TLD postal dose intercomparison for megavoltage units in Poland

R ADIDTHERAPY aO~~~~~~~ Radiotherapy and Oncology 36 (1995) 143-152

ELSEVIER

TLD postal dose intercomparison for megavoltage units in Poland J. Iiewska*, R. Gajewski, B. Gwiazdowska, M. Kania, J. Rostkowska Medical Physics Department. Cancer Centre. ul. Wawelska IS, M-973 Warsaw, Poland

Received 28 November 1994;revision received 7 April 1995;accepted 11 July 1995

The aim of the TLD pilot study was to investigate and to reduce the uncertainties involved in the measurementsof absorbed dose and to improve the consistency in dose determination in the regional radiotherapy centres in Poland. The intercomparison was organized by the SSDL. It covered absorbed dose measurementsunder reference conditions for Co-60, high energy X-rays and electron beams. LiF powder type MT-N was used for the irradiations and read with the Harshaw TLD reader model 2OOOB/2OOOC. The TLD systemwas set up and an analysis of the factors influencing the accuracy of absorbed dose measurements with TL-detectors was performed to evaluate and minimize the measurementuncertainty. A fading not exceeding 2% in 12 weeks was found. The relative energy correction factor did not exceed3% for X-rays in the range 4- 15 MV, and 4% for electron beams between6 and 20 MeV. A total of 34 beamswas checked.Deviation of f 3.5%stated and evaluated dosewas consideredacceptable for photons and f 5% for electron beams.The results for Co-60, high energy X-rays and electron beamsshowed that there were two, three and no centres, respectively, beyond acceptancelevels. The sourcesof errors for all deviations out of this range were thoroughly investigated, discussedand corrected, however two deviations remained unexplained. The pilot study resulted in an improvement of the accuracy and consistency of dosimetry in Poland. Keywords:

Radiotherapy; Quality Assurance; TLD intercomparison

1. IntroductioIl

There are 52 megavoltage facilities operating at the present time in Poland, in 18 regional oncological centres, delivering radiation therapy to cancer patients: 28 cobalt units and 24 linear accelerators. In these centres and in many other small hospitals and medical schools there are about 70 low energy and orthovoltage X-ray therapy machines. In Poland (population approx. 40 million), the approximate number of new oncology patients, which require radiotherapy, either curative or palliative, reaches about 50 000 per year [28]. It is generally accepted [9,25] that f 5% uncertainty in dose delivery to the irradiated volume is a safe limit causing no severeconsequencesdue to the treatment. Due to the complexity of procedures involved in radiotherapy, from the beam dosimetry, patient data acquisition and treatment planning, to the irradiation of the patient, the development and application of relevant quality assur* Corresponding author.

ante (QA) and quality control (QC) programmes seems to be a key factor in reducing overall uncertainty associated with subsequent steps of the radiotherapy chain. There exist many national and international recommendations [9,11,15,19,21,25]concerning quality control of high energy machines, and it is rather obvious that carrying out well established QC procedures in every-day hospital life is an essential condition for reducing some errors in the performance of radiotherapy

machines.

The uniformity and consistency of the basic dosimetry among different centres is another fundamental factor, which must be considered in contemporary radiotherapy. Until the late 1980sin Polish regional cancer centres, various dosimetry protocols were in use: ICRU-23 [lo], NACP [20], AAPM [l] and, most recently, IAEA [8]. In 1988,the Polish Secondary Standard Dosimetry Laboratory (SSDL) at the Cancer Centre in Warsaw was approved as a member of the IAEA SSDL network and took some responsibilities for carrying out a programme, the aim of which was to investigate and to

0167-8140/95/$09.50 0 1995Elsevier Science Ireland Ltd. All rights reserved SSDI 0167-8140(95)01604-F

144

J. Izewska et al. /Radiotherapy

reduce the uncertainties involved in the measurementof absorbed dose and to improve dose determination consistency in the regional radiotherapy centres. Starting from 1991,most of the Polish centres switched to a national code of practice of absorbed dose determination [4] which is based on TRS-277 IAEA protocol (81,and is extended to determination of absorbed dose to water basedon plastic phantom measurements,as well as giving coefficients for ionization chambers manufactured in Poland. A few countries [3,7,12,13,16,18,22,24,26,27] and, at the international level, a few organisations (IAEA [22,23], EORTC [6] and, recently, the EC network [2]) have established dosimetry intercomparison for radiotherapy centres. In Poland, the TLD postal dose intercomparison pilot study was organized by the SSDL in 1990-1993 and it covered intercomparison of beam outputs of Co-60 units, high energy X-rays and electron beams. Since the SSDL’s main interest was to study the accuracy and consistency of the basic dosimetry, in every subsequent year one cobalt beam, one high energy X-ray or electron beam per centre was checked. All centres participated in the TLD intercomparison voluntarily and showed an interest in continuation of the project.

and Oncology 36 (1995) 143- I52

dice indrcator

\

TLD capsules

? z:

0 Ln

1 TLD holder /

2. Material and methods 2.1. TLD system

Lithium fluoride thermoluminescent virgin powder type MT-N (LiF: Mg, Ti - natural abundance, doped with magnesiumand titanium) of Polish production was used for the intercomparison. Powder was encapsulated in waterproof perspex containers (4 mm inner diameter, 20 mm inner length and 0.5 mm wall thickness) in portions of about 280 mg, which were large enough to obtain about 15-18 independent readings. Powder from the samemanufacturing batch was always selectedand sifted before encapsulation to obtain reproducible mass of the aliquots from the reader dispenser. Irradiated powder was dispensedonto the platinum planchette by the dispenser attached to the Harshaw TLD reader model 2OOOC,which coupled with picoammeter 2000 B was used for all TLD readings. During all readouts the glow curves were recorded in order to eliminate possible errors due to the temperature shift of the reader. Capsules were irradiated in a water phantom at the referencedepth using a holder stand, as shown in Fig. 1. In photon beams,capsules were inserted into one of two lower holes at a depth of 5 or 10 cm (10 x 10 cm, SSD set up) depending on beam quality. The water level was adjusted precisely to the top of the holder and the axis of the beam aligned with the holder axis. For electron irradiations the capsules were inserted into the upper hole, positioned 1 cm from the top of the holder. The water level was adjusted with the help of the dis-

r

7

Fig. I. Schematicdrawing of the TLD holder stand designedfor Polish TLD intercomparisons. The holder should be put into the water tank for the irradiations. The TLD capsulesshould be inserted into one of three.openings: for photon beams at a depth 5 or 10 cm (depending on the beam quality), the water level is aligned with the top of the holder, the distance indicator is not used; for electrons TLDs should be put into the top opening and the water level adjusted with the use of the distance indicator to get TLDs at the d-, the distance indicator is removed during the irradiations.

tance indicator to have TLD at the depth of the maximum dosefor the particular electron beam. The distance indicator was removed from the holder before TLD irradiations. The holder was not designedfor X-ray quality index checks, since Dz,,lDlo is rather insensitive to small dose distribution changes and it allows to detect only large deviations of Dz,-JDlo. For this reason the quality index was not chosen for the intercomparisons. The determination of absorbed dose to water by the SSDL was basedon the TRS-277 IAEA protocol [8] and dosewas measuredusing the secondary standard ionization chamber type 2561 certified by the Primary Standard Dosimetry Laboratory in Warsaw.

J. Izewska et al. /Radiotherapy

The absorbed dose to water, D (Gy), at the location of TLD was calculated from the TL signal, R, registered by the TLD reader using the following formula D = R X &I

X &in X &ad X &n

(1)

where: R is the TLD reading normalized to the mass of aliquot of the powder and corrected for the reader’s daily fluctuations, &,t (Gy/pC) is the calibration factor of the TLD system; J&i is determined for 2 Gy from Co-60 beam, Ktin is the dose response linearity correction factor, Kfadis the fading correction, and Z&.,is the X-ray energy response correction. The calibration factor J&i was calculated as the ratio of dose to TL-signal for the dose to water equal to 2 Gy at the centre of the TLD capsule. The irradiations were done with Co-60, Theratron 780C. The capsuleswere inserted into the holder and placed in water phantom at a depth of 5 cm, 10 x 10 cm, SSD = 80 cm. Before TLD irradiations Co-60 beam output was checked with the ionization chamber type 2571 used as a local standard. The TLD calibration factor was determined directly before mailing the samples to the participating centres and was not corrected for the daily reader fluctuations. Only the readings of the sampleswere corrected for the reader’s sensitivity changes,which were examined a few times during each reading session with the sample irradiated to the standard dose of 2 Gy. The stability of the sensitivity of the reader depends on its electronics, optics and the reflectivity of the planchette. Apart from standard samplesreadings, the constancy checks of the reader stability were also done between the subsequent readings using the internal reference light source. The dose response linearity factor K,in corrects the TL-response as a function of dose delivered to the powder. To determine the linearity factor, TLD samples were irradiated with Co-60 beam at standard conditions with different doses in the range of 1.5-2.5 Gy. The TL-readings should also be corrected for fading with correction factor Kfad, if the irradiation of the sampleswas done at a different time than the TLDs used for the determination of the system calibration factor. The long-term fading for MT-N powder was determined from the readings taken over 12 weeks, at l-week intervals. A dose of 2 Gy was delivered to each of 13 TLD capsules the same day. The first reading was done one week after irradiation. For the energy response study, the capsules were irradiated with 2 Gy in a water phantom under the reference conditions, with X-rays in the energy range 4-15 MV, and electron beams in the range 6-20 MeV. For each energy beam, three capsules were irradiated. As a reference, one capsule was irradiated with a Co-60 beam. Samplestaken from the sameLiF batch were considered to have consistent calibration factor, dose response linearity, fading and energy characteristics.

