for the new EC film is much lower than that for conventional portal films because of the use of a microcubic film grain structure (1, 2). Also, the use of a phosphor screen increases the quantum efficiency of the EC-L cassettes, because X-ray quanta can interact directly in the phosphor screen as well as interacting in the metal
search) and Cancer Care Ontario is greatly appreciated. Acknowledgments—We thank Matt Mulligan, Tim Van Dijk, and Howard Siu for their help with some of the measurements with the ion chamber and the digital cassette. Accepted for publication 14 March 2001.
Reprint requests to: Peter Munro, London Regional Cancer Centre, 790 Commissioners Road East, London, Ontario N6A 4L6, Canada. Tel: (519) 685-8500, ext. 53317; Fax: (519) 6858658; E-mail: [email protected] Financial assistance from the Medical Research Council of Canada (now known as the Canadian Institutes of Health Re829
830
I. J. Radiation Oncology
●
Biology
●
Physics
Fig. 1. The “digital cassette” formed from a conventional film/ screen cassette, a 1-mm copper plate, a Lanex Fast (back) Gd2O2S:Tb phosphor screen, a liquid light guide, a photodiode, and an electrometer. The photodiode is shielded in a cylindrical lead container located beside the cassette.
maximum ratios, tissue-air ratios, field size factors, etc.) to generate values from which technique charts are calculated (6). However, our experience has shown that the technique charts generated in this way are not universally usable; small field sizes result in films that are too light, and large field sizes result in films that are too dark. The results of Keys et al. also show that optimum exposure is difficult to predict (7). Keys et al. found that adjusting the air gap between an EC-V cassette and the patient—to control the dose reaching the cassette during verification imaging— resulted in an inverse cube relationship, rather than the expected inverse square relationship (7). In this paper we describe a “digital cassette,” which consists of a metal plate/phosphor screen, a light guide, a photodiode sensor, and an electrometer. Using this cassette we can directly measure the amount of light emitted by the phosphor screen during an irradiation. Knowing the amount of light needed to optimally expose the EC film, we can use the measurements from the digital cassette to predict the irradiation that will optimally expose the EC film under a wide variety of conditions (e.g., field size, patient thickness, air gap between patient and cassette). We also present evidence that shows that low-energy scattered photons are the most likely cause of the discrepancy in response between the ion chamber measurements and the EC-L cassette.
METHODS AND MATERIALS Digital cassette The digital cassette (Fig. 1) was constructed using a standard kilovoltage film cassette from which the phosphor screens had been removed. The phosphor screens were replaced by a 1.0-mm copper plate in contact with a single Lanex Fast back screen. The gadolinium oxysulfide phosphor loading of this detector (134 mg/cm2) was only slightly less than that of the standard EC-L cassette (168 mg/cm2).
Volume 50, Number 3, 2001
A Series 800 liquid light guide, with a core diameter of 5 mm and a length of 1.5 m (Lumatec GMBH, Munich, Germany), was attached to the lid of the cassette using a mechanical coupling that had been fabricated locally. When the lid was closed, the light guide was in intimate contact with the phosphor screen. Another coupling was used to hold a series 5T photodiode, with an active area of 100 mm2 (Centronic, Newbury Park, CA) in contact with the other end of the light guide. The light guide allowed the photodiode to be positioned well outside the X-ray beam of the accelerator while measuring the light emitted from the phosphor screen. Optical filters (an infrared filter and a Wratten 15 filter) were placed between the light guide and the photodiode to limit the wavelengths of light (500 – 650 nm) reaching the photodiode. Since our measurements showed that direct interaction of X-ray radiation with the photodiode could generate large signals, the photodiode was shielded in a lead cylinder with a wall thickness of 3 cm. Additional lead blocks 5-cm thick surrounded the cylinder to ensure a high degree of shielding. A 30-m coax cable connected the photodiode to a Keithley Model 35617EBS electrometer, which was located outside the treatment room. Irradiations Irradiations used Varian 2100 C/D dual-energy linear accelerators, with either 6/18 or 4/10 X-ray energies. All measurements were made using a 6-MV X-ray beam, except for the results presented in one table, where measurements were made using a 4-MV X-ray beam. Lucite sheets 50 ⫻ 70 cm2 ⫻ ⬃1 cm thick