- Email: [email protected]
Are port films reliable for in vivo exit dose measurements?
Radiotherapy and Oncology, 25 (1992) 67-72 © 1992 Elsevier Science Publishers B.V. All rights reserved. 0167-8140/92/$05.00
67
RADION 01037
Short Communication
Are port films reliable for in vivo exit dose measurements? Jan Van Dam, Catherine Vaerman, Nino Blanckaert, Godelieve Leunens, Andr6e Dutreix and Emmanuel van der Schueren Department of Radiotherapy, U.Z. St. Rafael, Leuven, Belgium (Received 27 August 1991, revision received 7 May 1992, accepted 21 May 1992)
Key words: In vivo dosimetry; Port film; Quality assurance
Summary The possibility of using conventional port films as an in vivo method for obtaining relative exit dose distribution in patients is evaluated. Kodak "Verification" films in "Localization" cassettes are positioned in the usual clinical conditions behind an homogeneous polystyrene phantom as well as behind a phantom containing air, wood and aluminium inhomogeneities. Taking beam divergency into account the densitometric profiles are projected back to the exit side of the phantom. They are compared to the profiles obtained with an ionization chamber used under full backscatter conditions. The agreement between the off-axis ratios determined with either method are mostly better than 2 % and never exceed 5 %. These phantom measurements are completed by a comparison between off-axis ratios determined on a port film for a head and neck patient and those obtained by diode dosimetry applied on the patient at the exit side of the beam. A similar agreement as between film and ion chamber on the phantoms is obtained. These encouraging results illustrate the possibilities of using conventional port films for in vivo dosimetry.
Introduction The use of port films in radiotherapy is a well-established method for quality assurance. Their standard use is of a rather qualitative nature and consists mainly of visual evaluation of the field set-up. Quantitative evaluation of set-up can be made by measurement of the distances between the anatomical structures (bones, lung or air cavities) and the field edge, corrected for image magnification [5,8]. On the other hand, films are also used for phantom dosimetry for several decades. This method had indeed been shown [3] to have unique possibilities for relative dosimetry, provided the necessary precautions are taken: densitometric measurements of the films allow analysis of a complete dose distribution in the film plane with a short film exposure requiring the immobilization of the treatment unit for only a short time. From that respect, the use of port films for obtaining relative dose distributions seems only to be a logical extension of their standard qualitative use. Ying et al. [ 14] propose to use the dose distributions at the level of the port film to correct the CT data used for treatment planning purposes.
An alternative approach is to use the port film data for calculation of exit dose distributions at the level of the patient. Over recent years the interest for quality assurance in radiotherapy through in vivo dosimetry has steadily increased. Thermoluminescent [1,10] and semiconductor [4,6,7,9] detectors are the methods of choice for quantitative in vivo dosimetry. In that respect, use of port films for in vivo dosimetry would provide valuable and complementary information. Together with one TL or semiconductor detector applied on the patient's skin along the beam axis, this method would provide absolute exit dose distributions. A feasibility study of the method proposed has been performed. It is based on phantom measurements but has been completed by a few densitometric evaluations of patient port films. The aim of the present work is to compare densitometric distributions derived from port films, set at some distance from the phantom, with the dose distribution measured in full backscatter conditions at the level of the exit field. Materials and methods In the present work port films are used only to obtain exit dose distributions as a percentage of the central exit dose measured along the beam axis. This latter kind of measurement is done routinely in the Leuven radiotherapy department [7]. This central exit dose is measured with a diode and the signal is corrected in order to get the absorbed dose at the level of the exit field in full backscatter conditions [7]. These conditions are applied in order to be able to compare the results with the dose calculated by the treatment planning system. The port film method allows then to perform the same comparison for whole exit dose distributions.
The films Conventional port films are used and processed as done routinely in the radiotherapy department without any kind of modification, in order to check the validity of the method proposed in the most unfavourable (but clinically realistic) conditions encountered. Kodak "Verification" films are routinely used in Kodak "Localization" cassettes (with 1 mm copper in front and 0.25 mm lead at the back of the film). This combination of film and cassette has been selected for clinical use on the basis of visual assessment of image quality. Films are processed in a Gevaert automatic processing unit. In order to use only the linear part of the sensitometric curve of the verification film, the maximum absorbed dose to water
Address for correspondence: Jan Van Dam, U.Z. St. Rafael, 3000 Leuven, Belgium.
68 delivered at the film level in the p h a n t o m experiments was limited to 0.3 Gy [this dose level was, however, exceeded for the patient port films (see further)].
Phantom dose profile measurements The measurements were performed with the 6 M V X-ray beam o f a Siemens Mevatron 6700 linear accelerator. Port films were evaluated by densitometry and the profiles derived were compared to ionometric profiles measured within the p h a n t o m at 1.5 c m from the exit surface. This comparison was performed in three different phantoms. A homogeneous polystyrene p h a n t o m (phantom A) was used in the first part o f the study. In the second step (phantom B) some inhomogeneities were introduced in the phantom. As most o f the patients treated on this linac (6 MV) suffer from head and neck cancers or breast cancers, the thickness of the p h a n t o m s used has been adapted to those patients and was about 14 cm. In a third step (phantom C), large inhomogeneities were introduced in the phantom, in order to test the method in boundary conditions.
without a hole in it (Fig. 1A2). E x p a n d e d polystyrene spacers allowed positioning of the films at distances d corresponding to the clinical range (10 to 40 cm). For comparison with the ionometric profiles the densitometric profiles were "projected back" to the chamber level, taking into account beam divergency.
