- Email: [email protected]
Dose intercomparison at the radiotherapy centres in The Netherlands. 1. Photon beams under reference conditions and for prostatic cancer treatment
Radiotherapy and Oncology, 9 (1987) 33-44 Elsevier
33
RTO 00329
Dose intercomparison at the radiotherapy centres in The Netherlands. 1. Photon beams under reference conditions and for prostatic cancer treatment F . W . W i t t k / i m p e r l'z, B. J. M i j n h e e r 1 a n d H . J. v a n K l e f f e n s 2 1Radiotherapy Department, The Netherlands Cancer Institute (Antoni van Leeuwenhoekhuis), 1066 CX Amsterdam, The Netherlands, and 2Physics Department, Rotterdam Radio-Therapeutic Institute (Dr. Danii~l den Hoed kliniek ), 3075 EA Rotterdam, The Netherlands
(Received 13 September 1986, revision received 4 December 1986, accepted 23 December 1986)
Key words." Quality assurance in radiotherapy; Dose intercomparison; Photon beam
Summary In 1985, a dosimetry intercomparison was performed at all 20 radiotherapy centres in The Netherlands. Absorbed dose was determined with an ionization chamber under reference conditions in a water phantom for cobalt-60 g a m m a - r a y and megavoltage X-ray beams. The mean difference between measured and stated dose values was 0.5% with a standard deviation of 1.9%, but up to 6% at maximum. As soon as all institutes apply a common dosimetry protocol, this maximum difference will reduce to about 2%. In addition, an anthropomorphic phantom was irradiated to simulate the treatment of a prostatic cancer. The dose, determined with an ionization chamber at the isocentre and thermoluminescent dosimeters (TLD) powder at several points situated in the target volume, the bladder and the rectum, was compared with the stated dose calculated with the local planning system. Only small differences were found between the measured and stated dose at the isocentre: on the average 1.5%, with a standard deviation of 1.5%. The difference between stated and measured dose at several points situated in the target volume was on the average 0.4%, with a standard deviation of 5.2%. Almost the same result was found for a point situated in the bladder. In the rectum, the average difference was about 4%, however, with a large standard deviation, 18%, due to the relatively steep dose gradient at these points.
Introduction During the last decade, growing attention has been paid to quality control in radiation therapy. This Address for correspondence: F. W. Wittkiimper, Radiotherapy
Department, The Netherlands Cancer Institute (Antoni van Leeuwenhoekhuis), 1066 CX Amsterdam, The Netherlands.
is necessary because small changes in the dose of radiation delivered to a target volume in a homogeneous group of patients may drastically influence the probability of tumour control or the probability of severe normal tissue reactions. This effect is quantified in recent surveys of dose-effect curves for local tumour control and normal tissue comptica-
0167-8140/87/$03.50 9 1987 Elsevier Science Publishers B.V. (Biomedical Division)
34 tions as given by Brahme [1], Johansson [5] and Mijnheer et al. [9]. Each step in the dosimetry chain, from the calibration of the dosimeter to the delivery of radiation to a patient, contributes to the final uncertainty in the actual dose given to a patient. The five main steps in this chain are: (1) The calibration of the dosimetry equipment; (2) the dose determination at a reference point in a water phantom under reference conditions; (3) the determination of the relative dose distribution in a phantom under non-reference conditions, including corrections for oblique incidence, wedge filters etc.; (4) the assessment of the dose at relevant points in a patient; (5) the delivery of the radiation to the patient under treatment conditions. In principle, there are two ways to estimate the total uncertainty in the dose given to a patient: by in vivo dosimetry and by estimating the uncertainties in each of the steps in the dosimetry chain. The use of in vivo dosimetry is an ultimate check of the treatment but is difficult to perform in a systematic way. Methods have been described in 1CRU Report 24 [4]. Dose intercomparisons can be used to quantify uncertainties in the dosimetry chain. A number of dose intercomparisons have been carried out in order to investigate whether the procedures used in different radiotherapy centres, often based on different protocols, result in a satisfying agreement in dose determination [5-8,13,14,16]. Table I summarises the results of some dose intercomparisons related to step 2, in which the measurements were performed with an ionization chamber, whereas Table ll summarises the results of dose intercomparisons related to step 4 or 5. These data show that considerable differences can be observed between dose values stated by an institute and that measured by a visiting team. The purpose of the dose intercomparison reported here is to quantify the differences betweeen stated and measured dose values for megavoltage photon beams under reference conditions (step 2) and
to compare stated and measured dose values in an anthropomorphic phantom under treatment conditions (step 5). It should be noticed that when irradiating an anthropomorphic phantom, instead of a patient, not all parameters involved in step 5 are taken into account (e.g. patient movement during irradiation). Therefore, uncertainties due to replacement of a patient by an anthropomorphic phantom have to be added to the observed dose differences. In a separate paper, the results of measured relative dose distributions (step 3) will be compared with calculated dose distributions.
Methods and materials
The dosimetry intercomparison took place from March to December 1985. The measurements were performed by the same physicist and with the same equipment. All 20 radiotherapy centres were visited once at an agreed date. The dose at the reference point in a water phantom was measured for all megavoltage photon beams available in the institutes (see Table IlI). The anthropomorphic phantom measurements were performed in each institute only at one photon beam; the photon beam normally used in that particular institute to irradiate a prostatic tumour. The measurements, with this phantom were carried out according to a technique prescribed and a technique chosen by the institute. Dose determination at the reJ'erence point The absorbed dose to water at the reference point was measured in a water phantom with a graphite-walled thimble ionization chamber (Type NE* 2505/3A). The water phantom was irradiated horizontally under reference conditions: field size 10 cm x 10 cm at the surface, and the source-surface distance (SSD), equal to the source-axis (isocentre) distance (SAD). The reference point was situated on the central axis of the beam at 5 cm depth for energies up to 15 MV and at 7 cm depth for energies between 16 and 25 MV. The geometrical centre of * Nuclear Enterprises Ltd.
35 8
d d ~ d
d
I
d d d d
d d d d
dd
go
~.o..~.
~. ~. ~ . ~ . ~ . ~ .
~.~..
~.~.
>
g 0 "0
•
8~aa
X88
r. V A
~L 0
0
.g
ca t..,
a
e.~
>
.g
g
0
f.-
8 8
.g
.g d 0
4
4
~
~
z
$
3
r~
rm 0 o~
.g
~J
<
. ~ =
~=
;8
.g
36
,a
e~
gg
~'
lea
ir
gg
g,
9
a< 6 g
e~
g
o
g 9
.-=
.~
<
<
.-
g.g
o_ ~ ' =-
-~ =
0
~4 0 ".,~_~
a
<
~"
o
.g
.--2,.
o
0
9
<
37 TABLE III Survey of the number of cobalt-60 teletherapy units and megavolt linear accelerators available for radiotherapy on January 1, 1986 in The Netherlands. No. of photon beams Cobalt-60
11
4 MV
5
5 MV 6 MV 8 MV 10 MV 12 MV 15 MV 16 MV 18 MV 23 MV 25 MV
4 5 15 2 1 l 4 1 1 1
Total: spread over:
51 photon beams 11 cobalt-60 teletherapy units 35 linear accelerators
(2)
Dw = M - NK - Cw,u
the ionization chamber was positioned at the reference point. The measurements were converted to absorbed dose to water in two ways. Firstly, by using the equation: Dw = N x . R . C~
summarises the different photon beam dosimetry protocols applied in 1985 in The Netherlands. Secondly, the measurements were converted to absorbed dose to water applying the equation given in a recent code of practice for the dosimetry of high-energy photon beams [10] which is going to be used in The Netherlands (NCS code of practice):
(1)
in which Dw is the absorbed dose to water, Nx is the exposure calibration factor, R is the electrometer reading corrected for temperature and pressure, but not for recombination and Cz is the exposure to absorbed dose to water conversion factor. The ionization chamber was calibrated against a secondary ionization chamber calibrated at the Dutch Standards Laboratory, according to the old exposure value, to which the chambers of all Dutch institutes are still calibrated. For each institute, the Cz values were adopted normally used by that particular institute (institute protocol). Some institutes are using dosimetry protocols in which several conversion and correction factors are given separately, e.g. the N A C P protocol [12]. All these factors, not related to the instrument reading, have been combined to one C~ value for these protocols. Table IV
where Dw is the absorbed dose to water, M is the electrometer reading corrected for temperature, pressure and recombination, ArK is the air kerma calibration factor and Cw.u is the air kerma to absorbed dose to water conversion factor. In this code of practice Cw,u values for reference ionization chambers (including the N E 2505/3A) are given as a function of the quality index; the ratio of the ionizations measured at 20 cm and 10 cm depth in water for a fixed SDD, and aficld size of 10 cm x 10 cm at the geometrical centre of the detector [2]. The quality indices of the accelerator beams were derived from the stated central axis depth dose values at 10 cm and 20 cm depth in water for a 10 cm x 10 cm field size at 100 cm SSD, by using a semi-empirical relation between the ratio of these P D D values and the quality index [11]. The air kerma calibration factor NK was calculated from the exposure calibration factor Nx, applying the recently revised value for exposure at the Dutch Standards Laboratory, using the equation:
N~ = Nx" W/e. (1 - g ) in which
W/e is the
1
(3)
quotient of the average energy
TABLE IV Survey of the different dosimetry protocols in use on January 1, 1986 at the radiotherapy centres in The Netherlands. Dosimetry protocol
Number
ICRU (1969) [3] NACP (1980) [12] Other protocols Fricke dosimetry
12 3 4 1
38 expended to produce an ion pair in dry air and the electron charge and g is the fraction of energy of secondary charged particles that is converted to bremsstrahlung in air. For g, a value of 0.003 was applied and for W/e a value o f 33.97 J/C according to the recommendations given in the code of practice. Although the value for the exposure at the Standards Laboratory is revised, the consequence for Dw, because of this change is minor. The change in the exposure value is compensated by an almost equal change in the W/e value. The chamber was recalibrated against a secondary standard ionization chamber in a cobalt-60 g a m m a - r a y beam shortly before and after each visit to an institute. N o changes in the exposure calibration factor larger than 0.2% were observed.