were used to generate a phantom that varied in thickness between 8.5 and 52.5 cm. The Lucite sheets were placed on the treatment couch, and the cassette was placed either in the cassette holder of the couch (9.4 cm air gap) or on a mechanical stage that allowed the air gap to be adjusted to either 15 or 20 cm. The couch was not moved as the Lucite sheets were added, so that the source-to-exit surface of the phantom remained fixed at 135 cm. Collimator settings ranged between 4 ⫻ 4 and 40 ⫻ 40 cm2, which corresponded to field sizes between 5.4 ⫻ 5.4 and 54 ⫻ 54 cm2 when defined at the exit plane of the phantom. Care had to be taken to eliminate spurious signals generated in the digital cassette. The digital cassette was irradiated with an opaque material located between the phosphor screen and the liquid light guide. With the photodiode well outside the treatment beam and the photodiode shielded by the lead cylinder and lead blocks, a “nonphosphor” signal ⬃10% of the total optical signal was generated. Use of Wratten 15 and infrared filters reduced the “nonphosphor” signal by ⬃80% and reduced the optical signal to 60% of its original magnitude. Thus, after optical filtering, the “nonphosphor” signal was only ⬃3% of the optical signal. The Wratten 15 filter, which removes short wavelength light, and the IR filter, which removes long wavelength light, contributed equally to the reduction in the “nonphosphor” signal. Measurements showed that the “nonphosphor” signal changed from ⬃3% to ⬃4% of the total
EC film, technique charts, and X-ray scatter
signal as the field size changed. Thus, to simplify data acquisition and data correction, the small, almost constant, contribution of the “nonphosphor” signal was ignored. Another problem was that the measured signal decreased with increasing accumulated dose received by the digital cassette. The radiation created yellow products in the liquid light guide, thereby changing its transmission properties. To account for these changes, which could be as large as 7% for a 3000 monitor unit irradiation, measurements were repeated at standard field sizes and phantom thicknesses after every 5–7 irradiations. The raw readings were then corrected for any changes between the two standard measurements, assuming that the optical transmission of the light guide changed linearly with radiation dose. To minimize the total dose delivered to the digital cassette, all readings were acquired using 25 monitor unit irradiations. Thus, a technique chart (for one phantom to cassette air gap) could be measured using a total irradiation of only 3000 monitor units. Despite the small irradiations, repeatability was ⫾0.3%, and the signals were quite large, ranging between 4 and 65 nC. Receptor response and scatter measurements The response of an ion chamber, the digital cassette, and the EC-L film/cassette to a 6-MV radiation beam was measured. The ion chamber, a Capintec PR06C air-equivalent ion chamber connected to a Capintec 292 electrometer, was placed behind a 27.5-cm-thick Lucite phantom, which was sitting on the treatment couch with the couch top at 135 cm. The ion chamber was placed at a depth of 1.5 cm inside a 3-cm-thick Lucite plate to ensure that the chamber was at a depth of full buildup. With the air gap set to 9.4 cm, the signals from the ion chamber were recorded as the collimator settings varied between 4 ⫻ 4 cm2 and 40 ⫻ 40 cm2. The digital cassette was placed in the cassette holder of the treatment couch (air gap 9.4 cm), and the measurements for the same field sizes were repeated. Finally, the EC-L film/ cassette was placed in the cassette holder of the couch, and the monitor unit settings were adjusted with field size so that the films were all exposed to the same net optical density (OD) of ⬃1.5 OD. The ratio of monitor unit settings for each field size was taken to represent the relative response of the EC-L film/cassette. If the optical densities on the films were not exactly the same, the above ratio was corrected assuming a linear film response. This was a reasonable assumption because of the small differences between the two optical densities and because the optical densities were in the linear region of the film response curve. A 1.5-mm tungsten plate was used to measure the fraction of the signals generated in the film cassettes by low-energy photons. This plate was placed above the film cassettes and the ion chamber to absorb (i.e., filter) the low-energy components from the beam. The experiments described in the previous paragraph were then repeated to measure how the presence of the tungsten plate changed the response of the image receptors. The D vs. log (E) (H&D) curves of the EC film were also
● P. MUNRO et al.
