Phys. Med. Biol. 42 (1997) 23932409.

Printed in the UK

PII: S0031-9155(97)83024-2

A dosimetric intercomparison of electron beams in UK radiotherapy centres

A Nisbet and D I Thwaites
Department of Medical Physics and Medical Engineering, University of Edinburgh, Western General Hospital, Crewe Road, Edinburgh EH4 2XU, UK Received 4 April 1997, in nal form 15 August 1997 Abstract. A dosimetry intercomparison has been carried out for all 52 radiotherapy centres in the UK which possess electron treatment facilities. The intercomparison was carried out on one treatment unit in each centre and for three energies across the range of available energies. The position of the depth of maximum dose for a standard eld size was independently determined and a subsequent beam calibration made. The factor to convert the reading on a calibrated ionization chamber to absorbed dose in an electron beam is energy dependent, and hence to carry out an independent calibration measurement also requires the beam energy to be determined. In addition a quantitative measure of the difference in the calibration chains between the intercomparison equipment and the host departments eld instrument was carried out. In order to provide a follow-up to the initial IPSM national photon intercomparison, a photon beam calibration was measured in one photon beam in each centre. For 156 electron beam measurements, a mean ratio of intercomparison measured dose to locally measured dose of 0.994 was obtained with a standard deviation of 1.8%. For the 52 photon beam measurements, a mean ratio of intercomparison measured dose to locally measured dose of 1.003 was obtained with a standard deviation of 1.0%.

1. Introduction The demands on precision in radiotherapy dosimetry and treatment delivery are determined by the steepness of the relevant clinical doseeffect curves, both for tumour control and for normal tissue complications. Consideration of the clinical data leads to generally agreed recommendations on the required accuracy in clinical dosimetry for radical curative radiotherapy being given in ICRU Report 24 (ICRU 1976), which pointed to a need for at least 5% accuracy in the delivery of absorbed dose to the target volume in the patient. More recently, Brahme (1984) proposed a tolerance value on accuracy in dose delivery of 3% at the one standard deviation level, based on considering the consequences of dose variability on loss of tumour control probability. Similarly Mijnheer et al (1987) proposed 3.5% (one standard deviation) by considering limiting uncertainties for unacceptable increases in normal tissue complication risks. It is important to recognize that these gures are to be applied to the overall delivered dose, i.e. the dose received by the patient at the end point of all steps in the radiotherapy process. This includes patient data acquisition, treatment planning and the delivery of the prescribed, planned and accepted treatment to the patient under day to day conditions over the course of the treatment. The accuracy requirements
Present address: Department of Medical Physics, Sandringham Building, Leicester Royal Inrmary NHS Trust, Leicester LE1 5WW, UK. 0031-9155/97/122393+17$19.50 c 1997 IOP Publishing Ltd