Dose determination during the prostatic cancer treatment (anthropomorphic phantom) In each institute, an anthropomorphic phantom (type Alderson Rando) was irradiated as for treatment of a prostatic cancer according to a technique prescribed and a technique chosen by the institute. Some weeks before the visit, two sheets with a contour of the anthropomorphic phantom were sent to the institute. One contour was taken at the centre of the prostate and one contour was taken 2.5 cm off-axis (see Figs. 1 and 2). The target volume, with ~ lug.
p~n~om a4
a,1 central
p~ne
Fig. I. Contour of the anthropomorphic phantom in the central plane of the field. The target volume (-.-.-) and the outlines of the prostate, the rectum, the bone structures and four measuring points (0) are indicated.
9 u ~ ~ a3
oft-axis
plene
2.sere
e
lder
7
Fig. 2. Contour of the anthropomorphic phantom in the 2.5 cm off-axis plane of the field. The target volume (-.-.-) and the outlines of the vesiculaseminalis, the rectum, the bladder, the bone structures and five dose points (0) are indicated. the prostate and the vesicula seminalis, the rectum, the bladder and the bone structures were marked in the contours. The prescribed technique consisted of a three-field technique using an AP beam and two opposing lateral wedged beams. A wedge with an angle of 45 ~ or the nearest wedge angle awlilable, was requested. The isocentre of the prescribed technique, the entry points of the three beams and nine specific points for the dose intercomparison were also marked on the contours. CT pictures were taken from the central plane and the 2.5 cm off-axis plane. Large inhomogeneities in the bone structures were observed. CT numbers varying from - 6 0 0 to + 200 were found, which correspond roughly to a mass density of 0.4 g/cm 3 and 1.12 g/cm 3, respectively. In the central plane, the average mass density was 0.9 g/cm -~ and in the off-axis plane 1.0 g/cm 3. These large inhomogeneities in the bony material of this anthropomorphic p h a n t o m was also reported by Somerwil and van Kleffens [15]. The institutes were asked: (a) to give a dose of 2.5 Gy to the point indicated as the isocentre of the prescribed technique; (b) to calculate with their computer planning system the dose in the nine points. The dose at the isocentre was determined with a small ionization chamber and in the nine other points with thermoluminescent dosimeters (TLD). The ionization chamber has a 0.5 m m thick polystyrene wall and cavity dimensions o f 12 m m
39 (length) and 4 mm (diameter). The chamber was calibrated in a water phantom, irradiated with photon beams from cobalt-60 to 25 MV, against the calibrated graphite-walled ionization chamber applying twice Eqn. (2). As TLD material, 12 mg of powdered TLD-700 was used in each point, yielding two readings. The TLD was read out using a Pitman model 654 TLD reader, applying 7 s pre read-out heating at 135~ followed by 6 s at a temperature of 240~ Annealing was performed by 1.5 h at 400~ followed by a natural cooling down in the oven (Pitman model 622/B) till 80~ and finally the powder was kept during 16 h at this temperature. The TLD was recularly calibrated in a 4 MV photon beam. Measurements at other energies showed that no photon energy dependence could be observed in the calibration. In the dose obtained with TLD, a supra-linearity correction has been applied.
20
Institute
=
protocol
c o b a l t - S O beams
16-
mean S.9.
12-
1.000
O.OOS 0.018 10
delta :3
number
4-
0 0.90
stated absorbed dose / measured absorbed dose
Fig. 3. The distribution of the ratios o f stated and measured dose values for cobalt-60 g a m m a - r a y beams under reference conditions at the reference point (10 cm • 10 cm field size, SSD = SAD). The dose has been determined with an ionization chamber. The institute dosimetry protocol has been used. 2O
3
t net Itute
18o~
mean S.l].
12-
delta c
protocol
beams
x-rag
number
0.9qs 0.011 0.040 3B
4-
Results ~.
3o ' o . ' ~ 2 ' o . ' ~ 4 ' o . ' ~ '
Dose determination at the refi, rence point
The dose measured at the reference point has been compared with the dose stated by the institute. No distinction has been made between institutes where the stated dose values were determined by the local physicist the same day as the visit took place and institutes where the stated dose values were derived from the tables used for the treatment of patients. Firstly, the distribution was assessed of the ratios of the stated to the measured dose values by the visiting team, analysed according to the institute protocol, for cobalt-60 beams. Good agreement was found between the stated and measured dose values (Fig. 3). A similar distribution for the X-ray beams is presented in Fig. 4. This distribution features a larger variance. Next, the distribution of the stated to the measured dose values was analysed according to the NCS code of practice. Results for cobalt-60 and X-r~iy beams are presented in Figs. 5 and 6, respectively.
0.'~8'
a.'oo ' 1.'o2 ' i . ' o 4 ' l.'oG ' 1.'o2 ' 1 . 2 o
stated absorbed does / measured absorbed close
Fig. 4. The distribution of the ratios of stated and measured dose values for megavoltage X-ray beams under reference conditions at the reference point (10 cm • 10 cm field size, S S D = S A D ) . The dose has been determined with an ionization chamber. The institute dosimetry protocol has been used. 2O
. Is
o ~ g
beams
cobalt-60
iz-
mean S.I].
1.006 0.006 0.020 11
delta " 8-
prect Ice
HCS Code o f
number
4-
~.~
i
i
,
i
i
j
J
i
J
.
0 92 0 94 0 96 0 ~a 1 00
o,
i
,
i 02 I 04
,
~
1 06
,
l JOB'
10
stated absorbed dose / measured absorbed dose
Fig 5 The distribution of the ratios of stated and measureddose wdues for cobalt-60 g a m m a - r a y beams under reference conditions at the reference point (10 cm x 10 cm field size, SSD = SAD). The dose has been determined with an ionization chamber. The NCS code of practice for the dosimetry of highenergy photon beams [10] has been used.
40 2o
30. .
16-
x-rs~
" 12-
mean S.O. dolts number
a
HCS Code o f
.
prectlce
besme
24.
0.992 0.020 O.OgS 40
~ o
measuring
polnt(o)
moan S.D. number
0.985 O.OIS 36
Isocontre
1B"
12"
B" .
G.
4-
0 , , i , l , l , l , l , l , l , l , l , l , I , 0.90 0 . 9 2 0 . 9 4 0.~6 0.98 1.00 1.02
stated
absorbed
dose
/
measured
, 1.04
o
, , , , 1.06 1.08 1.10
absorbed
0.90 ' O.b2'
dose
Fig. 6. The distribution of the ratios of stated and measured dose values for megavoltage X-ray beams under reference conditions at the reference point (10 cm x l0 cm field size, S S D - S A D ) . The dose has been determined with an ionization chamber. The NCS code of practice for the dosimetry of high-energy photon beams [10] has been used.
O.b4'
stated
O.bS'
1.'00 ' I.'02 ' 1.'04 ' 1.'06 ' 1.'08 ' 1 . 1 0
O.b9'
absorbed
dose
/
measured
absorbed
dose
Fig. 7. The distribution of the ratios of stated and measured dose values in the anthropomorphic phantom, irradiated for the treatment of a prostatic cancer. The measuring point is the isocentre. The dose has been determined with an ionization chamber.