and Oncology 36 (1995) 143-152

145

Other factors, which might influence the accuracy of determination of absorbed dose from TLD readings have also beeninvestigated. One of them is the statistical distribution of TLD readings. Twenty TLD capsules providing 300 readings were irradiated to the samedose of 2 Gy from Co-60 and read in the same conditions within the same interval between irradiation and readout. The combined uncertainty in dose calculation from TLD measurementscomprises the uncertainty of dose determination by ionization chamber measurementsand the TLD system itself. The factors &I, Klin, Kfadand K,, used in Eq. (1) for the calculation of absorbed dose from the TLD reading R are all determined experimentally, or evaluated from experimentally determined functions, and all these factors have corresponding random errors. Since absorbed dose to water based on ion chamber measurements was determined using IAEA TRS 277 formalism [8], the contribution to the combined uncertainty originating from the IAEA Code of Practice interaction coefficients, as well as the calibration of secondarystandard, were disregarded due to the systematic nature. The SSDL applied IAEA TRS 277 [8] values for the uncertainties of absorbed dose at the reference point, which is, for Co-60 0.5%, and for high energy Xrays and electron beams 1.0%. The uncertainty of the fading correction was estimated from the standard error of the least square regression coefficient - slope of the fading function [ 171.The uncertainty of the energy correction factor is the combined uncertainty of the ion chamber dose determination for Co-60, high energy Xray or electron beams and TLD reading uncertainty. The total uncertainty of the TLD system is defined as square root of the sum of variances of individual coefficients, which is valid under an assumption that the uncertainties of the different coefficients are independent and follow a Gaussian distribution.

2.2. TLD intercomparisons In the three subsequent years, three separate intercomparisons were performed: (A) Co-60 intercomparison in 1991;(B) high energy X-rays in 1992;and (C) high energy electrons in 1993. Each participant of the study was identified by a code number to keep all results confidential. For each intercomparison radiotherapy centres were provided with: l l

l

a copy of national dosimetry recommendations [4]; an information sheet describing the method of irradiation of TL-detectors; a data sheet to enter the specifications of the therapy machine, measuring instrument, and also the method used for the absorbed dose to water determination, and the details concerning TLD capsules’ irradiation;

146 l

l

l

J. Izewska er al. /Radiotherapy

and Oncology 36 (1995) 143-152

the perspex holder stand, shown in Fig. 1, in which TLD capsuleswere placed for irradiation; three waterproof perspex capsulesfilled with lithium fluoride powder for irradiations; one perspex capsule (4 mm inner diameter, 15 mm inner length and 0.5 mm wall thickness) tilled with the same LiF powder serving as a background record.

The participants were requested to irradiate the capsules in sequenceto the absorbed dose of 2 Gy and to check the beam output with their dosimetry system before the irradiation of TLDs. The intercomparison was scheduledso that the irradiations in all participating centres were done at nearly the same time (within 1 week). At this time, the SSDL irradiated one reference sample per centre with 2 Gy from Co-60. For eachcapsulethe mean reading value and the standard error of the mean (SD) were determined. The average reading value of three capsules(evaluated absorbed dose) for each participant was calculated. The deviations A of reported and measured absorbed dose for each participant were calculated according to the formula recommended by the IAEA [8]:

200.0 3 0 5

199.0

Fl98.0

-

-

-w

f

(2)

_

1970 I--

A= Dp -- ljSsm , lw, 0 DSSDL

----_-

-E

‘950

1,,,,,,.,,,.,..,.,,,,,,,,,,,,.,,,,,,,,.,, 0 1 2

~/+/----

3

~i-~our]~

6

7

8

Time

where: &sDL is average absorbed dose determined by the SSDL; and D, is dose reported by the participant. The quotient of the mean reading of each capsule to the average reading from three capsules (measured dose/average dose) for each participant was also determined.

Fig. 2a. The histogram of 100measuCements of the internal light signal of the Harshaw reader. The readings are normally distributed (x2 = 7.59, Q = 0.5) with the standard deviation of u = 0.2%. Fig. 2b. The example of daily fluctuation of the Harshaw reader registered from the internal reference light source. Each point of the plot representsthe average of 10readings. An insignificant increaseof the readings in time does not exceed two standard deviations.

3. Results 3.1. TLD system

The example of fluctuations of the Harshaw reader registered from the internal reference light source is shown in Fig. 2. In Fig. 2a the histogram of 100 measurementsof the internal light signal is shown. The readings are normally distributed (x2 = 7.59, OL= 0.5) with a standard deviation of (I = 0.2%. In Fig. 2b the daily fluctations measuredwithin 8 h are plotted. Each point represents the average of 10 readings. A small increaseof the reading values in time was noted, but the maximum changewas lessthan 0.4%which was not considered to be significant. The histogram representing statistical distribution of 300 independent TLD readings (20 capsules irradiated to 2 Gy with Co-60 beam) is shown in Fig. 3. The TL-response follows again a Gaussian distribution (x2 = 12.5, a! = 0.25) with a mean value of 1.92 PC (K&t = 1.0417Gy/&) and standard deviation of a sin-

gle reading, (I = 2.4%. Standard deviation of the mean (SD) for a single TLD capsule does not exceed0.6% for Co-60 with its average value of 0.5%. For X-ray beams the deviation remains almost the same,but for electrons, due to non-uniform dose distribution within the capsule originated from the scattering from the walls of the perspexholder, the maximum standard deviation of the mean is 1.1% and its average value equals 0.7% (data from 36 capsules). The linearity correction for the dose within 2 f 0.5 Gy is &in = 1 (i 0.005) The fading of the TL-signal of the MT-N powder, shown in Fig. 4, did not exceed2% per 12 weeks. SSDL corrected the readings for the fading individually for each mailing, by irradiating referencecapsulesin the sameweek as the participants. The resulting uncertainty of the fading correction was found to be less than 0.5%. Table 1 shows the relative energycorrection factors for LiF powder in perspexcapsulesirradiated with X-rays and electrons in water under