Phantom B Three cm thick air, wood (density = 0.6 g/cm 3) and aluminium (density = 2.7 g/cm 3) inhomogeneities were introduced in the phantom (Fig. 1B, which corresponds to the geometry applied for film measurements), in order to simulate air cavities, lung and bone structures in the patient. These 5 c m wide inhomogeneities have, along the profile direction, a length of 2 cm for the air and of 8 c m for the wood and the aluminium. A similar comparison between profiles obtained by ionometry and densitometry was performed for the layer with the inhomogeneities placed at two different levels in the phantom: "close" to and "far" from the exit surface (Fig. 1B). Measurements were performed for three different positions o f the central beam axis:
Phantom A A flat 14 c m thick polystyrene p h a n t o m was positioned on the 1 cm thick Perspex insert (as provided by the manufacturer) o f the treatment couch and, for practical reasons, the gantry angle was turned at 180 ° (Fig. 1A 1 and 1A2). The upper part of the p h a n t o m was o f a variable geometry in order to allow measurements with an ionization chamber as well as with a film. In the ionization chamber version (Fig. 1A1) it consisted o f a separate 3 c m thick polystyrene layer, in which a hole was provided for a N E 0.6 cm 3 thimble chamber. The chamber centre was at a distance dmax (1.5 cm for 6 M V X-rays) from the exit side of the phantom. In order to ensure almost complete p h o t o n backscatter at the level o f the chamber (for the same reason as explained above for the exit dose measurements with a diode), the p h a n t o m was covered by a 1 0 c m thick polystyrene layer. The chamber-containing layer was moved manually across the field in order to obtain an ionometric profile. For film irradiation the backscatter block was removed and the block containing the chamber was replaced by a 3 cm thick layer
.
I
i ._7_ F A1
~
(1) In a first set-up the beam axis was adjusted through the centre o f the air cavity (Fig. 1B). A field size o f 10 x 10 cm at isocentre was adopted so that the two lateral parts o f the field went through wood and polystyrene, respectively. An off-axis ratio was determined for the two materials at 3 c m from the field edges (see left and right arrow on Fig. 4). These positions were chosen because they are in a region behind the inhomogeneity and are a compromise between the need to avoid the penumbra as well as the transition zone between the edges of different materials. Behind the air cavity the maximum offaxis ratio (see middle arrow on Fig. 4) was taken as the representative one. Polystyrene being the reference material, both the air and the w o o d values were expressed relatively to that for polystyrene. (2) In an other set-up, with the same field size of 10 x 10 cm, the beam axis was adjusted through the aluminium-polystyrene interface. The off-axis ratios for aluminium and polystyrene were again taken at 3 cm from the respective field edges. (3) In the third set-up the beam axis was put at the centre of the
F=tM¢ASSETI~ £
=
~.c
15
t
" "t" '
;
--IL:--III .... ~,...... t - ,
2,
"FAn"
._7_ ::oo F Az
F B
g C
Fig. 1. Schematic view of the phantoms used for present study. A homogeneous polystyrene phantom (phantom "A", see text) was positioned on the Perspex insert of the treatment couch and, for practical reasons, the gantry was turned at 180 ° . The ionization chamber (Fig. 1A~) was positioned at a distance of dmax (1.5 cm) below the exit surface of the phantom. It was moved manually across the field for measurement of the ionometric profile. The phantom was covered with a 10 cm thick polystyrene block in order to produce quasi-full backscatter conditions at the level of the chamber. For port film irradiation (Fig. 1A2) the backscatter block was removed and the film was positioned at a distance d from the phantom with four spacers. Similar experiments were performed after introduction of aluminium, wood and air inhomogeneities (Fig. 1B) in the polystyrene phantom (phantom "B", see text). The layer with the inhomogeneities was positioned either in position "far" or "close". When this layer was in one position, the other was occupied by a homogeneous polystyrene layer of the same dimensions. A few orientative measurements were also performed in boundary conditions, with a phantom (Fig. 1C) containing bulky pieces of wood and aluminium and large air cavities (phantom "C", see text).
69 phantom, crossing polystyrene only. In order to include all three inhomogeneities inside the field, the beam size was increased up to 20 x 20 cm. Off-axis ratios were determined either at the centre of the inhomogeneity for the air cavity or at 3 cm from the field edges (for wood and aluminium).
Phantom C Finally, a few measurements were performed with a more complex phantom (Fig. lC), following the model described by Wong et al. [ 13]. This p h a n t o m was used in order to produce boundary conditions regarding inhomogeneities. It contains bulky pieces of wood and aluminium and large air cavities. A field size o f 25 x 25 cm at isocentre was applied. A film was irradiated at 10 cm from the phantom and the density profile was compared with ionization chamber measurements.
Patient dose profile measurements For a patient treated for a head and neck malignancy with tho lateral fields, two semiconductor detectors were positioned for each field on the exit surface of the patient, one at the field center and the other one on an off-axis position. As the diodes for the first field were removed when the other one was irradiated, there was no problem of shadowing. A total of three different off-axis positions were explored in four treatment sessions (Fig. 2). During each session a verification port film was also irradiated with one of the two fields for the full duration o f the irradiation time (corresponding to standard clinical practice). The patient-film distance was 30 cm. The dose delivered to the film was about 0.50 Gy. On each o f the four port films a densitometric profile was measured along two lines: a line tangential to the shadow o f the central diode and another one tangential to the off-axis diode. The ratio of the optical densities at the level of the off- and onaxis diode was compared to that of the doses determined with these detectors. As the diodes are not necessarily at the same source distance because of the curvature of the patient exit surface, the off-axis ratios determined were corrected for differences in distance.
Results
Phantom dose profile measurements Reproducibility Ten films have been irradiated on different days and processed at different times of the day in the automatic processor used for routine port films. The variation of the off-axis ratio was within + 2}o. It includes the possible variations in beam homogeneity, film processing and densitometer reproducibility. This reproducibility is therefore the main limitation in the precision of the method.
Homogeneous polystyrene phantom A For a film-phantom distance of 10 cm a satisfactory agreement (deviations smaller than 1 - 2 % ) between ionometric and densitometric profiles is obtained over the whole field width (Fig. 3). When moving the film further away from the p h a n t o m increasing differences between the two methods are observed close to the field edges: for a phantom-film distance o f 30 cm the chamber values near the field edges are some 5% lower than the film values.