831
measured using the phantom setup described above. The dose at the plane of the cassette for 4 ⫻ 4 cm2, 10 ⫻ 10 cm2, and 40 ⫻ 40 cm2 collimator settings and various monitor unit settings was measured using the ion chamber. Films were exposed in the EC-L cassette using the same irradiations and were processed in a Kodak M35 processor using Kodak RP chemicals. The resulting optical density, which was measured using an X-rite Model 301 densitometer, was then plotted vs. log (dose) to yield the desired curves. Effective source location The measurements using the digital cassette were made with the phantom at extended distances, whereas most patients will not be treated at extended distances. Therefore, the effect on the measured signal of changing the position of the phantom, while keeping the air gap constant, needed to be determined. The phantom was moved over a range of 30 cm, with the position of the couch top changing between 135 and 105 cm SSD. Measurements were made for “nominal” collimator settings of 4 ⫻ 4 cm2 and 40 ⫻ 40 cm2 and Lucite thicknesses of 18.5 and 42 cm. The collimator settings were only “nominal” because, as the position of the phantom changed, the collimator setting was adjusted so that the same area, at the position of the treatment couch, was irradiated. The effective source location was then estimated from a plot of the square root of normalized dose vs. distance. A source offset was calculated from these measurements, which was used in the calculations of the technique charts (See Appendix A). Tests of the technique charts The technique charts were calculated using the approach described in Appendix A. The technique charts were tested by exposing EC-L film/cassettes under a variety of different fields sizes (square and rectangular) and patient thicknesses. If the predicted irradiation was not an integer value of monitor units, integer monitor unit irradiations greater than and less than the predicted value were used to expose two films. The optical density from both films was then used to estimate the optical density that would have resulted had a fractional monitor unit irradiation been possible, assuming a linear response between the two exposed films. RESULTS The D vs. log (E) curves are shown in Fig. 2 for a 27.5-cm-thick Lucite phantom (⬃30 cm water equivalent) with a 9.4-cm air gap between the phantom and the cassette irradiated by a 6-MV X-ray beam. The figure shows the effect of changing the collimator setting on the response of the EC film. The abscissa represents the dose reaching the plane of the image receptor as measured using the PR-06 ion chamber. The dose as measured by the ion chamber does not predict the optical density generated on the EC film. The number of developable film grains depends on the number of optical quanta reaching the film (8, 9), which in turn depends on the dose (energy) deposited in the EC-L cas-
832
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
Table 1. Effective source location Patient separation (cm)
FS (cm)
Source displacement (cm)
20 20 47 47
4⫻4 40 ⫻ 40 4⫻4 40 ⫻ 40
8 ⫺1 23 46
The apparent position of the X-ray source (SSDoff) as a function of field size and patient thickness. To simplify our calculations, SSDoff was set to 15 cm for all patients thicknesses and field sizes.
Fig. 2. The optical density of the EC film as a function of dose, where the dose is measured using an ion chamber. The EC film appears to need less dose to reach the same optical density as the collimator setting increases.
sette. Clearly, the dose measured by the ion chamber must not represent the actual dose deposited in the screen of the EC-L film/cassette. An explanation for the discrepancy between the response of the EC-L film/cassette and the dose measured by the ion chamber is shown in Fig. 3. Figure 3 (a) shows the signals generated in the ion chamber and the EC-L film/cassette as a function of collimator setting, with and without the presence of the 1.5-mm tungsten plate, when a 27.5-cm Lucite phantom is irradiated by a 6-MV beam. The presence of the tungsten plate does not change the response of the PR06 ion
chamber greatly; however, there is a large change in the response of the EC-L film/cassette. Indeed, after tungsten filtration, the EC-L film/cassette responds to the X-ray beam in a similar fashion as the ion chamber. Clearly, the tungsten plate is able to remove from the X-ray beam those components that are causing the overresponse of the EC-L film/ cassette, presumably low-energy scattered photons generated in the phantom. Figure 3 (b) shows the response of the digital cassette to the 6-MV beam. The response is very similar to that of the EC-L film/cassette, showing that the digital cassette is an excellent alternative for measuring technique charts for the EC-L film/cassette. The results for the measurement of the effective source location are shown in Table 1. In general, the change in signal as the Lucite phantom and image receptor are moved closer to the X-ray source is not as great as would be predicted by the inverse square relationship. The source appears to be farther away than the X-ray target. To calculate the correct change in signal using the inverse square formalism (See Appendix A), a source offset
Fig. 3. Panel (a) shows the effect of a 1.5-mm tungsten plate on the signals generated in the EC film and the ion chamber. The dotted and solid curves show the signals measured by the ion chamber with and without the tungsten plate, respectively. The solid circles and solid squares show the signals measured by the EC film with and without the tungsten plate, respectively. The tungsten plate appears to absorb a component of the beam that contributes substantially to the signals recorded by the EC film but that contributes relatively little to the signals generated in the ion chamber. Panel (b) shows the response of the “digital cassette” compared to that of the EC film and the ion chamber (without a tungsten plate). Irradiations used a 6-MV X-ray beam and a 27.5-cm Lucite phantom.