Dose determination during the prostatic cancer treatment (anthropomorphic phantom)
SO
In eight institutes, the technique chosen for the prostatic treatment was simihtr to the prescribed technique, although small ditferences in field sizes and wedge angles were precent, in six institutes, a four field box technique was applied, while two centres applied a (partly) rotation technique. Three institutes irradiated the p h a n t o m only according to the prescribed technique. In one institute, only the dose at the reference point was measured. The dose values determined at the 10 points have been divided into four groups. G r o u p I consists of the values determined at the isocentre, where an ionization chamber was placed. G r o u p 2 includes all points inside and on the outline of the target volume, except the isocentre (points 2,3,4,7,8,9). G r o u p 3 consists of the data determined at the point in the bladder, outside the target volume but not positioned on a large dose gradient (point 6), and finally, group 4 summarizes the results in the points in the rectum lying on a large dose gradient (points 1 and 5). The frequency distributions of the ratios of the stated and measured dose values were assessed for the four groups. These distributions are presented in Figs. 7 10 for group 1 to 4, respectively. No distinction has been made between the prescribed technique and the institute technique, because the results for both situations were almost
30-
40-
@ L q-
measuring
polnt(s)
mean S.D. number
1.004 0.052 207
2 , 3 , 4 , 7 , B,9 --
20.
10-
o
,
O.eO
,
, , , 9 ,
0.84
stated
0.88
,
0.92
absorbed
,
,
O.gG dose
,
,
1.00 /
,
,
1.04
,
,
1.08
measured
,
,
1.12
absorbed
,
,
,
I.IG
,
1.20
dose
Fig. 8. The distribution of the ratios of stated and measured dose values in the anthropomorphic phantom, irradiated for Ills treatment of a prostatic cancer. The measuring points are positioned within or o n the outline of the target volume. The dose has been determined with T L D powder.
20. 9
meesur
1~. 12. g"
In g
mean S.D. number
polnt(s)
G
1.000 0.047
33
a. 4
01 0 80
,rTq
I 0 re1 = O 8 8 = 0 =92
stated
absorbed
0 9E
does
I
OO
1 04
/
measured
I
r--q
0 8 ' 1.=12 = l . l l g
absorbed
I 1.20
dose
Fig. 9. The distribution of the ratios of stated t o measured dose values in the anthropomorphic phantom, irradiated for the treatment of a prostatic cancer. The measuring point is positioned outside the target volume in the bladder. The dose has been determined with T L D powder.
41 30
meseurlng polnt(s)
9 24-
mean
0.9S9
S.D.
0.182
number
1,5
66
[B'. 12"
6.
0
0.00
,
,
,
0.20
i
,
0.40
0.~0
0.80
1,00
1.20
1.40
v
1.60
,
1.'80
i
2.00
etsted ~beorbed dose / msesured absorbed dose
Fig. 10. The distribution of the ratio of stated and measured dose values in the anthropomorphic phantom, irradiated for the treatment of a prostatic cancer. The measuring points arc positioned outside the target volume on a large dose gradient in the rectum. The dose has been determined with TLD powder. identical. Table V summarizes the mean and standard deviation of the ratio of stated and measured dose values for each point individually.
Discussion
Dose determination at the reference point Applying Ca values according to the institute protocol, the only uncertainties in the measurement of TABLE V The mean and standard deviation (S.D.) of the ratios of stated and measured dose values for thc isocentrc and nine points in the anthropomorphic phantom (see Figs 1 and 2), Group
Point
Mean
S.D.
Delta
1
lsocentre
0.985
0.015
0.064
2
2 3 4 7 8 9
1.007 0.975 1.016 0.994 1.033 1.001
0.035 0.047 0.060 0.048 0.060 0.041
0.166 0.264 0.256 0.196 0.235 0.177
3
6
1.000
0.047
0.238
4
1 5
0.964 0.952
0. !48 0.215
0.644 0.803
The phantom is irradiated for the treatment of a prostatic cancer. The dose at the isocentre is determined with an ionization chamber and at the other points with TLD.
the absorbed dose, when compared with the stated dose values, are introduced by the exposure calibration factor and the electrometer reading (see Eqn. 1). The standard deviation of the set of calibration factors obtained over the measurement period was 0.1% with a maximum deviation of 0.2% from the mean value. The uncertainty in the electrometer reading, including the uncertainty in the positioning of the ionization chamber and the water phantom and the uncertainties in the pressure, temperature and recombination correction factors is estimated to be 0.5% (1 S.D.). The measurements were carried out with the beam directed horizontally, whereas most institutes determine the output of their machines with a vertical beam. Due to the gantry angle dependence of the accelerator output, a systematic uncertainty is introduced which has been estimated to be 0.5% (1 S.D.). In addition, there is an uncertainty due to fluctuations in the output of the accelerator which is estimated to be 0.75% (1 S.D.). The experimental uncertainty by the local physicist in his determination of the dose is dflficult to quantify. Assumed is that his uncertainty is about the same as discussed here, with the except of the uncertainty due to the tluctuations in the output of the accelerator. Taking the root of the sum of squares, the total uncertainty in the dose measurement, excluding the uncertainty in the Cx value, is therefore 0.7% (1 S.D.) for the cobalt-60 beams and 1.3% (1 S.D.) for the X-ray beams. Comparing these estimated uncertainties with the observed standard deviations in the distributions presented in Figs. 3 and 4 (0.5% and 1.1% for the cobalt-60 and X-ray beams, respectively) indicates that the experimental uncertainty is the main reason for the observed differences. Instead of following the same dosimetry protocols as applied by the institutes, a better estimate of the real differences in absorbed dose values in water at the reference point is obtained by adoption only one dosimetry protocol for the measurements of the visiting team. The results presented in Figs. 5 and 6 have been obtained by applying the NCS code of practice [10] for the dose measurement. A comparison of Fig. 3 with Fig. 5 and of Fig. 4 with Fig. 6 shows that the mean deviation is only slightly
42 changed. A more important change can, however, be observed in the frequency distributions presented in Figs. 4 and 6. The distribution is broader in Fig. 6 compared to Fig. 4, which is reflected in a larger standard deviation and a larger m a x i m u m difference between the photon beams, indicated in the figures as delta. For the X-ray beams the maximum difference is 9.5%. Such a difference in dose may be received by patients treated with these beams. If all institutes apply the same dosimetry protocol this maximum difference will be about 4%, as can be derived from the delta value given in Fig. 4. A comparison of our results, for the dose determination at the reference point, as presented in Figs. 5 and 6, with those summarised in Table l shows that the mean deviation for the cobalt-60 gamma-rays is very small, although in Table I, for some institutes considerable deviations are observed [7]. For X-ray beams the situation in The Netherlands is comparable to that in Scandinavia [6] but slightly better than the results obtained by Johansson et al. [7] during the dose intercomparison between the European institutes participating in E O R T C trials.
Dose determination during the prostatic cancer treatment (anthropomorphic phantom) Although the anthropomorphic p h a n t o m was placed in the treatment room a few hours before
the measurements were performed, there is an uncertainty in the temperature of the phantom. The r o o m temperature was taken for the temperature correction of the ionization chamber signal (open chamber). The ionization chamber reading was converted to absorbed dose to water applying Eqn. (2) (NCS code of practice). The resulting uncertainty in this conversion, excluding the uncertainty in the Cw,u value, is estimated at 0.9% (1 S.D.). In addition, there is an uncertainty ( < 1.0%) in the dose determination due to the difference in composition between water and the tissue equivalent material of the a n t h r o p o m o r p h i c phantom. The standard deviation in the frequency distribution of the stated to measured dose values in the anthropomorphic p h a n t o m at the isocentre is 1.5%. For the 36 beams used to irradiate this phantom, almost the same standard deviation is found in the distribution at the reference point in the water phantom (see Table VI). Correcting the measured dose at the isocentre for the difference found between the stated and measured dose value at the reference point did not change the result. In nine points (black dots in Figs. 1 and 2) the dose was determined with T L D giving two readings per point. The estimated uncertainty in the T L D dose determination is 2.5% (1 S.D.). For the six points within and on the outline of the target volume (group 2) the mean deviation between the stated and the measured dose values is only 0.4%. However, the standard deviation of the dose distri-
TABLE VI Summary of results of the dose intercomparison at the radiotherapy centres in The Netherlands for the anthropomorphic phantom. Type of irradiation
Reference Prostatic turnout Prostatic tumour Prostatic turnout Prostatic tumour
Measuring conditions
Results
Phantom
Type of dosimeter
Position
Mean
S.D.