Inhomogeneous phantom B For the inhomogeneity "far" from the ionization chamber the inhomogeneity corrections determined with this detector are larger than the corresponding values for the inhomogeneity in position "close". This is observed both for the 10 × 10 cm field (Table I) and for the 20 x 20 cm field (Table II). The statistical fluctuations in the results obtained with film for the 10 x 10 cm field size (Table I) do not allow the detection of any systematic influence of film-inhomogeneity distance on the inhomogeneity corrections, which could be expected from ionization measurements. In comparison, for the 20 x 20 cm field (Table II), the inhomogeneity corrections tend to increase with the distance between inhomogeneity and film. Under all conditions explored with inhomogeneous p h a n t o m B the off-axis ratios determined by film are in good agreement with the OAR 1120
~
:30cm
d-
loom
080
"060
0~,0 [FIELD 20crn -20cm ]C~o
20
0 --
i
-15cm
Fig. 2. Schematic representation of the head and neck patient port film image. During each session a diode was positioned in the field centre (position "C") on the exit surface of the field. A second diode was positioned on one of the three off-axis positions. These positions were all in the shadow of important bone structures: the vertebral body for position 1 and 3 and the mandibula for position 2. Densitometric profiles were recorded along the broken lines, which arc tangential to the shadow of the diodes on the film.
i
-10cm
i
-Scm
i
+Scm
CHAMBER FILM ,
i
• 10cm
* 15crn ~x
Fig. 3. Dose profiles [Off-Axis-Ratio (OAR) versus distance from field ccntrc] along one of the axes of a 20 x 20 cm field (at isocentre) for the homogeneous
polystyrene phantom (Figs IA l and 1A:). The densitometric profiles ( - - ) are obtained by back projection (taking beam divergency into account) of the cross plots read on the film for film-phantom distances of 10 and 30 cm. They are compared with the ionometric profile (o).
70 TABLE I Exit off-axis ratios with respect to polystryrene for wood, air and aluminium with the beam axis passing through an inhomogeneity (field size 10 × 10 cm). Inhomogeneity
Position beam axis
Nature
Relative position to detector
Wood
Close Far
Centrum air cavity
Air
Close Far
AI
Close Far
Chamber value
Film value at film-phantom distance 10 cm
20 cm
40 cm
1.04 1.06
1.04 1.03
1.05 1.03
1.05 1.05
Centrum air cavity
1.11 1.13
1.10 1.10
1.10 1.10
1.12 1.10
Al-polystyrene interface
0.88 0.85
0.86 0.85
0.84 0.88
0.84 0.87
TABLE II Exit off-axis ratios for wood, air and aluminium with the beam axis passing entirely through polystryrene (field size 20 × 20 cm). Inhomogeneity Nature
Chamber value
Off-axis ratio
Relative position to detector
Film value at film-phantom distance 10 cm
30 cm
Wood
Close Far
1.01 1.04
1.00 1.02
1.01 1.04
Air
Close Far
1.11 1.125
1.11 1.14
1.15 1.13
AI
Close Far
0.845 0.83
0.83 0.81
0.82 0.82
c h a m b e r values, mostly within 1 or 2 % . T h e m a x i m u m deviation does n o t exceed 5% (Fig. 4).
OA AIR
l - 1.20 wooo
Inhomogeneous phantom C
PS (REF.)op
D u e to the importance o f the inhomogeneities introduced in the p h a n t o m large dose variations are observed over the densitometric a n d ionometric profiles. T h e off-axis ratios range between about 0.8 a n d 1.5. Despite this i m p o r t a n t dose variation the densitometric profile is in good agreement with the ionometric one (deviations are mostly of the order of 1% over a substantial part of the profile) (Fig. 5).
080 0.60
Patient dose profile measurements 0.40
In general, the differences in dose observed between the field centre a n d the off-axis positions (Fig. 2) are m o r e p r o n o u n c e d for diodes t h a n for film (Table III). However, the deviations never exceed 5% for two m e a s u r e m e n t s performed during the s a m e session. T h e difference between the off-axis ratios determined for position "1" (see sessions 1 a n d 2 on Table III) is smaller t h a n 3% for either method. Discussion In the present m e a s u r e m e n t s o f exit dose profiles, three different types o f conditions have varied: the type of detector, the backscatter conditions a n d the distance between film a n d p h a n t o m . T h e dose profiles m e a s u r e d by films are m e a s u r e d "in air" at s o m e distance from the p h a n t o m while the dose profiles m e a s u r e d with an ionization c h a m b e r are m e a s u r e d in the p h a n t o m u n d e r full backscatter conditions. It should be stressed that this a p p r o a c h is essen-
/
I
-12cm
J
FIELD 10cm ~ 10cm l d = 20cm 0 CHAMBER
0.20
- -
I
I
I
-Scm
-/,cm
*/.cm
FI
LM
I
+8cm
I
*12cm I,X
Fig. 4. Similar data as presented in Fig. 3, but for the phantom containing air, wood and aluminium inhomogeneities (phantom B, see text). The field set-up corresponded to that indicated in Fig. 1B, with the beam axis through the center of the air cavity and a field size of 10 x 10 cm at isocentre. Polystyrene and wood were positioned in the left and fight part of the field, respectively. The film-phantom distance was equal to 20 cm and the inhomogeneities were in the position "far" (see text) from the detector. The OAR's have been expressed relatively to that for polystyrene, which has been taken at 3 cm within the field (see left arrow). The middle and fight arrow indicate the positions were the representative values (see Tables I and II) have been taken for air and wood, respectively.
71 OAR
t
175
.50
.25 OO~
~
1.00
075
X 6HV
050
d = 10 cm
025
--
FIELD
i
-15cm
i
i
-10crn
-5cm
25 x 25 or.
F I Li'4
i
*Scm
L
+10cm
i
"15cm bx
Fig. 5. Similar data as presented in Figs. 3 and 4 but for the phantom containing bulky inhomogeneities (phantom C, see text). The beam axis passed through Perspex, wood, aluminium and polystyrene. The field size at isocentre was 25 x 25 cm and the film-phantom distance was 10 cm. TABLE III Exit off-axis ratios determined on a head-and-neck patient from in vivo diode dosimetry and from portal films. Session
1 2 3 4
Off-axis position*
1 l 2 3
Off-axis ratio Diodes
Port film
0.88 0.86 0.90 0.79
0.91 0.90 0.92 0.79
* See Fig. 2 for position identification.
tially a pragmatic one. The in vivo dosimetry with port films is proposed for radiotherapists as a routine quality control tool for checking exit dose distributions calculated by the Treastment Planning System commercially available. Film dosimetry is a critical method in the sense that film response is known to show pronounced spectral dependence. If the ionometric and densitometric profiles obtained under different scattering conditions, as specified above, agree with each other, it may be by chance. It can probably be explained by compensation phenomena. A detailed analysis of correlation between film and ionization chamber is underway and will be the subject of a more physically oriented paper. The beam profiles measured at a large distance from the phantom correspond essentially to the profile of the primary fluence since most of the scattered photons are scattered at a large angle and are progressively lost when the distance between the detector and the phantom increases. As the profiles are normalized on the beam axis, the expected variation can be estimated as follows for the homogeneous phantom: the primary fluence is relatively constant throughout the beam while the fluence of the scattered radiation decreases when the distance from the beam axis increases. Because part of the scattered radiation is lost, the detector overestimates the dose near the edge of the beam: the larger the field size and the larger the detector-phantom distance, the greater the overestimation.