EC film, technique charts, and X-ray scatter
● P. MUNRO et al.
833
Table 3. Technique charts— 4 MV
Table 2. Technique charts— 6 MV 9.4-cm air gap
9.4-cm air gap
Sep\FS
5.4
13.5
27
40.5
54
Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.512 0.844 1.053 1.347 1.671 2.077 2.723 3.410
0.407 0.616 0.752 1.000 1.209 1.431 1.886 2.313
0.311 0.419 0.494 0.626 0.749 0.894 1.160 1.433
0.260 0.324 0.373 0.462 0.545 0.651 0.843 1.035
0.230 0.278 0.315 0.385 0.451 0.536 0.687 0.846
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.449 0.874 1.104 1.593 2.059 2.683 3.745 4.841
0.332 0.574 0.730 1.000 1.276 1.647 2.266 2.892
0.231 0.342 0.415 0.552 0.696 0.882 1.218 1.524
0.184 0.250 0.297 0.384 0.481 0.601 0.822 1.035
0.163 0.211 0.248 0.315 0.387 0.487 0.656 0.830
15-cm air gap
15-cm air gap
Sep\FS
5.4
13.5
27
40.5
54
Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.549 0.915 1.131 1.456 1.797 2.183 2.845 3.446
0.440 0.686 0.831 1.053 1.285 1.547 2.006 2.418
0.332 0.461 0.547 0.676 0.815 0.974 1.250 1.512
0.271 0.347 0.406 0.494 0.592 0.705 0.901 1.085
0.236 0.289 0.336 0.405 0.484 0.573 0.731 0.881
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.496 0.965 1.219 1.759 2.274 2.962 4.135 5.346
0.367 0.634 0.806 1.104 1.409 1.819 2.502 3.194
0.255 0.378 0.458 0.610 0.769 0.974 1.345 1.683
0.203 0.276 0.328 0.424 0.531 0.664 0.907 1.143
0.180 0.233 0.273 0.348 0.427 0.537 0.724 0.917
20-cm air gap
20-cm air gap
Sep\FS
5.4
13.5
27
40.5
54
Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.594 1.002 1.253 1.624 1.998 2.418 3.192 3.853
0.483 0.777 0.951 1.205 1.472 1.764 2.305 2.778
0.369 0.538 0.640 0.796 0.955 1.136 1.472 1.760
0.299 0.407 0.473 0.582 0.692 0.822 1.062 1.270
0.255 0.337 0.388 0.473 0.559 0.664 0.855 1.021
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.577 1.123 1.419 2.048 2.647 3.449 4.814 6.224
0.427 0.738 0.939 1.286 1.641 2.117 2.913 3.718
0.296 0.440 0.534 0.710 0.895 1.134 1.566 1.959
0.237 0.321 0.382 0.494 0.619 0.772 1.056 1.330
0.209 0.271 0.318 0.405 0.497 0.626 0.843 1.067
The relative response of the digital cassette as a function of field size (FS), patient thickness (SEP), and air gap for a 6-MV X-ray beam. The values in the table represent the irradiation necessary to give the same optical density on an EC film irradiated in an EC-L cassette. The field size, which is given as the side of the square field, is defined at the exit surface of the phantom at 135 cm from the source. All values have been normalized to a 13.5 ⫻ 13.5 field (10 ⫻ 10 collimator setting), patient thickness of 30.5 cm (water equivalent), and an air gap of 9.4 cm.