Delta
Water Alderson Alderson Alderson Alderson
Ionisation chamber Ionisation chamber TLD TLD TLD
Reference Isocentre Target volume Bladder Rectum
0.994 0,985 1.004 1.000 0.959
0.013 0.015 0.052 0.047 0.182
0.045 0.064 0.30 0.24 0.80
The mean and standard deviation (S.D.) of the ratios of stated and measured dose values and the largest difference observed between the ratios (delta) are given. Also the results of the intercomparison at the reference point in water, restricted to the 36 photon beams used to irradiate the anthropomorphic phantom, are given.
43 bution at these points is 5.2%. (Fig. 8). The mean mass density of the bony structures in the two planes were determined with CT slices. The mean mass density in the central plane was 0.9 g/cm 3 and in the off-axis plane 1.0 g/cm 3. These prescribed densities were stated in the contours. CT information was not applied in any dose calculation by one of the institutes. Deviations in the density of the bone structures from the prescribed mean mass density will have a systematic effect on the results. This refers especially to point 3 where for the two lateral beams more than half of the material in front of point 3 consists of "bone material". As can be seen from Table V, the dose measured in point 3 is 2.5% higher than stated, which might indicate the presence of inhomogeneities with an average mass density lower than 0.9 g 9 cm 3 in the bone material in front of point 3. In addition, in one institute the dose in the marked points, situated in the central plane (isocentre and points 2,3,4), was recalculated using CT data. Normalised to the dose in the isocentre, the dose calculated in points 3 was about 3% higher in comparison with the dose calculated using the average mass density of 0.9 g/cm 3 of the bone material. This increase of dose in point 3, due to the low density bone material corresponds approximately to the dose deviation observed in point 3. Points 2,4,7 and 8 are positioned on the outline of the target volume. Examining the isodose plots for the prescribed and the institute techniques shows that point 2 and 7 are situated well inside the treatment volume while points 4 and 8 are situated on the outline of the treatment volume near large dose gradients. Small uncertainties in the set up of the phantom and treatment unit will therefore have more effect on the measured dose at points 4 and 8 than at points 2 and 7. It can be seen from Table V that the standard deviations in the ratios of the stated and measured dose for the points 4 and 8 are indeed larger than for points 2 and 7. The 3.3% deviation between the stated and measured dose in point 8, the latter being lower, indicates that the real position of point 8 in the phantom might be a little bit further from the isocentre. The dose at point 6 is only slightly influenced by
uncertainties in the density of the bone structures. The mean ratio of measured and stated dose values is 1.000 (Table V), while the standard deviation is 4.7%. This indicates that the dose determination in the bladder point can be performed with about the same accuracy as in points in the target volume. The average difference between the stated and measured dose in points 1 and 5 in the rectum is about 4%, the latter being higher. These data have, however, a larger standard deviation and delta value (see Tables V and VI), due to the relatively steep dose gradient at these points. The results of this part of the dose intercomparison can be compared only to a certain extent with those of others as presented in Table II, because the non-reference conditions differ for all these intercomparisons. Only with the Alderson phantom measurements of Johansson et al. [8] is a more direct comparison possible. For the combined results of two points inside the target volume, the isocentre and point 9, the distribution of the ratio of stated and measured dose values, reflected in the standard deviation (3.1%) and the maximum difference between two ratios (18%) is almost similar to the distribution observed by Johansson et al. [8] for a tonsil tumour treatment (see Table II). Conclusions Deviations between the prescribed dose and the dose delivered to a patient can arise during all steps in the dosimetry chain. Dose intercomparisons as reported here might reveal some of these deviations and allow quantifications of uncertainties in some of the steps in the dosimetry chain. Differences between stated and measured dose under reference conditions in a water phantom were on the average 0.5% with a standard deviation of 1.9%, but up to 6% at maximum. If all institutes apply the same set of recommended values for the physical quantities, this maximum difference will reduce to about 2%, thus showing the importance of using a common dosimetry protocol. It indicates in addition that no major systematic uncertainties are introduced in the dose determination, as discussed by others [4,11,12].
44 A comparison between stated and measured dose at the isocentre during the prostatic tumour treatment of the anthropomorphic phantom shows that the deviation is about the same as during the water phantom irradiation under reference circumstances. This very good agreement indicates that no additional systematic uncertainties are introduced in the dose calculation at the institutes, e.g. in the applied wedge factor. It must, however, be pointed out that in this intercomparison institutes were asked to give a specified dose to the isocentre and not to the tumour or target volume. The latter method introduces additional dose differences due to different methods of dose specification applied by the institutes. The difference between stated and measured dose in the points situated in the target volume is on the average 0.4% with a standard deviation of 5.2%. This increase in standard deviation compared to the results for the isocentre can only partly be explained by using T L D instead of an ionization chamber. Other reasons are differences in the actual dose distribution and those available in the computer planning systems and possible limitations of dose calculation algorithms. Quantification of these deviations needs further study.
Acknowledgements The authors would like to express their gratitude to the radiotherapy centres in The Netherlands for their extensive cooperation and hospitality during this intercomparison. We also wish to thank B. G6bel for the preparation and read-out of the T L D powder. This work was financially supported by The Netherlands Cancer Foundation, K.W.F. Grant N.K.I. 84-12.
References 1 Brahme, A. Dosimetric precision requirements in radiation therapy. Acta Radiol. Oncol. 23:379 391, 1984. 2 Dutreix, A. The French dosimetry protocol. Radiother. Oncol. 4: 301-304, 1985. 3 1CRU report 14. Radiation dosimetry: X rays and gamma
4
5
6
7
8
9
10
I1
12
13
14
15 16
rays with maximum photon energies between 0.6 and 50 MeV. International Commission on Radiological Units and Measurements, 4201 Connecticut Avenue, N.W., Washington, D.C. 20008, U.S.A., 1969. ICRU report 24. Determination of absorbed dose in a patient irradiated by means of X or gamma rays in radiotherapy procedures. International Commission on Radiological Units and Measurements, 7910 Woodmont Avenue, Washington, D.C. 20014, U.S.A., 1976. Johansson, K. A. Studies of different methods of absorbed dose determination and a dosimetric intercomparison at the Nordic radiotherapy centres. Thesis University of Gothenburg, Sweden, 1982. Johansson, K. A., Mattsson, L. O. and Svensson, H. Dosimetric intercomparison at the Scandinavian radiation therapy centres. 1. Absorbed dose intercomparison. Acta Radiol. Ther. Phys. Biol. 21:1 10, 1982. Johansson, K. A., Horiot, J. C., Van Dam, J., Lepinoy, D., Sentenac, 1. and Sernbo, G. Quality assurance control in the EORTC cooperative group of radiotherapy. 2. Dosimetric intercomparison. Radiother. Oncol. 7: 269---279,. 1986. Johansson, K. A., Horiot, J. C. and van der Schueren, E. Quality assurance control in the EORTC cooperative group of radiotherapy. 3. lntercomparison in an anatomical phantom. Radiother. Oncol., Submined, 1986. Mijnheer, B. J., Battermann, J. J. and Wambersie, A. Which accuracy is required and can be achieved in clinical photon and neutron dosimetry?. Radiother. Oncol. Mijnheer, B. J., Aalbers, A. H. L., Visser, A. G. and Wittk~imper, F. W. Consistency and simplicity in the determination of absorbed dose to water in high-energy photon beams: a new code of practice. Radiothcr. Oncol. 7:371 384, 1986. Mijnheer, B. J., Wittk~imper, F. W., Aalbcrs, A. H. L. and van Dijk, E. Experimental verification of the air kerma to absorbed dose conversion t:actor C~.~. Radiother. Oncol. 8: 49-56, [987. NACP (Nordic Association of Clinical Physics). Procedures in external radiation therapy between 1 and 50 MeV. Acta Radiol. Oncol. 19: 55--79, 1980. Samulski, T., Dubuque, G. L., Cacak, R. K., Courlas, G., De Werd, L. A., Hilko, R., Humphries, L. J., Jones, D., Masterson, M. E., Miller, D. W,, Stovall, M. and Wochos, J.F. Radiation therapy dosimelry reviews by the centers [k~r radiological physics. Int. J. Radiat. Oncol. Biol. Phys. 7: 379 383, 1981. Shalek, R. J., Kennedy, P., Stovall, M., Cundiff, J. H., Gagnon, W. F., Grant IIl, W. and Hanson, W. F. Quality assurance for measurements in therapy. N.B.S. SP 457, 1976. Somerwil, A. and van Kleffens, H. J. Experience with the Alderson Rando Phantom. Br. J. Radiol. 50: 295-296, 1977. Svensson, H. Dosimetric measurements at the nordic medical accelerators, ll. Absorbed dose measurements. Acta Radiol. Ther. Phys. Biol. 10:631 654, 1971.