The beam profiles measured in a penumbra region have been shown to depend on detector size [ 11 ]. Due to the relatively large size of an ionization chamber (e.g. diameter 6 mm for the chamber used for present study), the dose profiles are "spread out" in a zone with a high dose gradient, compared to a film. On top of the factors already discussed, part of the differences observed may also be due to positional uncertainties between ionization and film measurements. Despite the pronounced variation in conditions between reference chamber measurements and film measurements, the agreement between the two kinds of profiles obtained for the homogeneous phantom is surprisingly good (deviations never exceeding 5 %, even for large fields and large film-phantom distances). The same general rationale may be applied for the complex situation of inhomogeneous phantoms: the relative contributions of primary and scattered radiation at the level of the detector depend on detector-inhomogeneity distance. This dependence is expected to be more pronounced for larger field sizes. The results obtained with phantoms B and C indeed tend to confirm these theoretical expectations and are consistent with other published data [2,12]. For low density material, reported data show an increase of the inhomogeneity correction factor as a function of the detector - inhomogeneity distance. Although some calculation methods show a decrease behind high density material, the measurements do not show any decrease (apart from points in the interface region, in which we are not interested). As the spread on present data is too large to reach statistical significance, additional experiments are necessary to confirm the trend observed. For head and neck patient measurements it should be noted that the film underestimates the dose variation somewhat and therefore the dose inhomogeneity through the irradiated volume. Part of this effect could possibly be related to saturation effects in the film values, which are caused by the higher doses to the film than delivered in the phantom experiments (see above). However, tbese results concern only one patient and additional measurements are needed to confirm these preliminary findings.
Conclusions The present results show the possibility to use conventional port films, without sophisticated devices, to evaluate the dose variation throughout the exit field. The simultaneous measurements with port films and in vivo dosimeters on the beam axis allow for the determination, during the same session, of the accuracy of the dose delivered on the beam axis and the relative dose variation throughout the exit field. The maximum deviation of 5% in the dose variation which can be expected from the present work could probably be reduced by slight improvements in the method, e.g. correction for the non-linear dose-density curve or by optimization of the choice of the film, the cassette or the patient-film distance. The next step will be the comparison of the measured exit doses with the theoretical ones, as calculated by the treatment planning system. Finally, on top of the optimization of the port film method, the long-term objective is the use of real time images for in vivo dosimetry.
Acknowledgements The authors gratefully acknowledge the valuable comments and suggestions by Dr. C. Ling and Dr. B. Mijnheer.
72 References 1 Chavaudra, J., Marinello, G., BrOl6, A. M. and Nguyen, J. Utilisation pratique du borate de lithium eta dosim6trie par thermoluminescence. J. Radiol. Electrol. 57: 435-455, 1976. 2 Dutreix, A. Dosimetry of high energy photon beams in a non homogeneous medium. In: Advances in Radiation Protection and Dosimetry in Medicine, pp. 345-365. Editors: Ralph H. Thomas and Victor PerrezMendez. Plenum Press, New York, 1980. 3 Dutreix, J. and Dutreix, A. Film dosimetry of high-energy electrons. Ann. N.Y. Acad. Sci. 161: 33-43, 1969. 4 Heukelom, S., Lanson, J. H. and Mijnheer, B.J. Comparison of entrance and exit dose measurements using ionization chambers and silicon diodes. Phys. Med. Biol. 36: 47-59, 1991. 5 Leong, J. and Shimm, D. A method for consistent precision radiation therapy. Radiother. Oncol. 3: 89-92, 1985. 6 Leunens, G., Van Dam, J., Dutreix, A. and van der Schueren, E. Quality assurance in radiotherapy by in vivo dosimetry. 1. Entrance dose measurements, a reliable procedure. Radiother. Oncol. 17: 141-151, 1990. 7 Leunens, G., Van Dam, J., Dutreix, A. and van der Schueren, E. Quality assurance in radiotherapy by in vivo dosimetry. 2. Determination of the target absorbed dose. Radiother. Oncol. 19: 73-87, 1990.
8 Mitine, C., Leunens, G., Verstraete, J., Blanckaert, N., Van Dam, J., Dutreix, A. and van der Schueren, E. Is it necessary to repeat quality control procedures for head and neck patients? Radiother. Oncol. 21: 201-210, 1991. 9 Nilsson, B., Rud6n, B. I. and Sorcini, B. Characteristics of silicon diodes as patient dosemeters in external beam therapy. Radiother. Oncol. 11: 279-288, 1988. 10 Rud6n, B.I. Evaluation of the clinical use of TLD. Acta Radiol. Ther. Phys. Biol. 15: 447-464, 1976. 11 Sibata, C.H., Mota, H.C., Beddar, A.S., Higgins, P.D. and Shin, K.H. Influence of detector size in photon beam profile measurements. Phys. Med. Biol. 36: 621-631, 1991. 12 Wong, J. W. and Purdy, J.A. On methods of inhomogeneity corrections for photon transport. Med. Phys. 17: 807-814, 1990. 13 Wong, J. W., Slessinger, E.D. Hermes, R. E., Offutt, C. J., Roy, T. and Vannier, M.W. Portal dose images I: quantitative treatment plan verification. Int. J. Radiat. Oncol. Biol. Phys. 18: 1455-1463, 1990. 14 Ying, X.,Geer, L.Y. andWong, J.W. Portal dose images lI: patient dose estimation. Int. J. Radiat. Oncol. Biol. Phys. 18: 1465-1475, 1990.