The relative response of the digital cassette as a function of field size (FS), patient thickness (SEP), and air gap for a 4-MV X-ray beam. The values in the table represent the irradiation necessary to give the same optical density on an EC film irradiated in an E C-L cassette. The field size, which is given as the side of the square field, is defined at the exit surface of the phantom at 135 cm from the source. All values have been normalized to a 13.5 ⫻ 13.5 field (10 ⫻ 10 collimator setting), patient thickness of 30.5 cm (water equivalent), and an air gap of 9.4 cm.
has been introduced into the inverse square calculation. For simplicity, we chose a single value of 15 cm, even though the source offset ranged from 0 to 45 cm, depending on field size and phantom thickness. This value was obtained by averaging all the numbers in Table 1 and rounding off the average value to the lower multiple of 5 cm. Tables 2 and 3 give the technique charts for the EC-L film/cassette for 4-MV and 6-MV X-ray beams. All the measurements have been normalized to the 13.5 ⫻ 13.5 cm2 field size (10 ⫻ 10 cm2 collimator setting), the 30.5 cm (27.5 cm Lucite) phantom thickness, and an air gap of 9.4 cm. The values in the tables are the relative exposures required to generate the same optical density on the film. Therefore, to use these tables in his own clinical practice, the user needs only to determine the irradiation (in monitor
units) to properly expose the EC-L film/cassette at the reference conditions. Table 4 gives the ratio of exposures predicted by ion chamber measurements divided by the exposures predicted by the digital cassette. (Note that the ion chamber measurements have been made under identical conditions to those used for the digital cassette [i.e., 6-MV X-ray beam, 9.4-cm air gap, the same field sizes, and phantom thicknesses]. Also note that both sets of data have been normalized to 13.5 ⫻ 13.5 cm2 field size [10 ⫻ 10 cm2 collimator setting] and the 30.5-cm phantom thickness before the ratios have been calculated.) At large field sizes, the ion chamber measurements predict that the irradiation required for optimal exposure is up to 82% larger (54 ⫻ 54 cm2; 35-cm separation) than exposures predicted by the digital cassette. At small field sizes, the ion chamber measurements predict an irra-
834
I. J. Radiation Oncology
●
Biology
●
Physics
Table 4. Ion chamber over- and under-response
Volume 50, Number 3, 2001
thickness used for the measurements. Figure 4 shows that we are able to predict the optical density of the exposed films to ⫾0.2 OD.
9.4-cm air gap Sep\FS
5.4
13.5
27
40.5
54
9.5 20.0 24.5 30.5 35.0 40.0 47.0 52.0
0.95 0.95 0.94 0.94 0.97 0.91 0.92 0.85
1.05 1.06 1.05 1.00 1.04 1.02 1.02 0.97
1.22 1.30 1.32 1.28 1.33 1.28 1.28 1.20
1.38 1.54 1.57 1.55 1.61 1.55 1.53 1.44
1.51 1.71 1.76 1.75 1.82 1.75 1.73 1.62
The relative response of the ion chamber divided by that of the digital cassette, when both sets of data have been normalized to the reference conditions. When the numbers in the table are greater than 1.0, they indicate that the ion chamber predicts a larger irradiation than is necessary to properly expose the EC film.
diation that is 15% smaller (5.4 ⫻ 5.4 cm2; 52 cm separation). These results are consistent with our clinical observations when we use technique charts generated by ion chambers: small field sizes result in light films, and large field sizes result in dark films. One can generate the results in Table 4 by taking the ratio of the ion chamber and digital cassette curves in Fig. 3. However, the normalization is for a 13.5 ⫻ 13.5 cm2 field size (10 ⫻ 10 cm2 collimator setting) in Table 4, and this is why some values are less than 1.0. The accuracy of our technique charts in predicting the optical density on irradiated EC films is shown in Fig. 4. All films were exposed with an air gap of 9.4 cm. The height of the bars represent the optical density on the film, while the labels for each bar indicate the field size and phantom
Fig. 4. The accuracy of the technique charts when used to expose EC film. The goal was to obtain an optical density of 1.5.