33
RTO 00329
Dose intercomparison at the radiotherapy centres in The Netherlands. 1. Photon beams under reference conditions and for prostatic cancer treatment F . W . W i t t k / i m p e r l'z, B. J. M i j n h e e r 1 a n d H . J. v a n K l e f f e n s 2 1Radiotherapy Department, The Netherlands Cancer Institute (Antoni van Leeuwenhoekhuis), 1066 CX Amsterdam, The Netherlands, and 2Physics Department, Rotterdam Radio-Therapeutic Institute (Dr. Danii~l den Hoed kliniek ), 3075 EA Rotterdam, The Netherlands
(Received 13 September 1986, revision received 4 December 1986, accepted 23 December 1986)
Key words." Quality assurance in radiotherapy; Dose intercomparison; Photon beam
Summary In 1985, a dosimetry intercomparison was performed at all 20 radiotherapy centres in The Netherlands. Absorbed dose was determined with an ionization chamber under reference conditions in a water phantom for cobalt-60 g a m m a - r a y and megavoltage X-ray beams. The mean difference between measured and stated dose values was 0.5% with a standard deviation of 1.9%, but up to 6% at maximum. As soon as all institutes apply a common dosimetry protocol, this maximum difference will reduce to about 2%. In addition, an anthropomorphic phantom was irradiated to simulate the treatment of a prostatic cancer. The dose, determined with an ionization chamber at the isocentre and thermoluminescent dosimeters (TLD) powder at several points situated in the target volume, the bladder and the rectum, was compared with the stated dose calculated with the local planning system. Only small differences were found between the measured and stated dose at the isocentre: on the average 1.5%, with a standard deviation of 1.5%. The difference between stated and measured dose at several points situated in the target volume was on the average 0.4%, with a standard deviation of 5.2%. Almost the same result was found for a point situated in the bladder. In the rectum, the average difference was about 4%, however, with a large standard deviation, 18%, due to the relatively steep dose gradient at these points.
Introduction During the last decade, growing attention has been paid to quality control in radiation therapy. This Address for correspondence: F. W. Wittkiimper, Radiotherapy
Department, The Netherlands Cancer Institute (Antoni van Leeuwenhoekhuis), 1066 CX Amsterdam, The Netherlands.
is necessary because small changes in the dose of radiation delivered to a target volume in a homogeneous group of patients may drastically influence the probability of tumour control or the probability of severe normal tissue reactions. This effect is quantified in recent surveys of dose-effect curves for local tumour control and normal tissue comptica-
0167-8140/87/$03.50 9 1987 Elsevier Science Publishers B.V. (Biomedical Division)
34 tions as given by Brahme [1], Johansson [5] and Mijnheer et al. [9]. Each step in the dosimetry chain, from the calibration of the dosimeter to the delivery of radiation to a patient, contributes to the final uncertainty in the actual dose given to a patient. The five main steps in this chain are: (1) The calibration of the dosimetry equipment; (2) the dose determination at a reference point in a water phantom under reference conditions; (3) the determination of the relative dose distribution in a phantom under non-reference conditions, including corrections for oblique incidence, wedge filters etc.; (4) the assessment of the dose at relevant points in a patient; (5) the delivery of the radiation to the patient under treatment conditions. In principle, there are two ways to estimate the total uncertainty in the dose given to a patient: by in vivo dosimetry and by estimating the uncertainties in each of the steps in the dosimetry chain. The use of in vivo dosimetry is an ultimate check of the treatment but is difficult to perform in a systematic way. Methods have been described in 1CRU Report 24 [4]. Dose intercomparisons can be used to quantify uncertainties in the dosimetry chain. A number of dose intercomparisons have been carried out in order to investigate whether the procedures used in different radiotherapy centres, often based on different protocols, result in a satisfying agreement in dose determination [5-8,13,14,16]. Table I summarises the results of some dose intercomparisons related to step 2, in which the measurements were performed with an ionization chamber, whereas Table ll summarises the results of dose intercomparisons related to step 4 or 5. These data show that considerable differences can be observed between dose values stated by an institute and that measured by a visiting team. The purpose of the dose intercomparison reported here is to quantify the differences betweeen stated and measured dose values for megavoltage photon beams under reference conditions (step 2) and
to compare stated and measured dose values in an anthropomorphic phantom under treatment conditions (step 5). It should be noticed that when irradiating an anthropomorphic phantom, instead of a patient, not all parameters involved in step 5 are taken into account (e.g. patient movement during irradiation). Therefore, uncertainties due to replacement of a patient by an anthropomorphic phantom have to be added to the observed dose differences. In a separate paper, the results of measured relative dose distributions (step 3) will be compared with calculated dose distributions.
Methods and materials
The dosimetry intercomparison took place from March to December 1985. The measurements were performed by the same physicist and with the same equipment. All 20 radiotherapy centres were visited once at an agreed date. The dose at the reference point in a water phantom was measured for all megavoltage photon beams available in the institutes (see Table IlI). The anthropomorphic phantom measurements were performed in each institute only at one photon beam; the photon beam normally used in that particular institute to irradiate a prostatic tumour. The measurements, with this phantom were carried out according to a technique prescribed and a technique chosen by the institute. Dose determination at the reJ'erence point The absorbed dose to water at the reference point was measured in a water phantom with a graphite-walled thimble ionization chamber (Type NE* 2505/3A). The water phantom was irradiated horizontally under reference conditions: field size 10 cm x 10 cm at the surface, and the source-surface distance (SSD), equal to the source-axis (isocentre) distance (SAD). The reference point was situated on the central axis of the beam at 5 cm depth for energies up to 15 MV and at 7 cm depth for energies between 16 and 25 MV. The geometrical centre of * Nuclear Enterprises Ltd.
35 8
d d ~ d
d
I
d d d d
d d d d
dd
go
~.o..~.
~. ~. ~ . ~ . ~ . ~ .
~.~..
~.~.
>
g 0 "0
•
8~aa
X88
r. V A
~L 0
0
.g
ca t..,
a
e.~
>
.g
g
0
f.-
8 8
.g
.g d 0
4
4
~
~
z
$
3
r~
rm 0 o~
.g
~J
<
. ~ =
~=
;8
.g
36
,a
e~
gg
~'
lea
ir
gg
g,
9
a< 6 g
e~
g
o
g 9
.-=
.~
<
<
.-
g.g
o_ ~ ' =-
-~ =
0
~4 0 ".,~_~
a
<
~"
o
.g
.--2,.
o
0
9
<
37 TABLE III Survey of the number of cobalt-60 teletherapy units and megavolt linear accelerators available for radiotherapy on January 1, 1986 in The Netherlands. No. of photon beams Cobalt-60
11
4 MV
5
5 MV 6 MV 8 MV 10 MV 12 MV 15 MV 16 MV 18 MV 23 MV 25 MV
4 5 15 2 1 l 4 1 1 1
Total: spread over:
51 photon beams 11 cobalt-60 teletherapy units 35 linear accelerators
(2)
Dw = M - NK - Cw,u
the ionization chamber was positioned at the reference point. The measurements were converted to absorbed dose to water in two ways. Firstly, by using the equation: Dw = N x . R . C~
summarises the different photon beam dosimetry protocols applied in 1985 in The Netherlands. Secondly, the measurements were converted to absorbed dose to water applying the equation given in a recent code of practice for the dosimetry of high-energy photon beams [10] which is going to be used in The Netherlands (NCS code of practice):
(1)
in which Dw is the absorbed dose to water, Nx is the exposure calibration factor, R is the electrometer reading corrected for temperature and pressure, but not for recombination and Cz is the exposure to absorbed dose to water conversion factor. The ionization chamber was calibrated against a secondary ionization chamber calibrated at the Dutch Standards Laboratory, according to the old exposure value, to which the chambers of all Dutch institutes are still calibrated. For each institute, the Cz values were adopted normally used by that particular institute (institute protocol). Some institutes are using dosimetry protocols in which several conversion and correction factors are given separately, e.g. the N A C P protocol [12]. All these factors, not related to the instrument reading, have been combined to one C~ value for these protocols. Table IV
where Dw is the absorbed dose to water, M is the electrometer reading corrected for temperature, pressure and recombination, ArK is the air kerma calibration factor and Cw.u is the air kerma to absorbed dose to water conversion factor. In this code of practice Cw,u values for reference ionization chambers (including the N E 2505/3A) are given as a function of the quality index; the ratio of the ionizations measured at 20 cm and 10 cm depth in water for a fixed SDD, and aficld size of 10 cm x 10 cm at the geometrical centre of the detector [2]. The quality indices of the accelerator beams were derived from the stated central axis depth dose values at 10 cm and 20 cm depth in water for a 10 cm x 10 cm field size at 100 cm SSD, by using a semi-empirical relation between the ratio of these P D D values and the quality index [11]. The air kerma calibration factor NK was calculated from the exposure calibration factor Nx, applying the recently revised value for exposure at the Dutch Standards Laboratory, using the equation:
N~ = Nx" W/e. (1 - g ) in which
W/e is the
1
(3)
quotient of the average energy
TABLE IV Survey of the different dosimetry protocols in use on January 1, 1986 at the radiotherapy centres in The Netherlands. Dosimetry protocol
Number
ICRU (1969) [3] NACP (1980) [12] Other protocols Fricke dosimetry
12 3 4 1
38 expended to produce an ion pair in dry air and the electron charge and g is the fraction of energy of secondary charged particles that is converted to bremsstrahlung in air. For g, a value of 0.003 was applied and for W/e a value o f 33.97 J/C according to the recommendations given in the code of practice. Although the value for the exposure at the Standards Laboratory is revised, the consequence for Dw, because of this change is minor. The change in the exposure value is compensated by an almost equal change in the W/e value. The chamber was recalibrated against a secondary standard ionization chamber in a cobalt-60 g a m m a - r a y beam shortly before and after each visit to an institute. N o changes in the exposure calibration factor larger than 0.2% were observed.