J. Izewska

et al. /Radiotherapy

and Oncology

36 (1995)

141

143-152

Table 1 The relative energy correction factors for LiF powder in perspex cap sules irradiated in water under reference conditions Electrons

Photons

0

,"""'"I""""'I"'"""I""'"" 1.88

1.78

1.83

1.98

1.93

2.03

Reading [PC]

Fig. 3. The histogram of 300 independent TLD readings taken from 20 capsulesirradiated to 2 Gy with a Co-60 beam. The TL-response follows the Gaussian distribution (x2 = 12.5, Q = 0.25) with a mean value of 1.92 CC (Qt = 1.0417Gy/&) and standard deviation of a single reading u = 2.4%.

the referenceconditions. Assuming that the uncertainty of ion chamber dose determination for Co-60 is 0.5% and for X-rays 1% (IAEA [8], Table XXIII), and uncertainty of TL-reading is 0.5% for photons, the combined uncertainty of the energy correction for X-rays is 1.3% SD. For electrons, due to larger uncertainty in dose

Nominal energy WV)

TPRZO,s Correction Energy factor

1.25 4.0 9.0 15.0

0.64 0.72 0.76

1

1.01 1.02 1.03

6sEs9MeV IO s E s I8MeV 20 MeV

Correction factor 1.04 1.03 1.02

reproducibility of 2.4%, which comprises dose determination as well as geometry and TL-reading uncertainty, the energy correction is evaluated with an uncertainty of 2.5%. The total uncertainty of the TLD system was estimatedto be of about 1.O%for Co-60 and about 1.7% for high energy X-rays and about 2.7% for electrons. Table 2 summarisesthe contribution of the individual uncertainties to the total uncertainty of the Warsaw SSDL TLD system.Basedon theseuncertainties, and to find a compromise for photon beams, an acceptance level of 3.5% has been chosen for both Co-60 and X-ray intercomparisons (22SD) and 5% (approx. 2SD) for electrons. All deviations detected during the intercomparison being beyond the acceptancelevel were carefully investigated and the reasons for errors examined and discussedwith the participants.

Table 2 Combined uncertainties of the individual coefficients to the absorbed dose to water determined from TLD readings Coefficient

Uncertainty (%)

Ion chamber dose determination: co-60 0.5 X-rays 1.0 Electrons 1.0

10

4 0

1

2

3

4

5

0

7

0

9

10

11

12

Time intfzrval [weeks]

Fig. 4. The fading of TL-signal of the MT-N powder. TLD capsules were irradiated the same day and read in subsequent 12 weeks. The readings were corrected for the daily fluctuations of the reader sensitivity, checked with freshly irradiated reference samples. The fading does not exceed 2% per 12 weeks.

Capsule reading Calibration coefftcient Linearity correction Fading correction

co.7 0.7 0.5 0.5

Energy correction: X-rays Electrons

1.3 2.5

Total Co-60 Total X-rays Total electrons

1.0 1.7 2.7

J. Izewska et al. /Radiotherapy and Oncology 36 (1995) 143-152

148

Table 3 Data concerning detertnination of absorbed dose to water for mailed TLD intercomparison for Co-60 beams Beamcude number Model of cobalt unit Year of installation/source replacement Type of electrometer Type of ion chamber Measured/average dose, TLD Deviation A (%)

9001 Philips 82

9002 Theratron 780C 86

9003 9004 9005 Gamma- Alcyon II Gammatron 3 tron s 78f90 83 7of90

Ionex 2500/3 NE 2505/3A 0.999 1.002 0.999 0.9

Ionex Ionex 250013 DM 2590 NE 2581 NE 2505l3A 1.006 1.008 0.996 0.997 0.998 0.996 1.0 -1.5

9007 Theratron 780 86

9012 Gammatron S 78184

9013 Gammatron S 78190

Farmer Ionex Ionex Farmer Farmer Fartner PTW 2570/1B DL4/DI4 250013 2570A 2570A 2570/lB 2500/3 NE 2571 NE 2571 NE 2571 NE 2581 NE 2571 NE2581 PTW 23331 0.994 1.000 1.000 1.001 1.002 1.004 1.007 1.004 1.001 1.000 1.002 0.997 0.999 0.995 1.002 0.999 1.000 0.998 1.001 0.997 0.999 -5.0 -0.5 0.4 -0.7 0.4 2.5 12.1

Ionex 2570/l NE 2571

3.2. TLD intercomparisons

Out of 12 Polish regional centres performing Co-60 and -13 accelerator radiotherapy, the cooperation of most centres was obtained (11 centres for Co-60, 11 for X-rays and 12 for electron beams). Tables 3-5 present selecteddata gathered from the centres concerning their radiotherapy units, dosimetry systems and the determination of absorbed dose to water by the ionometric method and TLD for Co-60, high energy X-rays and electron beam intercomparison, respectively. Ionization chambers (0.6 cc) manufactured by the Nuclear Enterprises type 2571 and 2581, connected to Ionex 2500/3or Farmer 2570dosimeters,are in common usein Poland. A review of the information comprised in the data sheets,confirmed that most of the radiotherapy centres use dosimeters calibrated in the preceding 3 years; two centres used non-valid SSDL calibration certificates. Fortunately, the deviations of the beams in