67
RADION 01037
Short Communication
Are port films reliable for in vivo exit dose measurements? Jan Van Dam, Catherine Vaerman, Nino Blanckaert, Godelieve Leunens, Andr6e Dutreix and Emmanuel van der Schueren Department of Radiotherapy, U.Z. St. Rafael, Leuven, Belgium (Received 27 August 1991, revision received 7 May 1992, accepted 21 May 1992)
Key words: In vivo dosimetry; Port film; Quality assurance
Summary The possibility of using conventional port films as an in vivo method for obtaining relative exit dose distribution in patients is evaluated. Kodak "Verification" films in "Localization" cassettes are positioned in the usual clinical conditions behind an homogeneous polystyrene phantom as well as behind a phantom containing air, wood and aluminium inhomogeneities. Taking beam divergency into account the densitometric profiles are projected back to the exit side of the phantom. They are compared to the profiles obtained with an ionization chamber used under full backscatter conditions. The agreement between the off-axis ratios determined with either method are mostly better than 2 % and never exceed 5 %. These phantom measurements are completed by a comparison between off-axis ratios determined on a port film for a head and neck patient and those obtained by diode dosimetry applied on the patient at the exit side of the beam. A similar agreement as between film and ion chamber on the phantoms is obtained. These encouraging results illustrate the possibilities of using conventional port films for in vivo dosimetry.
Introduction The use of port films in radiotherapy is a well-established method for quality assurance. Their standard use is of a rather qualitative nature and consists mainly of visual evaluation of the field set-up. Quantitative evaluation of set-up can be made by measurement of the distances between the anatomical structures (bones, lung or air cavities) and the field edge, corrected for image magnification [5,8]. On the other hand, films are also used for phantom dosimetry for several decades. This method had indeed been shown [3] to have unique possibilities for relative dosimetry, provided the necessary precautions are taken: densitometric measurements of the films allow analysis of a complete dose distribution in the film plane with a short film exposure requiring the immobilization of the treatment unit for only a short time. From that respect, the use of port films for obtaining relative dose distributions seems only to be a logical extension of their standard qualitative use. Ying et al. [ 14] propose to use the dose distributions at the level of the port film to correct the CT data used for treatment planning purposes.
An alternative approach is to use the port film data for calculation of exit dose distributions at the level of the patient. Over recent years the interest for quality assurance in radiotherapy through in vivo dosimetry has steadily increased. Thermoluminescent [1,10] and semiconductor [4,6,7,9] detectors are the methods of choice for quantitative in vivo dosimetry. In that respect, use of port films for in vivo dosimetry would provide valuable and complementary information. Together with one TL or semiconductor detector applied on the patient's skin along the beam axis, this method would provide absolute exit dose distributions. A feasibility study of the method proposed has been performed. It is based on phantom measurements but has been completed by a few densitometric evaluations of patient port films. The aim of the present work is to compare densitometric distributions derived from port films, set at some distance from the phantom, with the dose distribution measured in full backscatter conditions at the level of the exit field. Materials and methods In the present work port films are used only to obtain exit dose distributions as a percentage of the central exit dose measured along the beam axis. This latter kind of measurement is done routinely in the Leuven radiotherapy department [7]. This central exit dose is measured with a diode and the signal is corrected in order to get the absorbed dose at the level of the exit field in full backscatter conditions [7]. These conditions are applied in order to be able to compare the results with the dose calculated by the treatment planning system. The port film method allows then to perform the same comparison for whole exit dose distributions.
The films Conventional port films are used and processed as done routinely in the radiotherapy department without any kind of modification, in order to check the validity of the method proposed in the most unfavourable (but clinically realistic) conditions encountered. Kodak "Verification" films are routinely used in Kodak "Localization" cassettes (with 1 mm copper in front and 0.25 mm lead at the back of the film). This combination of film and cassette has been selected for clinical use on the basis of visual assessment of image quality. Films are processed in a Gevaert automatic processing unit. In order to use only the linear part of the sensitometric curve of the verification film, the maximum absorbed dose to water
Address for correspondence: Jan Van Dam, U.Z. St. Rafael, 3000 Leuven, Belgium.
68 delivered at the film level in the p h a n t o m experiments was limited to 0.3 Gy [this dose level was, however, exceeded for the patient port films (see further)].
Phantom dose profile measurements The measurements were performed with the 6 M V X-ray beam o f a Siemens Mevatron 6700 linear accelerator. Port films were evaluated by densitometry and the profiles derived were compared to ionometric profiles measured within the p h a n t o m at 1.5 c m from the exit surface. This comparison was performed in three different phantoms. A homogeneous polystyrene p h a n t o m (phantom A) was used in the first part o f the study. In the second step (phantom B) some inhomogeneities were introduced in the phantom. As most o f the patients treated on this linac (6 MV) suffer from head and neck cancers or breast cancers, the thickness of the p h a n t o m s used has been adapted to those patients and was about 14 cm. In a third step (phantom C), large inhomogeneities were introduced in the phantom, in order to test the method in boundary conditions.
without a hole in it (Fig. 1A2). E x p a n d e d polystyrene spacers allowed positioning of the films at distances d corresponding to the clinical range (10 to 40 cm). For comparison with the ionometric profiles the densitometric profiles were "projected back" to the chamber level, taking into account beam divergency.