Dose determination during the prostatic cancer treatment (anthropomorphic phantom) In each institute, an anthropomorphic phantom (type Alderson Rando) was irradiated as for treatment of a prostatic cancer according to a technique prescribed and a technique chosen by the institute. Some weeks before the visit, two sheets with a contour of the anthropomorphic phantom were sent to the institute. One contour was taken at the centre of the prostate and one contour was taken 2.5 cm off-axis (see Figs. 1 and 2). The target volume, with ~ lug.
p~n~om a4
a,1 central
p~ne
Fig. I. Contour of the anthropomorphic phantom in the central plane of the field. The target volume (-.-.-) and the outlines of the prostate, the rectum, the bone structures and four measuring points (0) are indicated.
9 u ~ ~ a3
oft-axis
plene
2.sere
e
lder
7
Fig. 2. Contour of the anthropomorphic phantom in the 2.5 cm off-axis plane of the field. The target volume (-.-.-) and the outlines of the vesiculaseminalis, the rectum, the bladder, the bone structures and five dose points (0) are indicated. the prostate and the vesicula seminalis, the rectum, the bladder and the bone structures were marked in the contours. The prescribed technique consisted of a three-field technique using an AP beam and two opposing lateral wedged beams. A wedge with an angle of 45 ~ or the nearest wedge angle awlilable, was requested. The isocentre of the prescribed technique, the entry points of the three beams and nine specific points for the dose intercomparison were also marked on the contours. CT pictures were taken from the central plane and the 2.5 cm off-axis plane. Large inhomogeneities in the bone structures were observed. CT numbers varying from - 6 0 0 to + 200 were found, which correspond roughly to a mass density of 0.4 g/cm 3 and 1.12 g/cm 3, respectively. In the central plane, the average mass density was 0.9 g/cm -~ and in the off-axis plane 1.0 g/cm 3. These large inhomogeneities in the bony material of this anthropomorphic p h a n t o m was also reported by Somerwil and van Kleffens [15]. The institutes were asked: (a) to give a dose of 2.5 Gy to the point indicated as the isocentre of the prescribed technique; (b) to calculate with their computer planning system the dose in the nine points. The dose at the isocentre was determined with a small ionization chamber and in the nine other points with thermoluminescent dosimeters (TLD). The ionization chamber has a 0.5 m m thick polystyrene wall and cavity dimensions o f 12 m m
39 (length) and 4 mm (diameter). The chamber was calibrated in a water phantom, irradiated with photon beams from cobalt-60 to 25 MV, against the calibrated graphite-walled ionization chamber applying twice Eqn. (2). As TLD material, 12 mg of powdered TLD-700 was used in each point, yielding two readings. The TLD was read out using a Pitman model 654 TLD reader, applying 7 s pre read-out heating at 135~ followed by 6 s at a temperature of 240~ Annealing was performed by 1.5 h at 400~ followed by a natural cooling down in the oven (Pitman model 622/B) till 80~ and finally the powder was kept during 16 h at this temperature. The TLD was recularly calibrated in a 4 MV photon beam. Measurements at other energies showed that no photon energy dependence could be observed in the calibration. In the dose obtained with TLD, a supra-linearity correction has been applied.
20
Institute
=
protocol
c o b a l t - S O beams
16-
mean S.9.
12-
1.000
O.OOS 0.018 10
delta :3
number
4-
0 0.90
stated absorbed dose / measured absorbed dose
Fig. 3. The distribution of the ratios o f stated and measured dose values for cobalt-60 g a m m a - r a y beams under reference conditions at the reference point (10 cm • 10 cm field size, SSD = SAD). The dose has been determined with an ionization chamber. The institute dosimetry protocol has been used. 2O
3
t net Itute
18o~
mean S.l].
12-
delta c
protocol
beams
x-rag
number
0.9qs 0.011 0.040 3B
4-
Results ~.
3o ' o . ' ~ 2 ' o . ' ~ 4 ' o . ' ~ '
Dose determination at the refi, rence point
The dose measured at the reference point has been compared with the dose stated by the institute. No distinction has been made between institutes where the stated dose values were determined by the local physicist the same day as the visit took place and institutes where the stated dose values were derived from the tables used for the treatment of patients. Firstly, the distribution was assessed of the ratios of the stated to the measured dose values by the visiting team, analysed according to the institute protocol, for cobalt-60 beams. Good agreement was found between the stated and measured dose values (Fig. 3). A similar distribution for the X-ray beams is presented in Fig. 4. This distribution features a larger variance. Next, the distribution of the stated to the measured dose values was analysed according to the NCS code of practice. Results for cobalt-60 and X-r~iy beams are presented in Figs. 5 and 6, respectively.
0.'~8'
a.'oo ' 1.'o2 ' i . ' o 4 ' l.'oG ' 1.'o2 ' 1 . 2 o
stated absorbed does / measured absorbed close
Fig. 4. The distribution of the ratios of stated and measured dose values for megavoltage X-ray beams under reference conditions at the reference point (10 cm • 10 cm field size, S S D = S A D ) . The dose has been determined with an ionization chamber. The institute dosimetry protocol has been used. 2O
. Is
o ~ g
beams
cobalt-60
iz-
mean S.I].
1.006 0.006 0.020 11
delta " 8-
prect Ice
HCS Code o f
number
4-
~.~
i
i
,
i
i
j
J
i
J
.
0 92 0 94 0 96 0 ~a 1 00
o,
i
,
i 02 I 04
,
~
1 06
,
l JOB'
10
stated absorbed dose / measured absorbed dose
Fig 5 The distribution of the ratios of stated and measureddose wdues for cobalt-60 g a m m a - r a y beams under reference conditions at the reference point (10 cm x 10 cm field size, SSD = SAD). The dose has been determined with an ionization chamber. The NCS code of practice for the dosimetry of highenergy photon beams [10] has been used.
40 2o
30. .
16-
x-rs~
" 12-
mean S.O. dolts number
a
HCS Code o f
.
prectlce
besme
24.
0.992 0.020 O.OgS 40
~ o
measuring
polnt(o)
moan S.D. number
0.985 O.OIS 36
Isocontre
1B"
12"
B" .
G.
4-
0 , , i , l , l , l , l , l , l , l , l , l , I , 0.90 0 . 9 2 0 . 9 4 0.~6 0.98 1.00 1.02
stated
absorbed
dose
/
measured
, 1.04
o
, , , , 1.06 1.08 1.10
absorbed
0.90 ' O.b2'
dose
Fig. 6. The distribution of the ratios of stated and measured dose values for megavoltage X-ray beams under reference conditions at the reference point (10 cm x l0 cm field size, S S D - S A D ) . The dose has been determined with an ionization chamber. The NCS code of practice for the dosimetry of high-energy photon beams [10] has been used.
O.b4'
stated
O.bS'
1.'00 ' I.'02 ' 1.'04 ' 1.'06 ' 1.'08 ' 1 . 1 0
O.b9'
absorbed
dose
/
measured
absorbed
dose
Fig. 7. The distribution of the ratios of stated and measured dose values in the anthropomorphic phantom, irradiated for the treatment of a prostatic cancer. The measuring point is the isocentre. The dose has been determined with an ionization chamber.