9008 Theratron 780 85

9009 Gammatron S 80187

9011 Theratron 780 88

0.996 1.006 0.998 1.5

these two centres were found to be within acceptance limits. Most participants determined absorbed dose to water in the referencepoint using a water phantom, one participant uses a perspex phantom for beam output calibrations for photons and a polystyrene phantom for electrons. A majority of the participants, with the exception of one centre, have declared their intention to follow the TRS-277 IAEA [S] recommendations, however a few mistakes and inconsistenciesrevealed by the data sheets prove that, sometimes, an understanding of the employed protocol is not full or someminor mistakes in interpretation are made. All these mistakes and misinterpretations were carefully discussedwith the participants, and explained and corrected. Table 6 summarizes the results of our study in the context of other recent TLD intercomparisons performed by different national and international institutions [2,5-7,18,22,23]. To compare the results of this work

Table 4 Data concerning determination of absorbed dose to water for mailed TLD intercomparison for photon beams Beam code number Model of accelerator Nominal X-ray beam energy WV) TPRZO/l0 JlON20 Type of electrometer Type of ion chamber Measured/average dose, TLD Deviation A [“‘I

9201 Satume II+ 10

9202 Satume 43 15

0.74 Fanner 250313 NE 250513A I .OOO 1.005 0.995 0.8

1.52 Farmer 257Of1B NE 2571 1.008 0.982 1.011 -2.6

9203 Neptun lop 9

9204 Neptun lop 9

0.73 1.61 1.607 Farmer Ionex 25023 250013 NE NE 2571 2505/3A 1.004 0.994 1.008 1.002 0.988 1.004 -7.2 0.9

9205 Neptun lop 9

9207 Neptun lop 9

0.709 1.62 Ionex PTW 250013 D14-7751 NE 2571 PTW 23331 0.998 0.997 0.991 1.012 1.012 0.992 -6.4 -0.6

9208 Neptun lop 9

9209 Neptun lop 9

9210 Neptun lop 9

9211 Neptun lop 9

0.73

0.716

Ionex Fartner 2500/3 2570llB NE 2571 NE2581

0.731 1.602 Ionex 250013 NE2581

0.70 1.64 1.626 Ionex Farmer 250013 257011 NE 2571 NE 2581

1.022 0.990 0.988 4.1

1.005 0.994 1.002 -I

0.993 1.001 1.007 -0.6

0.993 1.026 0.981 -3.5

9212 Neptun lop 9

0.986 1.013 I.@30 -3.2

Deviation A (%)

Type of ion chamber Measured/average dose, TLD

Beam code number Model of accelerator Nominal electron beam energy (MeV) Average energy at the phantom surface EO WV) Depth of dose maximum (a) Type of electrometer

9302 Satume 43 10.5

9.79

2.4

Farmer 2570/l NE 2534 0.96 1.02 1.02 -2.5

9301 Satume II+ 9

8.6

2.2

Ionex Dosemaster NE 2571 0.99 1.03 0.98 I.0

2.4

2.35

2.3

1.7

6.85 2.2

8.70

Ionex 2500/3 Ionex 250013Ionex 250013lonex 250013Farmer 257011 NE 2505J3A NE 2571 NE 2571 NE 2581 NE 2581 1.03 1.02 1.04 0.93 1.04 0.97 1.02 1.00 1.04 0.96 I.00 0.98 0.97 1.03 -0.5 -4.9 -2.1 -2.9 2.0

9.20

8.90

8.85

Farmer 2570llB NE 2581 1.00 0.98 1.02 -3.8

2.2

9.00

2.2

9.3

1.8

7.27

Ionex 250013Ionex 250013Farmer 257011 NE 2581 NE 2571 NE 2571 0.99 0.99 1.05 1.00 1.03 0.99 1.01 0.98 0.97 1.8 2.0 0.0

3.0

13.3

NE 2581 1.02 1.02 0.97 4.7

Ionex 250013

1.7

7.25

9303 9304 9305 9306 9307 9309 9310 9311 9312 9313 Neptun IOp Neptun 1Op Neptun IOp Neptun IOp Neptun 1Op Neptun 1Op Satume II+ Neptun IOp Neptun 1Op Neptun 1Op 10 10 10 8 10 10 15 10 8 8

Table 5 Data concerning determination of absorbed dose to water for mailed TLD intercomparison for electron beams

150

J. Izewska et al. /Radiotherapy

Table 6 Results of a few recent TLD photon and electron beam intercomparisons in referenceconditions Reference

Study

Dutreix at al.

EC 91-93 photons

PI Hanson at al.

151 Hansson at al.

I61

Huntley at al. (71 Kiyak at al.

1181 Penchev at al.

WI Svenssonat al.