Phantom B Three cm thick air, wood (density = 0.6 g/cm 3) and aluminium (density = 2.7 g/cm 3) inhomogeneities were introduced in the phantom (Fig. 1B, which corresponds to the geometry applied for film measurements), in order to simulate air cavities, lung and bone structures in the patient. These 5 c m wide inhomogeneities have, along the profile direction, a length of 2 cm for the air and of 8 c m for the wood and the aluminium. A similar comparison between profiles obtained by ionometry and densitometry was performed for the layer with the inhomogeneities placed at two different levels in the phantom: "close" to and "far" from the exit surface (Fig. 1B). Measurements were performed for three different positions o f the central beam axis:
Phantom A A flat 14 c m thick polystyrene p h a n t o m was positioned on the 1 cm thick Perspex insert (as provided by the manufacturer) o f the treatment couch and, for practical reasons, the gantry angle was turned at 180 ° (Fig. 1A 1 and 1A2). The upper part of the p h a n t o m was o f a variable geometry in order to allow measurements with an ionization chamber as well as with a film. In the ionization chamber version (Fig. 1A1) it consisted o f a separate 3 c m thick polystyrene layer, in which a hole was provided for a N E 0.6 cm 3 thimble chamber. The chamber centre was at a distance dmax (1.5 cm for 6 M V X-rays) from the exit side of the phantom. In order to ensure almost complete p h o t o n backscatter at the level o f the chamber (for the same reason as explained above for the exit dose measurements with a diode), the p h a n t o m was covered by a 1 0 c m thick polystyrene layer. The chamber-containing layer was moved manually across the field in order to obtain an ionometric profile. For film irradiation the backscatter block was removed and the block containing the chamber was replaced by a 3 cm thick layer
.
I
i ._7_ F A1
~
(1) In a first set-up the beam axis was adjusted through the centre o f the air cavity (Fig. 1B). A field size o f 10 x 10 cm at isocentre was adopted so that the two lateral parts o f the field went through wood and polystyrene, respectively. An off-axis ratio was determined for the two materials at 3 c m from the field edges (see left and right arrow on Fig. 4). These positions were chosen because they are in a region behind the inhomogeneity and are a compromise between the need to avoid the penumbra as well as the transition zone between the edges of different materials. Behind the air cavity the maximum offaxis ratio (see middle arrow on Fig. 4) was taken as the representative one. Polystyrene being the reference material, both the air and the w o o d values were expressed relatively to that for polystyrene. (2) In an other set-up, with the same field size of 10 x 10 cm, the beam axis was adjusted through the aluminium-polystyrene interface. The off-axis ratios for aluminium and polystyrene were again taken at 3 cm from the respective field edges. (3) In the third set-up the beam axis was put at the centre of the
F=tM¢ASSETI~ £
=
~.c
15
t
" "t" '
;
--IL:--III .... ~,...... t - ,
2,
"FAn"
._7_ ::oo F Az
F B
g C
Fig. 1. Schematic view of the phantoms used for present study. A homogeneous polystyrene phantom (phantom "A", see text) was positioned on the Perspex insert of the treatment couch and, for practical reasons, the gantry was turned at 180 ° . The ionization chamber (Fig. 1A~) was positioned at a distance of dmax (1.5 cm) below the exit surface of the phantom. It was moved manually across the field for measurement of the ionometric profile. The phantom was covered with a 10 cm thick polystyrene block in order to produce quasi-full backscatter conditions at the level of the chamber. For port film irradiation (Fig. 1A2) the backscatter block was removed and the film was positioned at a distance d from the phantom with four spacers. Similar experiments were performed after introduction of aluminium, wood and air inhomogeneities (Fig. 1B) in the polystyrene phantom (phantom "B", see text). The layer with the inhomogeneities was positioned either in position "far" or "close". When this layer was in one position, the other was occupied by a homogeneous polystyrene layer of the same dimensions. A few orientative measurements were also performed in boundary conditions, with a phantom (Fig. 1C) containing bulky pieces of wood and aluminium and large air cavities (phantom "C", see text).
69 phantom, crossing polystyrene only. In order to include all three inhomogeneities inside the field, the beam size was increased up to 20 x 20 cm. Off-axis ratios were determined either at the centre of the inhomogeneity for the air cavity or at 3 cm from the field edges (for wood and aluminium).
Phantom C Finally, a few measurements were performed with a more complex phantom (Fig. lC), following the model described by Wong et al. [ 13]. This p h a n t o m was used in order to produce boundary conditions regarding inhomogeneities. It contains bulky pieces of wood and aluminium and large air cavities. A field size o f 25 x 25 cm at isocentre was applied. A film was irradiated at 10 cm from the phantom and the density profile was compared with ionization chamber measurements.
Patient dose profile measurements For a patient treated for a head and neck malignancy with tho lateral fields, two semiconductor detectors were positioned for each field on the exit surface of the patient, one at the field center and the other one on an off-axis position. As the diodes for the first field were removed when the other one was irradiated, there was no problem of shadowing. A total of three different off-axis positions were explored in four treatment sessions (Fig. 2). During each session a verification port film was also irradiated with one of the two fields for the full duration o f the irradiation time (corresponding to standard clinical practice). The patient-film distance was 30 cm. The dose delivered to the film was about 0.50 Gy. On each o f the four port films a densitometric profile was measured along two lines: a line tangential to the shadow o f the central diode and another one tangential to the off-axis diode. The ratio of the optical densities at the level of the off- and onaxis diode was compared to that of the doses determined with these detectors. As the diodes are not necessarily at the same source distance because of the curvature of the patient exit surface, the off-axis ratios determined were corrected for differences in distance.
Results
Phantom dose profile measurements Reproducibility Ten films have been irradiated on different days and processed at different times of the day in the automatic processor used for routine port films. The variation of the off-axis ratio was within + 2}o. It includes the possible variations in beam homogeneity, film processing and densitometer reproducibility. This reproducibility is therefore the main limitation in the precision of the method.
Homogeneous polystyrene phantom A For a film-phantom distance of 10 cm a satisfactory agreement (deviations smaller than 1 - 2 % ) between ionometric and densitometric profiles is obtained over the whole field width (Fig. 3). When moving the film further away from the p h a n t o m increasing differences between the two methods are observed close to the field edges: for a phantom-film distance o f 30 cm the chamber values near the field edges are some 5% lower than the film values.