Dose determination during the prostatic cancer treatment (anthropomorphic phantom)
SO
In eight institutes, the technique chosen for the prostatic treatment was simihtr to the prescribed technique, although small ditferences in field sizes and wedge angles were precent, in six institutes, a four field box technique was applied, while two centres applied a (partly) rotation technique. Three institutes irradiated the p h a n t o m only according to the prescribed technique. In one institute, only the dose at the reference point was measured. The dose values determined at the 10 points have been divided into four groups. G r o u p I consists of the values determined at the isocentre, where an ionization chamber was placed. G r o u p 2 includes all points inside and on the outline of the target volume, except the isocentre (points 2,3,4,7,8,9). G r o u p 3 consists of the data determined at the point in the bladder, outside the target volume but not positioned on a large dose gradient (point 6), and finally, group 4 summarizes the results in the points in the rectum lying on a large dose gradient (points 1 and 5). The frequency distributions of the ratios of the stated and measured dose values were assessed for the four groups. These distributions are presented in Figs. 7 10 for group 1 to 4, respectively. No distinction has been made between the prescribed technique and the institute technique, because the results for both situations were almost
30-
40-
@ L q-
measuring
polnt(s)
mean S.D. number
1.004 0.052 207
2 , 3 , 4 , 7 , B,9 --
20.
10-
o
,
O.eO
,
, , , 9 ,
0.84
stated
0.88
,
0.92
absorbed
,
,
O.gG dose
,
,
1.00 /
,
,
1.04
,
,
1.08
measured
,
,
1.12
absorbed
,
,
,
I.IG
,
1.20
dose
Fig. 8. The distribution of the ratios of stated and measured dose values in the anthropomorphic phantom, irradiated for Ills treatment of a prostatic cancer. The measuring points are positioned within or o n the outline of the target volume. The dose has been determined with T L D powder.
20. 9
meesur
1~. 12. g"
In g
mean S.D. number
polnt(s)
G
1.000 0.047
33
a. 4
01 0 80
,rTq
I 0 re1 = O 8 8 = 0 =92
stated
absorbed
0 9E
does
I
OO
1 04
/
measured
I
r--q
0 8 ' 1.=12 = l . l l g
absorbed
I 1.20
dose
Fig. 9. The distribution of the ratios of stated t o measured dose values in the anthropomorphic phantom, irradiated for the treatment of a prostatic cancer. The measuring point is positioned outside the target volume in the bladder. The dose has been determined with T L D powder.
41 30
meseurlng polnt(s)
9 24-
mean
0.9S9
S.D.
0.182
number
1,5
66
[B'. 12"
6.
0
0.00
,
,
,
0.20
i
,
0.40
0.~0
0.80
1,00
1.20
1.40
v
1.60
,
1.'80
i
2.00
etsted ~beorbed dose / msesured absorbed dose
Fig. 10. The distribution of the ratio of stated and measured dose values in the anthropomorphic phantom, irradiated for the treatment of a prostatic cancer. The measuring points arc positioned outside the target volume on a large dose gradient in the rectum. The dose has been determined with TLD powder. identical. Table V summarizes the mean and standard deviation of the ratio of stated and measured dose values for each point individually.
Discussion
Dose determination at the reference point Applying Ca values according to the institute protocol, the only uncertainties in the measurement of TABLE V The mean and standard deviation (S.D.) of the ratios of stated and measured dose values for thc isocentrc and nine points in the anthropomorphic phantom (see Figs 1 and 2), Group
Point
Mean
S.D.
Delta
1
lsocentre
0.985
0.015
0.064
2
2 3 4 7 8 9
1.007 0.975 1.016 0.994 1.033 1.001
0.035 0.047 0.060 0.048 0.060 0.041
0.166 0.264 0.256 0.196 0.235 0.177
3
6
1.000
0.047
0.238
4
1 5
0.964 0.952
0. !48 0.215
0.644 0.803
The phantom is irradiated for the treatment of a prostatic cancer. The dose at the isocentre is determined with an ionization chamber and at the other points with TLD.
the absorbed dose, when compared with the stated dose values, are introduced by the exposure calibration factor and the electrometer reading (see Eqn. 1). The standard deviation of the set of calibration factors obtained over the measurement period was 0.1% with a maximum deviation of 0.2% from the mean value. The uncertainty in the electrometer reading, including the uncertainty in the positioning of the ionization chamber and the water phantom and the uncertainties in the pressure, temperature and recombination correction factors is estimated to be 0.5% (1 S.D.). The measurements were carried out with the beam directed horizontally, whereas most institutes determine the output of their machines with a vertical beam. Due to the gantry angle dependence of the accelerator output, a systematic uncertainty is introduced which has been estimated to be 0.5% (1 S.D.). In addition, there is an uncertainty due to fluctuations in the output of the accelerator which is estimated to be 0.75% (1 S.D.). The experimental uncertainty by the local physicist in his determination of the dose is dflficult to quantify. Assumed is that his uncertainty is about the same as discussed here, with the except of the uncertainty due to the tluctuations in the output of the accelerator. Taking the root of the sum of squares, the total uncertainty in the dose measurement, excluding the uncertainty in the Cx value, is therefore 0.7% (1 S.D.) for the cobalt-60 beams and 1.3% (1 S.D.) for the X-ray beams. Comparing these estimated uncertainties with the observed standard deviations in the distributions presented in Figs. 3 and 4 (0.5% and 1.1% for the cobalt-60 and X-ray beams, respectively) indicates that the experimental uncertainty is the main reason for the observed differences. Instead of following the same dosimetry protocols as applied by the institutes, a better estimate of the real differences in absorbed dose values in water at the reference point is obtained by adoption only one dosimetry protocol for the measurements of the visiting team. The results presented in Figs. 5 and 6 have been obtained by applying the NCS code of practice [10] for the dose measurement. A comparison of Fig. 3 with Fig. 5 and of Fig. 4 with Fig. 6 shows that the mean deviation is only slightly
42 changed. A more important change can, however, be observed in the frequency distributions presented in Figs. 4 and 6. The distribution is broader in Fig. 6 compared to Fig. 4, which is reflected in a larger standard deviation and a larger m a x i m u m difference between the photon beams, indicated in the figures as delta. For the X-ray beams the maximum difference is 9.5%. Such a difference in dose may be received by patients treated with these beams. If all institutes apply the same dosimetry protocol this maximum difference will be about 4%, as can be derived from the delta value given in Fig. 4. A comparison of our results, for the dose determination at the reference point, as presented in Figs. 5 and 6, with those summarised in Table l shows that the mean deviation for the cobalt-60 gamma-rays is very small, although in Table I, for some institutes considerable deviations are observed [7]. For X-ray beams the situation in The Netherlands is comparable to that in Scandinavia [6] but slightly better than the results obtained by Johansson et al. [7] during the dose intercomparison between the European institutes participating in E O R T C trials.
Dose determination during the prostatic cancer treatment (anthropomorphic phantom) Although the anthropomorphic p h a n t o m was placed in the treatment room a few hours before
the measurements were performed, there is an uncertainty in the temperature of the phantom. The r o o m temperature was taken for the temperature correction of the ionization chamber signal (open chamber). The ionization chamber reading was converted to absorbed dose to water applying Eqn. (2) (NCS code of practice). The resulting uncertainty in this conversion, excluding the uncertainty in the Cw,u value, is estimated at 0.9% (1 S.D.). In addition, there is an uncertainty ( < 1.0%) in the dose determination due to the difference in composition between water and the tissue equivalent material of the a n t h r o p o m o r p h i c phantom. The standard deviation in the frequency distribution of the stated to measured dose values in the anthropomorphic p h a n t o m at the isocentre is 1.5%. For the 36 beams used to irradiate this phantom, almost the same standard deviation is found in the distribution at the reference point in the water phantom (see Table VI). Correcting the measured dose at the isocentre for the difference found between the stated and measured dose value at the reference point did not change the result. In nine points (black dots in Figs. 1 and 2) the dose was determined with T L D giving two readings per point. The estimated uncertainty in the T L D dose determination is 2.5% (1 S.D.). For the six points within and on the outline of the target volume (group 2) the mean deviation between the stated and the measured dose values is only 0.4%. However, the standard deviation of the dose distri-
TABLE VI Summary of results of the dose intercomparison at the radiotherapy centres in The Netherlands for the anthropomorphic phantom. Type of irradiation
Reference Prostatic turnout Prostatic tumour Prostatic turnout Prostatic tumour
Measuring conditions
Results
Phantom
Type of dosimeter
Position
Mean
S.D.