1231 This work

RPC 84-92 photons electrons EORTC 87-92 photons Australia 92 photons Turkey 89-91 CO-60 Bulgaria 75-92 co-60 IAEA 69-87 co-60 Poland 91-93 photons electrons

Number Mean of beams

Standard deviation

125 1198

0.970 0.985a

0.095 0.025a

8895 5215

1.000 0.993

0.024 0.032

357

1.007

I.040

30

0.993

0.033

45

1.025

0.048

173

0.986

0.033

1945

0.998

0.067

22 12

1.004 1.004

0.038 0.027

BExcluding deviations > 12%. The mean values and the standard deviations of the distributions of measured-to-stated doses are reported.

with other studies, we recalculated mean and standard deviation of the distribution of ratios of measured-tostated doses. 4. Diion 4.1. Results of the intercomparisons A. TLD intercomparison for Co-60 beams

As can be seenfrom Table 3, the centres did their irradiations with Co-60 beams produced by Theratrons (four beams), Gammatrons (five beams), Alcyon (one beam) and Philips (one beam). The reported percentage depth dose (PDD) at 5 cm for Co-60 ranged from 77.2 to 78.9%, which is interesting, since some participants employ published isodose data for Co-60, and someuse data measured by the department. The deviations A of the quoted absorbed dose from the averageof the SSDL evaluations were within f 3.5% acceptancelevel for nine participants out of eleven. For two centres, differences in stated to evaluated dose becamemore serious (-5% and 12.1%).Thesecaseswere thoroughly discussed with the participants, and the sourcesof errors traced during on-site visits. During the first visit it was explained that -5% deviation was caused by a few small errors in geometry and calculation, which is additive. The reason for the 12.1%discrepancy was due to calculation mistake: wrong chamber factor,

and Oncology 36 (1995) 143-152

exposure factor N, instead of absorbed dose factor No was used. In both casesthe mistakes were made during intercomparison and the outputs used for patient irradiations were not influenced. In the latter case, however, the physicist did not compare the output used for patient treatments with that reported in the TLD data sheet. The averagequotient of the measureddose for a single capsule to an averageof three capsuleswas found to be 0.3% for all participants, which shows very good reproducibility of TLD irradiations and measurements. B. High energy X-ray intercomparison

For high energy X-ray checks (Table 4), the participating centres typically used 9 MV beam from Neptuns 1OP(9 beams),however Satume II+ and Satume 43 beams (10 and 15 MV) were also checked. The deviations A of the quoted absorbed dose from the averageof the SSDL evaluations were within f 3.5% acceptancelevel for eight participants out of eleven (see Table 4). In one case value of the quoted dose differed from SSDL value by 4.1%. In two cases,large errors were detected (-6.4 and -7.2%): both were caused, according to the participants, by geometry errors during TLD irradiation; 4.1% deviation remained unexplained. All three beams were immediately remeasured by the participants, and the outputs used for patients agreed with recheckedand reported values, which suggeststhat patients were probably not at risk. The reason for the bad TLD results may be thus related to insufficient care taken during TLD irradiations. Review of the data sheetsrevealed a few inconsistencies in the values of the displacement of the effective point of measurement,in the applied perturbation correction factor, and in the stopping power ratio water-toair - which were incorrectly read from Fig. 14 and Table XVIII of the TRS-277 recommendations [8]. Calculation errors were also made in the dose delivered to capsules.These factors only slightly (< 1%) influenced the value of the dose calculated as compared to the dose reported by the participants. As was calculated from Table 4, the averagedeviation of the dose from the single capsule from the average of three capsules was 0.9%. C. Intercomparison for electron beams

Most of the participants in the electron project (Table 5) irradiated their capsuleswith nominal beam energy of 8-10 MeV using Neptuns 1OPaccelerator. Only one centre irradiated TLDs with the energy of 15 MeV. From the review of data sheetsit appeared that four centres made small mistakes (< 1%) and one centre made a large error (19%) in the chamber perturbation correction factor. The physicist wrongly used for chamber radius a value for the diameter, and extrapolated perturbation corrections from Table XI of

J. Izewska et al. /Radiotherapy

TRS-277 [8] and reached an extremely large correction. There were also minor mistakes in the values of the stopping power water-to-air ratio. A few inconsistencies in values of the displacement of the effective point of measurement appeared. Calculation mistakes were also made, however most of these mistakes were small (< 1%) and did not significantly influence a value of the absorbed dose delivered to the capsule. In one caseserious errors were made; but combined with a 19% error of perturbation factor they luckily cancelled out! One may expect, that in this centre, the patient treatment may be potentially affected. It can be seenfrom Table 5 that the deviations A of the quoted absorbed dose from the average of SSDL measurementswere all within iS%. It was observed that the standard error of the mean for each capsule is larger for electrons (0.4-l. 1%) than that for X-rays (0.3-0.6%). The reason may have been inhomogeneous dose distribution within the capsule due to scattered electrons from the holder walls. The average deviation of the dose from a single capsule to an average of three capsules, calculated from Table 5, was 2.4%, which indicated that the reproducibility of the TL-signal of samples irradiated with electrons had a larger spread than TLDs irradiated with photons (0.3% for Co-60,0.9% for high energy Xrays). Since it is rather unlikely that the stability of all checked linacs (mainly Neptuns 1OP)is generally worse for electron beams than for photons, the larger spread of electron beam data may be because of the larger uncertainty in the reproducibility of irradiation geometry. 4.2, Comparison with other TLD studies The requirement of l 5% accuracy in the delivery of absorbed dose to a target volume in a patient, as it is generally accepted [9,25], presupposes that the uncertainty in the dose determination in a reference point in a phantom is smaller. The EORTC in Europe and RPC in the USA [5,14] set a tolerance level of *3% for the uncertainty of the calibration of the therapy machinesas an acceptable practice in radiotherapy. Using mailed TLdosimetry with its limited precision due to uncertainties incorporated into the method (see Section 3.1), we set acceptance limits of *3.5% for photons and f 5% for electrons. Our method doesnot allow the satisfactory identification of small deviations to check if the beams meet the criterion of *3%, however the acceptance levels in this work remain consistent with other TLD networks with acceptancelimits between 3 and 5% [5-71. As can be seen from Table 6, the mean value of the ratio of measured-to-stateddosesis, in this study, 1JO4 for photons and 1.004for electrons, indicating, that no large systematicerror was detected, even if the statistics