Inhomogeneous phantom B For the inhomogeneity "far" from the ionization chamber the inhomogeneity corrections determined with this detector are larger than the corresponding values for the inhomogeneity in position "close". This is observed both for the 10 × 10 cm field (Table I) and for the 20 x 20 cm field (Table II). The statistical fluctuations in the results obtained with film for the 10 x 10 cm field size (Table I) do not allow the detection of any systematic influence of film-inhomogeneity distance on the inhomogeneity corrections, which could be expected from ionization measurements. In comparison, for the 20 x 20 cm field (Table II), the inhomogeneity corrections tend to increase with the distance between inhomogeneity and film. Under all conditions explored with inhomogeneous p h a n t o m B the off-axis ratios determined by film are in good agreement with the OAR 1120
~
:30cm
d-
loom
080
"060
0~,0 [FIELD 20crn -20cm ]C~o
20
0 --
i
-15cm
Fig. 2. Schematic representation of the head and neck patient port film image. During each session a diode was positioned in the field centre (position "C") on the exit surface of the field. A second diode was positioned on one of the three off-axis positions. These positions were all in the shadow of important bone structures: the vertebral body for position 1 and 3 and the mandibula for position 2. Densitometric profiles were recorded along the broken lines, which arc tangential to the shadow of the diodes on the film.
i
-10cm
i
-Scm
i
+Scm
CHAMBER FILM ,
i
• 10cm
* 15crn ~x
Fig. 3. Dose profiles [Off-Axis-Ratio (OAR) versus distance from field ccntrc] along one of the axes of a 20 x 20 cm field (at isocentre) for the homogeneous
polystyrene phantom (Figs IA l and 1A:). The densitometric profiles ( - - ) are obtained by back projection (taking beam divergency into account) of the cross plots read on the film for film-phantom distances of 10 and 30 cm. They are compared with the ionometric profile (o).
70 TABLE I Exit off-axis ratios with respect to polystryrene for wood, air and aluminium with the beam axis passing through an inhomogeneity (field size 10 × 10 cm). Inhomogeneity
Position beam axis
Nature
Relative position to detector
Wood
Close Far
Centrum air cavity
Air
Close Far
AI
Close Far
Chamber value
Film value at film-phantom distance 10 cm
20 cm
40 cm
1.04 1.06
1.04 1.03
1.05 1.03
1.05 1.05
Centrum air cavity
1.11 1.13
1.10 1.10
1.10 1.10
1.12 1.10
Al-polystyrene interface
0.88 0.85
0.86 0.85
0.84 0.88
0.84 0.87
TABLE II Exit off-axis ratios for wood, air and aluminium with the beam axis passing entirely through polystryrene (field size 20 × 20 cm). Inhomogeneity Nature
Chamber value
Off-axis ratio
Relative position to detector
Film value at film-phantom distance 10 cm
30 cm
Wood
Close Far
1.01 1.04
1.00 1.02
1.01 1.04
Air
Close Far
1.11 1.125
1.11 1.14
1.15 1.13
AI
Close Far
0.845 0.83
0.83 0.81
0.82 0.82
c h a m b e r values, mostly within 1 or 2 % . T h e m a x i m u m deviation does n o t exceed 5% (Fig. 4).
OA AIR
l - 1.20 wooo
Inhomogeneous phantom C
PS (REF.)op
D u e to the importance o f the inhomogeneities introduced in the p h a n t o m large dose variations are observed over the densitometric a n d ionometric profiles. T h e off-axis ratios range between about 0.8 a n d 1.5. Despite this i m p o r t a n t dose variation the densitometric profile is in good agreement with the ionometric one (deviations are mostly of the order of 1% over a substantial part of the profile) (Fig. 5).
080 0.60
Patient dose profile measurements 0.40
In general, the differences in dose observed between the field centre a n d the off-axis positions (Fig. 2) are m o r e p r o n o u n c e d for diodes t h a n for film (Table III). However, the deviations never exceed 5% for two m e a s u r e m e n t s performed during the s a m e session. T h e difference between the off-axis ratios determined for position "1" (see sessions 1 a n d 2 on Table III) is smaller t h a n 3% for either method. Discussion In the present m e a s u r e m e n t s o f exit dose profiles, three different types o f conditions have varied: the type of detector, the backscatter conditions a n d the distance between film a n d p h a n t o m . T h e dose profiles m e a s u r e d by films are m e a s u r e d "in air" at s o m e distance from the p h a n t o m while the dose profiles m e a s u r e d with an ionization c h a m b e r are m e a s u r e d in the p h a n t o m u n d e r full backscatter conditions. It should be stressed that this a p p r o a c h is essen-
/
I
-12cm
J
FIELD 10cm ~ 10cm l d = 20cm 0 CHAMBER
0.20
- -
I
I
I
-Scm
-/,cm
*/.cm
FI
LM
I
+8cm
I
*12cm I,X
Fig. 4. Similar data as presented in Fig. 3, but for the phantom containing air, wood and aluminium inhomogeneities (phantom B, see text). The field set-up corresponded to that indicated in Fig. 1B, with the beam axis through the center of the air cavity and a field size of 10 x 10 cm at isocentre. Polystyrene and wood were positioned in the left and fight part of the field, respectively. The film-phantom distance was equal to 20 cm and the inhomogeneities were in the position "far" (see text) from the detector. The OAR's have been expressed relatively to that for polystyrene, which has been taken at 3 cm within the field (see left arrow). The middle and fight arrow indicate the positions were the representative values (see Tables I and II) have been taken for air and wood, respectively.
71 OAR
t
175
.50
.25 OO~
~
1.00
075
X 6HV
050
d = 10 cm
025
--
FIELD
i
-15cm
i
i
-10crn
-5cm
25 x 25 or.
F I Li'4
i
*Scm
L
+10cm
i
"15cm bx
Fig. 5. Similar data as presented in Figs. 3 and 4 but for the phantom containing bulky inhomogeneities (phantom C, see text). The beam axis passed through Perspex, wood, aluminium and polystyrene. The field size at isocentre was 25 x 25 cm and the film-phantom distance was 10 cm. TABLE III Exit off-axis ratios determined on a head-and-neck patient from in vivo diode dosimetry and from portal films. Session
1 2 3 4
Off-axis position*
1 l 2 3
Off-axis ratio Diodes
Port film
0.88 0.86 0.90 0.79
0.91 0.90 0.92 0.79
* See Fig. 2 for position identification.
tially a pragmatic one. The in vivo dosimetry with port films is proposed for radiotherapists as a routine quality control tool for checking exit dose distributions calculated by the Treastment Planning System commercially available. Film dosimetry is a critical method in the sense that film response is known to show pronounced spectral dependence. If the ionometric and densitometric profiles obtained under different scattering conditions, as specified above, agree with each other, it may be by chance. It can probably be explained by compensation phenomena. A detailed analysis of correlation between film and ionization chamber is underway and will be the subject of a more physically oriented paper. The beam profiles measured at a large distance from the phantom correspond essentially to the profile of the primary fluence since most of the scattered photons are scattered at a large angle and are progressively lost when the distance between the detector and the phantom increases. As the profiles are normalized on the beam axis, the expected variation can be estimated as follows for the homogeneous phantom: the primary fluence is relatively constant throughout the beam while the fluence of the scattered radiation decreases when the distance from the beam axis increases. Because part of the scattered radiation is lost, the detector overestimates the dose near the edge of the beam: the larger the field size and the larger the detector-phantom distance, the greater the overestimation.