DISCUSSION AND CONCLUSIONS Our results suggest that X-ray scatter from the patient can generate a large fraction of the signal recorded on EC film, depending upon field size, patient thickness, and air gap. This X-ray scatter complicates the measurements of technique charts for EC film, because the X-ray scatter does not generate as large a signal in an ion chamber as in the film/cassette. As a result, we have had to develop a method to directly measure the light generated by a Gd2O2S screen when irradiated by megavoltage X-ray beams. We found that the largest differences between the response of the ion chamber and that of the EC film occur as a function of field size rather than as a function of depth. A simple method to account for the effects of X-ray scatter would be to make measurements using ion chambers, but use different film response curves (e.g., Fig. 2) for each field size. In other words, use different “doses” (as measured by the ion chamber) at different field sizes to optimally expose the EC film. Whereas such an approach would not be as accurate as our direct measurements, the approach might generate clinically usable technique charts with equipment widely available in radiation therapy departments. Certainly, use of different film response curves would be considerably more accurate than using one response curve. Although our measurement approach worked well, there are some possible improvements. One of the difficulties with our experimental procedure was the change in transmission of the liquid light guide. Over the course of many measurements, the transmission dropped by a factor of five, making the later measurements more difficult because of the smaller signals generated in the photodiode. An improvement would be to use a quartz rod, or fiber optic materials (such as those used in scintillation detectors), to transport the light outside of the radiation beam (10, 11). These materials are less susceptible to radiation coloring than the liquid light guide and thus should result in a more stable measuring tool. Another difficulty was the “nonphosphor” signals (probably mostly Cerenkov radiation [12, 13]) generated in the liquid light guide. Our unpublished calculations suggest that given the 1/3 dependence of Cerenkov radiation intensity, the transmission of the light guide, and the response of the silicon photodiode, the “nonphosphor” signal is likely to have a uniform value at all wavelengths (12, 13). (Silicon photodiodes have very high optical sensitivities at red and infrared wavelengths, so even a small number of optical photons at these wavelengths can result in large signals being generated in the photodiode.) We used readily available optical filters to create an optical passband of 500 to 650 nm to minimize the spurious optical signals reaching our photodiode. An improvement would be to use an optical interference filter with a narrower passband centered on the main emission peak of Gd2O2S:Tb located at
EC film, technique charts, and X-ray scatter
545 nm. The main emission peak generates 54% of the optical signal from the Gd2O2S:Tb phosphor, and it can be completely transmitted by a filter with a 20-nm passband. In principle, such a filter would reduce our signal to ⬃54% of its original value (the current filters currently remove some of the signal from the main emission peak) but might reduce the “nonphosphor” signals by the ratio of the passband widths (150/20 ⫽ 7.5). This would make the “nonphosphor” signals even less of a concern. The effective source location measurements in Table 1 appear to have some uncertainties, since we would expect the source location displacement to be larger for larger field sizes where X-ray scatter is more important. However, for the 20-cm-thick phantom, this trend does not occur. The measurements were made when the transmission of the light pipe was quite low. So we believe that experimental uncertainty is the cause of the unpredictable trends. This is why we decided on one value for the source offset. Despite the assumption of a single source offset our table performs well. Let us assume that we are in error in our estimate of the source offset by 10 cm. This would make an error in our estimates of dose of 1052/1152 ⫽ 0.834. Using Fig. 2, such a change in dose for the high-gradient region of the film would result in a change in optical density of 0.55. However, our clinical experiences suggest that the error in our estimate of the source offset is not this large. During the calculation of the technique charts, we used a number of models to predict the equivalent square field for rectangular fields. The most accurate model related the side of the equivalent square field with the square root of the area of the rectangular field (See Appendix A). This model predicts a larger change in equivalent square field, as the aspect ratio of the rectangular fields become greater, than other models (e.g., BJR 17) typically used to predict equivalent square fields for megavoltage beams. This suggests that only a component of the X-ray scatter generated by the patient contributes to the spurious signals generated in the EC film and that this component is more widely scattered than the total X-ray scatter. This results in larger scatter contributions from the edges of the field. We plan to perform a Monte Carlo study to examine this phenomenon more thoroughly. The results in Table 4 also suggest that the scatter is a complex phenomenon. The largest value in Table 4 does not occur for the largest phantom thickness but for a patient thickness of 35 cm. Therefore, X-ray scatter components generating signals in the digital cassette decrease more rapidly (or increase less rapidly) with increasing phantom thickness beyond 35 cm than the X-ray scatter components that generate signals in the ion chamber. Presumably this is because some of the low-energy scatter components are absorbed in the patient as the patient thickness increases. Thus, the volumes of the patient contributing to the highand low-energy scatter components differ. We can predict the irradiation required to optimally expose the EC film with a high degree of confidence for a large range of field sizes, patient thicknesses, and air gaps.