and Oncology 36 (1995) 143-152

151

is basedonly on 22 checksfor photons and 12checksfor electron beams. The mean of measured-to-stateddoses ranges from 0.970 for the EC network (including deviations > 12%)to 1.025for Turkey intercomparisons. The standard deviation of the results in our study is SD = 0.038 for photons and SD = 0.027 for electrons, whereasfor other intercomparisons it ranges from 0.024 (RPC) to 0.095 (EC network, including deviations > 12%)for photons and 1.032(SD) for electrons (RPC). As was reported by Dutreix et al. [2], excluding deviations larger than 12%(approximately 5% of the results), the mean and standard deviation of the distribution become 0.985 and SD = 0.025, respectively. In the IAEA Co-60 intercomparison, which covers the SSDL network and many radiotherapy departments mainly in developing countries, deviations exceeding 30% (approximately 1%of results) have also beenexcluded from the analysis. The IAEA standard deviations taken only for European countries, as was reported by Svensson [23], vary from 0.019 to 0.083 for different countries. The EORTC intercomparison covers only selected radiotherapy centres, which take part in clinical trials and for which lower deviations might have been expected, whereas national intercomparisons are organized for all radiotherapy centreswilling to participate. The figures from this work compare well with the statistics of other intercomparisons, even if the number of beams checked in our study is small. In this pilot study we focusedon the protocol compliance only, omitting clinical aspectsof the dosimetry, since our goal was to test the consistency of the basic dosimetry in Poland as a first step of the larger QA programme. Conclusive verification of absorbed dose actually delivered to a patient during radiation treatments may be performed by in vivo dosimetry. 5. c0nc1usI0ns (1) The intercomparison pilot study has made it possible to test the consistency of high energy photon and electron beam dosimetry in regional radiotherapy centres in Poland. (2) The TLD intercomparison has shown that the high energy photon and electron beam dosimetry in most centres in Poland remains within acceptable limits. (3) Out of 34 checksof Co-60, high energy X-rays and electron beams(seeTables 3-5), in 29 casesthe deviation of the absorbed dose stated by the participants compared with values measured by the SSDL did not exceedthe acceptancelevel of *3.5% for photons and *5% for electrons. (4) The standard error of the mean for a single TLD capsule,which is lessthan 0.6% for X-rays and lessthan 1.1%for electrons showed that the precision of the procedure developed by the SSDL was satisfactory. (5) It follows from the experienceacquired in the pilot

152

J. Izewska et al. /Radiotherapy and Oncology 36 (1995) 143-152

study, that there is a permanent need for contact with physicists from regional centres to discussthe dosimetry problems they face in professional life, and also that the support they receiveis fully appreciated. All participants are in favour of continuation of the project. (6) The pilot study confirmed that there is a need for further clinical dosimetry training of physicists employed in oncological centres performing radiotherapy. Acknowledgements The authors wish to acknowledge Danuta Milkowska for her help with the TLD readings as well as other colleagues from the Medical Physics Department of the Cancer Centre in Warsaw, who were temporarily involved in the TLD intercomparisons. This work was partially supported by Research Contracts 6013/RB and 6013/RB/Rl awarded by the IAEA. References I11 AAPM. A protocol for the determination of absorbed dose from high energy photon and electron beams. Med. Phys. 10: 741-771, 1983. 121Dutreix, A., Derreumaux, S., Chavaudra, J. and van der Schueren,E. Quality control of radiotherapy centres in Europe: beam calibration. Radiother. Oncol. 32: 256-264, 1994. [31 Gajewski, R., Eska, J. and Gwiaxdowska, B. Absorbed dose intercomparison studiesfor Co-60 therapy units in Poland. Part 1. TLdetector responsesinvestigation. Post. Fix. Med. 27: 71-75, 1992. 141 Gajewski, R., Iiewska, J., Rostkowska, J. and Kania, M. National recommendationsfor absorbed dose determination from high energy photon and electron beams applied in teleradiotherapy. Polish Society for Medical Physics, Warsaw, 1991, in Polish. 151Hanson, W.F., Stovah, M. and Kennedy, P. Review of dose intercomparison at a reference point, Proceedings of the Interregional Seminar for Europe, the Middle East and Africa organ&d by the IAEA/ESTRO. Leuven, 16-20 Sept. 1991, IAEA-TecDoc-734, 121-130, 1994. I61 Hansson, U., Johansson, K.A., Horiot, J.C. and Bemier, J. Mailed TL dosimetry for machine output check and clinical application in the EORTC radiotherapy group. Radiother. Oncol. 29: 85-90, 1993. [71 Huntley, R.B., Bera, P. and Nette, P. IAEA/WHO TLD radiotherapy dosimetry intercomparison for Australia. Measurement Assurance in Dosimetry. Proceedings of a Symposium, Vienna, 24-27 May 1993,IAEA-SM-330/70, 177- 190, Vienna 1994. PI IAEA. Absorbed Dose Determination in Photon and Electron Beams.An International Code of Practice Tech. Rep. No. 277. IAEA Pub., Vienna, 1987. [91 ICRU. Prescribing, Recording and Reporting Photon Beam Therapy. Report No. 50, ICRU, Washington, 1993. 1101ICRU. Measurementof absorbed dose in a phantom irradiated by a single beam of X- or -y-rays. ICRU Report 23, Bethesda, 1973. [Ill IEC. Safety of medical electrical equipment, part 2: particular