The beam profiles measured in a penumbra region have been shown to depend on detector size [ 11 ]. Due to the relatively large size of an ionization chamber (e.g. diameter 6 mm for the chamber used for present study), the dose profiles are "spread out" in a zone with a high dose gradient, compared to a film. On top of the factors already discussed, part of the differences observed may also be due to positional uncertainties between ionization and film measurements. Despite the pronounced variation in conditions between reference chamber measurements and film measurements, the agreement between the two kinds of profiles obtained for the homogeneous phantom is surprisingly good (deviations never exceeding 5 %, even for large fields and large film-phantom distances). The same general rationale may be applied for the complex situation of inhomogeneous phantoms: the relative contributions of primary and scattered radiation at the level of the detector depend on detector-inhomogeneity distance. This dependence is expected to be more pronounced for larger field sizes. The results obtained with phantoms B and C indeed tend to confirm these theoretical expectations and are consistent with other published data [2,12]. For low density material, reported data show an increase of the inhomogeneity correction factor as a function of the detector - inhomogeneity distance. Although some calculation methods show a decrease behind high density material, the measurements do not show any decrease (apart from points in the interface region, in which we are not interested). As the spread on present data is too large to reach statistical significance, additional experiments are necessary to confirm the trend observed. For head and neck patient measurements it should be noted that the film underestimates the dose variation somewhat and therefore the dose inhomogeneity through the irradiated volume. Part of this effect could possibly be related to saturation effects in the film values, which are caused by the higher doses to the film than delivered in the phantom experiments (see above). However, tbese results concern only one patient and additional measurements are needed to confirm these preliminary findings.
Conclusions The present results show the possibility to use conventional port films, without sophisticated devices, to evaluate the dose variation throughout the exit field. The simultaneous measurements with port films and in vivo dosimeters on the beam axis allow for the determination, during the same session, of the accuracy of the dose delivered on the beam axis and the relative dose variation throughout the exit field. The maximum deviation of 5% in the dose variation which can be expected from the present work could probably be reduced by slight improvements in the method, e.g. correction for the non-linear dose-density curve or by optimization of the choice of the film, the cassette or the patient-film distance. The next step will be the comparison of the measured exit doses with the theoretical ones, as calculated by the treatment planning system. Finally, on top of the optimization of the port film method, the long-term objective is the use of real time images for in vivo dosimetry.
Acknowledgements The authors gratefully acknowledge the valuable comments and suggestions by Dr. C. Ling and Dr. B. Mijnheer.
72 References 1 Chavaudra, J., Marinello, G., BrOl6, A. M. and Nguyen, J. Utilisation pratique du borate de lithium eta dosim6trie par thermoluminescence. J. Radiol. Electrol. 57: 435-455, 1976. 2 Dutreix, A. Dosimetry of high energy photon beams in a non homogeneous medium. In: Advances in Radiation Protection and Dosimetry in Medicine, pp. 345-365. Editors: Ralph H. Thomas and Victor PerrezMendez. Plenum Press, New York, 1980. 3 Dutreix, J. and Dutreix, A. Film dosimetry of high-energy electrons. Ann. N.Y. Acad. Sci. 161: 33-43, 1969. 4 Heukelom, S., Lanson, J. H. and Mijnheer, B.J. Comparison of entrance and exit dose measurements using ionization chambers and silicon diodes. Phys. Med. Biol. 36: 47-59, 1991. 5 Leong, J. and Shimm, D. A method for consistent precision radiation therapy. Radiother. Oncol. 3: 89-92, 1985. 6 Leunens, G., Van Dam, J., Dutreix, A. and van der Schueren, E. Quality assurance in radiotherapy by in vivo dosimetry. 1. Entrance dose measurements, a reliable procedure. Radiother. Oncol. 17: 141-151, 1990. 7 Leunens, G., Van Dam, J., Dutreix, A. and van der Schueren, E. Quality assurance in radiotherapy by in vivo dosimetry. 2. Determination of the target absorbed dose. Radiother. Oncol. 19: 73-87, 1990.
8 Mitine, C., Leunens, G., Verstraete, J., Blanckaert, N., Van Dam, J., Dutreix, A. and van der Schueren, E. Is it necessary to repeat quality control procedures for head and neck patients? Radiother. Oncol. 21: 201-210, 1991. 9 Nilsson, B., Rud6n, B. I. and Sorcini, B. Characteristics of silicon diodes as patient dosemeters in external beam therapy. Radiother. Oncol. 11: 279-288, 1988. 10 Rud6n, B.I. Evaluation of the clinical use of TLD. Acta Radiol. Ther. Phys. Biol. 15: 447-464, 1976. 11 Sibata, C.H., Mota, H.C., Beddar, A.S., Higgins, P.D. and Shin, K.H. Influence of detector size in photon beam profile measurements. Phys. Med. Biol. 36: 621-631, 1991. 12 Wong, J. W. and Purdy, J.A. On methods of inhomogeneity corrections for photon transport. Med. Phys. 17: 807-814, 1990. 13 Wong, J. W., Slessinger, E.D. Hermes, R. E., Offutt, C. J., Roy, T. and Vannier, M.W. Portal dose images I: quantitative treatment plan verification. Int. J. Radiat. Oncol. Biol. Phys. 18: 1455-1463, 1990. 14 Ying, X.,Geer, L.Y. andWong, J.W. Portal dose images lI: patient dose estimation. Int. J. Radiat. Oncol. Biol. Phys. 18: 1465-1475, 1990.