● P. MUNRO et al.
835
The accuracy is ⫾0.2 OD. This accuracy is significantly better than what is required clinically and is better than what can actually be achieved in practice, because of the inability of linear accelerators to deliver irradiations of fractional monitor units. A random check of 25 EC radiographs from our center for treatments in the pelvis and head and neck regions has shown a range of optical densities of 0.84 to 1.93. In our experience, EC film generates extremely highquality portal films—if exposed properly. Our ability to generate optimal exposures has improved the clinical utility of EC film. And although our results have been measured on Varian linear accelerators, there is every reason to expect that Tables 2 and 3 would apply to other accelerators. The large change in signals generated in the EC-L cassette/film is due to X-ray scatter from the patient. Apart from minor differences in X-ray spectrum between accelerators, which would change the penetration of the primary beam slightly and which would change X-ray scatter spectrum slightly, we would expect our results to apply to all accelerators. APPENDIX A: CALCULATION OF TECHNIQUE CHARTS The calculation of the technique charts uses the following equation: MUi ⫽ MUref ⴱ REC-L共FSexit, S, G兲/INV
(1)
where MUi is the irradiation in monitor units required to optimally expose the EC film; MUref is the irradiation in monitor units required to optimally expose the EC film at the reference conditions (field size ⫽ 13.5 ⫻ 13.5 [10 ⫻ 10 collimator setting], patient separation ⫽ 30.5 cm of water, and air gap ⫽ 9.4 cm); REC-L(FSexit, S, G) is the measured response of the EC-L film/cassette given in Tables 2 and 3; FSexit is the field size defined at the exit surface of the patient; S is the patient separation (thickness); G is the air gap between the back of the patient and the film cassette; and INV is the inverse square law correction that corrects for the difference in distance from the X-ray source between the current location of the patient and the location of the phantom that was used to generate Tables 2 and 3. The reader needs only to determine the three quantities in Eq. 1 to calculate the irradiation required to optimally expose the EC film. The quantity MUref can be measured by irradiating a 30.5-cm-thick water-equivalent phantom and an EC-L film/ cassette to different exposures while at the reference conditions, plotting the film/cassette response in a format similar to Fig. 2, selecting the desired optical density on the plot, and determining MUref from the plot. Calculation of the correct field size is necessary to select the correct value of REC-L (FSexit, S, G) from Tables 2 or 3. We match the field size defined at the exit surface of the patient with the field sizes in Tables 2 and 3, which represent the field sizes, defined at the exit surface of the phan-
836
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
冉
冊
SFDm ⫺ SSDoff SFDT ⫺ SSDoff
2
tom, that were used to measure the tables. Generally the patient will be positioned at a different source–axis distance than was used for the phantom measurements, so the divergence of the X-ray beams and hence the volume irradiated will be slightly different. However, the differences in divergence are quite small. Matching the field sizes at the exit surface, where the largest amount of scatter would be most likely to originate (14), seemed a priori the best approach to estimate the film response from the tables. The field size at the exit surface of the patient is calculated as follows:
where SSDoff is an offset to correct for the (empirically measured) apparent position of the X-ray source, SFDm is the source-to-film distance used to measure the tables with the source-to-exit distance fixed at 135 cm;
FSexit ⫽ FScoll共SAD ⫺ d ⫹ S兲
and SFDT is the treatment source-to-film distance, which is given as follows:
(2)
where FSexit is the field size used to search Tables 2 and 3, FScoll is the equivalent square for the collimator settings of the linear accelerator for the treatment field, SAD is the source–axis distance, d is the treatment depth, and S is the patient separation. Our empiric studies show that the most accurate method of estimating the length of the equivalent square for a rectangular field is to take the square root of the area of the field. FScoll ⫽ 共X ⴱ Y兲1/ 2
(3)
where X and Y are the field length and field width collimator setting on the treatment machine. We can calculate INV, the inverse square term, using the following formula:
INV ⫽
SFDm ⫽ 135 ⫹ G
SFDT ⫽ SAD ⫺ d ⫹ S ⫹ G
(4)
(5)
(6)
In summary, to calculate the irradiation that will optimally expose an EC film for their own clinical situations, readers need only determine MUref and SSDoff. The equivalent square field can be calculated using Eqs. 2 and 3. Then, knowing patient thickness, the correct value of REC-L (FSexit, S, G) can be interpolated from Table 2 or 3. The inverse square factor, INV, can be calculated using Eq. 4 if SSDoff is known. Then Eq. 1 is used to calculate the irradiation, MUi. The above equations have been added to our clinical monitor unit calculation software. Since most of the information needed is already entered for the monitor unit calculations, the only remaining parameters are S (patient separation) and G (air gap). Default values of S ⫽ 2 * d and G ⫽ 9.4 cm work well in our clinical environment.
REFERENCES 1. Dickerson RE, Haus AG, Huff KE. Development of a novel, high contrast, cassette/film/screen system for radiation therapy portal localization imaging. Proc SPIE 1997;3032:520 –529. 2. Munro P, Rawlinson JA, Fenster A. Therapy imaging: A signal-to-noise analysis of metal plate/film detectors. Med Phys 1987;14:975–984. 3. Radcliffe T, Barnea G, Wowk B, et al. Monte Carlo optimization of metal/phosphor screens at megavoltage energies. Med Phys 1993;20:1161–1169. 4. Munro P. Megavoltage radiography for treatment verification. In: Van Dyk J, editor. The modern technology of radiation oncology: A compendium for medical physicists and radiation oncologists. Madison, WI: Medical Physics Publishing; 1999. p. 481–508. 5. Haus AG, Dickerson RE, Huff KE, et al. Evaluation of a cassette-screen-film combination for radiation therapy portal localization imaging with improved contrast. Med Phys 1997; 24:1605–1608. 6. Lee PC, Glasgow GP. Technique charts for Kodak’s new film-screen systems for portal localization. Med Dosim 1998; 23:113–116. 7. Keys R, Marks JE, Haus A. Evaluation of a cassette-screenfilm combination with improved contrast for verification por-
8. 9. 10.
11.
12. 13.
14.
tal imaging during radiation therapy treatments (Abstr.). Med Phys 1998;25:A198. Marriage A. How many quanta? J Photogr Sci 1961;9:93–100. Farnell GC, Chanter JB. The quantum sensitivity of photographic emulsion grains. J Photogr Sci 1961;9:73– 83. Beddar AS, Mackie TR, Attix FH. Water-equivalent plastic scintillation detectors for high-energy beam dosimetry. II. Properties and measurements. Phys Med Biol 1992;37:1901– 1913. Beddar AS, Mackie TR, Attix FH. Water-equivalent plastic scintillation detectors for high-energy beam dosimetry. I. Physical characteristics and theoretical consideration. Phys Med Biol 1992;37:1883–1900. de Boer SF, Beddar AS, Rawlinson JA. Optical filtering and spectral measurements of radiation-induced light in plastic scintillation dosimetry. Phys Med Biol 1993;38:945–958. Arnfield MR, Gaballa HE, Zwicker RD, et al. Radiationinduced light in optical fibers and plastic scintillators: Application to brachytherapy dosimetry. IEEE Trans Nuc Sci 1996; 43:2077–2084. Sheikh-Bagheri D, Munro P. A Monte Carlo study of verification imaging in HDR brachytherapy. Med Phys 1997;25: 404 – 414.