requirements for medical electron accelerators in the range of l-50 MeV. International Electrotechnical Commission. IEC Publication, Geneva, 601-2-L 1990. WI Iiewska, J., Gajewski, J., Gwiaxdowska, B., Kania, M. and Rostkowska, J. Absorbed dose intercomparison studies for Co-60 therapy units in Poland. Part 2. Absorbed dose measurements.Post. Fix. Med. 27: 77-83, 1992. 1131 Johansson,K.A., Mattson, L.O. and Svensson,H. Dosimetric intercomparison at the Scandinavian radiotherapy centres.Acta Radiol. 21: I-10, 1982. [I41 Johansson, K.A., Horiot, J.C., Van Dam, J., Lepinoy, D., Sentenac, I. and Sembo, G. Quality assurancecontrol in the EORTC cooperative group of radiotherapy. Dosimetric intercomparison. Radiother. Oncol. 7: 269-279, 1986. 1151 Johansson,K.A., Hanson, W.F. and Horiot, J.C. Workshop of the EORTC Radiotherapy Group on quality assurancein cooperative trial of radiotherapy: A recommendation for EORTC Cooperative Groups. Radiother. Oncol. 11: 201-203, 1988. I161Kirby, T.H., Hanson, W.F. et al. Mailable TLD systemfor photon and electron therapy beams. Int. J. Radiat. Biol. Phys. 12: 261-265, 1985. u71 Kirby, T.H., Hansson, W.F. and Johnston, D.A. Uncertainty analysis of absorbed dose calculations from thermohuninescencedosimeters. Med. Phys. 19: 1427-1433, 1992. Kiyak, N., Yasar, S. and Balkan, H. Intercomparison programme of absorbed dose measurement for Co-60 teletherapy units in Turkey, Measumment Assurance in Dosimetry, Proceedingsof a Symposium, Vienna, 24-27 May 1993,IAEA-SM33Ot32,157-163, Vienna 1994. 091 Kutcher, G.J. et al. ComprehensiveQA for radiation Gncology: Report of AAPM Radiation Therapy Committee Task Group 40. Med. Phys. 21 (4): 581-619, 1994. I201NACP. Proceduresin external radiation therapy dosimetry with electron and photon beams with maximum energies between 1 and 50 MeV. Acta Radio]. Gncol. 19: 55-79, 1980. PU NCRP. Dosimetry of X-ray and ‘y-ray beams for radiation therapy in the energy range 10keV to 50 MeV, National Council on Radiation Protection and Measurements,Report 69, Bethesda, 1981. WI Penchev,V., Bouchakliev, Z., Constantinov, B., Poppitx, R. and Ivanova, K., Quality Assurance of therapy level measurements at the Secondary Standard Dosimetry Laboratory, Sofia, Measurement Assurance in Dosimetry, Proceedings of a Symposium, Vienna, 24-27 May 1993, IAEA-SM-330/3, 177-190, Vienna 1994. v31 Svensson,H., Hanson, G.P. and Zsdansky, K. The IAEA/WHO TL dosimetry Service for Radiotherapy Centres 1969-1987. Acta Gncol. 29: 461-467, 1990. v41 Thwaites, D.I., Williams, J.R., Aird, E.G., Klevenhagen, S.C. and Williams, P.C. A dosimetric intercomparison of megavoltage photon beams in UK radiotherapy centres. Phys. Med. Biol. 37 (2): 445-461, 1992. ~251WHO. Quality Assurance in Radiotherapy. WHO, Geneva, Switzerland, 1988. I261Wittkamper, F.W. and Mijnheer, B.J. Dose intercomparison at the radiotherapy centres in The Netherlands. 3. Characteristics of electron beams. Radiother. Oncol. 27: 156-163, 1993. 1271 Wittkamper, F.W., Mijnheer, B.J. and van Kleffens, H.J. Dose intercomparison at the radiotherapy centresin The Netherlands. 1. Photon beam under reference conditions and for prostate cancer treatment. Radiother. Gncol. 9: 33-44, 1987. I281Zatonski, W. and Tycxynski, J. Cancer in Poland in 1990.Internal Report of Cancer Centre, Warsaw, 1993.