Delta
Water Alderson Alderson Alderson Alderson
Ionisation chamber Ionisation chamber TLD TLD TLD
Reference Isocentre Target volume Bladder Rectum
0.994 0,985 1.004 1.000 0.959
0.013 0.015 0.052 0.047 0.182
0.045 0.064 0.30 0.24 0.80
The mean and standard deviation (S.D.) of the ratios of stated and measured dose values and the largest difference observed between the ratios (delta) are given. Also the results of the intercomparison at the reference point in water, restricted to the 36 photon beams used to irradiate the anthropomorphic phantom, are given.
43 bution at these points is 5.2%. (Fig. 8). The mean mass density of the bony structures in the two planes were determined with CT slices. The mean mass density in the central plane was 0.9 g/cm 3 and in the off-axis plane 1.0 g/cm 3. These prescribed densities were stated in the contours. CT information was not applied in any dose calculation by one of the institutes. Deviations in the density of the bone structures from the prescribed mean mass density will have a systematic effect on the results. This refers especially to point 3 where for the two lateral beams more than half of the material in front of point 3 consists of "bone material". As can be seen from Table V, the dose measured in point 3 is 2.5% higher than stated, which might indicate the presence of inhomogeneities with an average mass density lower than 0.9 g 9 cm 3 in the bone material in front of point 3. In addition, in one institute the dose in the marked points, situated in the central plane (isocentre and points 2,3,4), was recalculated using CT data. Normalised to the dose in the isocentre, the dose calculated in points 3 was about 3% higher in comparison with the dose calculated using the average mass density of 0.9 g/cm 3 of the bone material. This increase of dose in point 3, due to the low density bone material corresponds approximately to the dose deviation observed in point 3. Points 2,4,7 and 8 are positioned on the outline of the target volume. Examining the isodose plots for the prescribed and the institute techniques shows that point 2 and 7 are situated well inside the treatment volume while points 4 and 8 are situated on the outline of the treatment volume near large dose gradients. Small uncertainties in the set up of the phantom and treatment unit will therefore have more effect on the measured dose at points 4 and 8 than at points 2 and 7. It can be seen from Table V that the standard deviations in the ratios of the stated and measured dose for the points 4 and 8 are indeed larger than for points 2 and 7. The 3.3% deviation between the stated and measured dose in point 8, the latter being lower, indicates that the real position of point 8 in the phantom might be a little bit further from the isocentre. The dose at point 6 is only slightly influenced by
uncertainties in the density of the bone structures. The mean ratio of measured and stated dose values is 1.000 (Table V), while the standard deviation is 4.7%. This indicates that the dose determination in the bladder point can be performed with about the same accuracy as in points in the target volume. The average difference between the stated and measured dose in points 1 and 5 in the rectum is about 4%, the latter being higher. These data have, however, a larger standard deviation and delta value (see Tables V and VI), due to the relatively steep dose gradient at these points. The results of this part of the dose intercomparison can be compared only to a certain extent with those of others as presented in Table II, because the non-reference conditions differ for all these intercomparisons. Only with the Alderson phantom measurements of Johansson et al. [8] is a more direct comparison possible. For the combined results of two points inside the target volume, the isocentre and point 9, the distribution of the ratio of stated and measured dose values, reflected in the standard deviation (3.1%) and the maximum difference between two ratios (18%) is almost similar to the distribution observed by Johansson et al. [8] for a tonsil tumour treatment (see Table II). Conclusions Deviations between the prescribed dose and the dose delivered to a patient can arise during all steps in the dosimetry chain. Dose intercomparisons as reported here might reveal some of these deviations and allow quantifications of uncertainties in some of the steps in the dosimetry chain. Differences between stated and measured dose under reference conditions in a water phantom were on the average 0.5% with a standard deviation of 1.9%, but up to 6% at maximum. If all institutes apply the same set of recommended values for the physical quantities, this maximum difference will reduce to about 2%, thus showing the importance of using a common dosimetry protocol. It indicates in addition that no major systematic uncertainties are introduced in the dose determination, as discussed by others [4,11,12].
44 A comparison between stated and measured dose at the isocentre during the prostatic tumour treatment of the anthropomorphic phantom shows that the deviation is about the same as during the water phantom irradiation under reference circumstances. This very good agreement indicates that no additional systematic uncertainties are introduced in the dose calculation at the institutes, e.g. in the applied wedge factor. It must, however, be pointed out that in this intercomparison institutes were asked to give a specified dose to the isocentre and not to the tumour or target volume. The latter method introduces additional dose differences due to different methods of dose specification applied by the institutes. The difference between stated and measured dose in the points situated in the target volume is on the average 0.4% with a standard deviation of 5.2%. This increase in standard deviation compared to the results for the isocentre can only partly be explained by using T L D instead of an ionization chamber. Other reasons are differences in the actual dose distribution and those available in the computer planning systems and possible limitations of dose calculation algorithms. Quantification of these deviations needs further study.
Acknowledgements The authors would like to express their gratitude to the radiotherapy centres in The Netherlands for their extensive cooperation and hospitality during this intercomparison. We also wish to thank B. G6bel for the preparation and read-out of the T L D powder. This work was financially supported by The Netherlands Cancer Foundation, K.W.F. Grant N.K.I. 84-12.
References 1 Brahme, A. Dosimetric precision requirements in radiation therapy. Acta Radiol. Oncol. 23:379 391, 1984. 2 Dutreix, A. The French dosimetry protocol. Radiother. Oncol. 4: 301-304, 1985. 3 1CRU report 14. Radiation dosimetry: X rays and gamma
4
5
6
7
8
9
10
I1
12
13
14
15 16
rays with maximum photon energies between 0.6 and 50 MeV. International Commission on Radiological Units and Measurements, 4201 Connecticut Avenue, N.W., Washington, D.C. 20008, U.S.A., 1969. ICRU report 24. Determination of absorbed dose in a patient irradiated by means of X or gamma rays in radiotherapy procedures. International Commission on Radiological Units and Measurements, 7910 Woodmont Avenue, Washington, D.C. 20014, U.S.A., 1976. Johansson, K. A. Studies of different methods of absorbed dose determination and a dosimetric intercomparison at the Nordic radiotherapy centres. Thesis University of Gothenburg, Sweden, 1982. Johansson, K. A., Mattsson, L. O. and Svensson, H. Dosimetric intercomparison at the Scandinavian radiation therapy centres. 1. Absorbed dose intercomparison. Acta Radiol. Ther. Phys. Biol. 21:1 10, 1982. Johansson, K. A., Horiot, J. C., Van Dam, J., Lepinoy, D., Sentenac, 1. and Sernbo, G. Quality assurance control in the EORTC cooperative group of radiotherapy. 2. Dosimetric intercomparison. Radiother. Oncol. 7: 269---279,. 1986. Johansson, K. A., Horiot, J. C. and van der Schueren, E. Quality assurance control in the EORTC cooperative group of radiotherapy. 3. lntercomparison in an anatomical phantom. Radiother. Oncol., Submined, 1986. Mijnheer, B. J., Battermann, J. J. and Wambersie, A. Which accuracy is required and can be achieved in clinical photon and neutron dosimetry?. Radiother. Oncol. Mijnheer, B. J., Aalbers, A. H. L., Visser, A. G. and Wittk~imper, F. W. Consistency and simplicity in the determination of absorbed dose to water in high-energy photon beams: a new code of practice. Radiothcr. Oncol. 7:371 384, 1986. Mijnheer, B. J., Wittk~imper, F. W., Aalbcrs, A. H. L. and van Dijk, E. Experimental verification of the air kerma to absorbed dose conversion t:actor C~.~. Radiother. Oncol. 8: 49-56, [987. NACP (Nordic Association of Clinical Physics). Procedures in external radiation therapy between 1 and 50 MeV. Acta Radiol. Oncol. 19: 55--79, 1980. Samulski, T., Dubuque, G. L., Cacak, R. K., Courlas, G., De Werd, L. A., Hilko, R., Humphries, L. J., Jones, D., Masterson, M. E., Miller, D. W,, Stovall, M. and Wochos, J.F. Radiation therapy dosimelry reviews by the centers [k~r radiological physics. Int. J. Radiat. Oncol. Biol. Phys. 7: 379 383, 1981. Shalek, R. J., Kennedy, P., Stovall, M., Cundiff, J. H., Gagnon, W. F., Grant IIl, W. and Hanson, W. F. Quality assurance for measurements in therapy. N.B.S. SP 457, 1976. Somerwil, A. and van Kleffens, H. J. Experience with the Alderson Rando Phantom. Br. J. Radiol. 50: 295-296, 1977. Svensson, H. Dosimetric measurements at the nordic medical accelerators, ll. Absorbed dose measurements. Acta Radiol. Ther. Phys. Biol. 10:631 654, 1971.