2393

2394

A Nisbet and D I Thwaites

on each contributing stage must be correspondingly better in order to achieve these overall requirements. Such considerations make it necessary to establish comprehensive quality assurance systems at the local level in each centre (Bleehan 1991, AAPM 1994, Thwaites et al 1995). In particular, the basis of consistent dosimetry is the use of carefully designed and applied national dosimetry protocols and codes of practice (e.g. HPA 1985, IPSM 1990, 1992, IPEMB 1996a, b), ensuring traceability of dosimetry to national and international standards. However, even with protocols in place errors may occur, for example due to inexact implementation or misinterpretation of recommendations, equipment problems or mistakes. Dosimetry intercomparisons are designed to establish the accuracy and precision of dosimetry at a given level in the dosimetry chain and to assess consistency between centres. They also act as an audit that can reveal the presence of errors. Coupled with procedural audit of the local centres quality assurance system and records, they provide a powerful overall audit of local clinical dosimetry. Dosimetry intercomparison techniques form the basis of the practical methodology employed in the developing radiotherapy dosimetry audit networks (Svensson et al 1994, Hanson et al 1994, Derreumaux et al 1995, Thwaites 1996). Whilst audit systems can employ mailed TLD or site visits using ionization chambers, the latter approach provides the most accurate and exible means to carry out a dosimetry intercomparison. Dosimetry intercomparison methods and results have been reviewed recently by Thwaites and Williams (1994) and Thwaites (1994). Briey they can be divided into three categories. As the rst level in the dosimetry chain, primary standard laboratories regularly carry out intercomparisons of their provision of standard dosimetric quantities and agreement has consistently been within a few tenths of a per cent (Niatel et al 1975, Boutillon et al 1994). Secondly there have been a number of intercomparisons carried out to assess the consistency of treatment beam calibration between radiotherapy centres, making measurements under xed conditions which either duplicate or approximate reference calibration conditions (Johansson et al 1982, 1986, Wittkamper et al 1987, Thwaites et al 1992, Hoornaert et al 1993, Wittkamper and Mijnheer 1993, Hanson et al 1994). The third group essentially assess uncertainties at other levels of the clinical dosimetry chain, and range from additional measurements in non-reference conditions to more complex intercomparisons in anatomical phantoms (Worsnop 1968, Johansson 1987, Johansson et al 1986, Wittkamper et al 1987, Thwaites et al 1992). Tables 1 and 2 summarize some results for photon and electron intercomparisons at the beam calibration level. A number of general statements can be made as a result of reviewing intercomparisons (Thwaites et al 1992, Thwaites and Williams 1994): nearly all have shown some clinically signicant variations; the standard deviations on the observed distributions increase on going from cobalt-60 to megavoltage x-rays to electrons, and on going from reference point measurements to more complex situations involving more factors; and standard deviations and the incidence of major discrepancies decrease in repeated intercomparisons. The UK-wide dosimetry intercomparison of megavoltage photons (Thwaites et al 1992) provided quantitative information on the consistency of clinical dosimetry for megavoltage photons. It improved the quality of clinical dosimetry by highlighting some problems, which in turn led to improvements in practice. The study was a one-off exercise with the aim of measuring the achieved dosimetric precision in radiotherapy across the UK, yet at the same time it provided a baseline set of data which can be used as a standard for subsequent continuing regional audits in the UK radiotherapy dosimetry audit network to work from and refer to. Indeed that intercomparison stimulated the establishment of, and provided the basic methods, for on-going regional audits (Thwaites 1992, 1996, Bonnett et al 1994,

UK radiotherapy dosimetry intercomparison

2395

Table 1. Ratios of measured to stated doses of recent photon dosimetry intercomparisons in reference conditions. Reference Johansson et al (1982) Johansson et al (1986) Wittkamper et al (1987) Hanson et al (1984) Thwaites et al (1992) Region Scandinavia Co-60 x-rays Europe Co-60 x-rays Netherlands Co-60 x-rays Mainly USA Co-60 and x-rays UK C0-60 x-rays No 22 50 59 16 11 40 740 61 100 Mean 1.001 1.017 1.001 1.024 0.994 1.008 1.008 1.002 1.003 SD 0.014 0.023 0.019 0.033 0.006 0.02 0.019 0.014 0.015 Range 0.05 0.10 0.10 0.14 0.02 0.10 0.14 0.08 0.10

Table 2. Ratios of measured to stated doses of recent electron dosimetry intercomparisons in reference conditions. Reference Johansson et al (1982) Johansson et al (1986) Wittkamper et al (1993) Region Scandinavia > 10 MeV < 10 MeV Europe 425 MeV Netherlands 620 MeV No 59 68 148 54 Mean 0.989 0.996 1.017 1.002 SD 0.027 0.034 0.047 0.025 Range 0.15 0.18 0.24 0.11

Thwaites and Allahverdi 1995). The megavoltage photon intercomparison also pointed to the need for a similar assessment of the consistency of electron beam dosimetry in the UK, partly because around 90% of UK radiotherapy centres now have electron treatment facilities, a signicant number having obtained an electron capacity relatively recently, and partly because the potential for problems in electron dosimetry may be greater than for megavoltage photons. This has been observed in other intercomparisons which have included electron beams (Johansson et al 1982, 1986, Wittkamper and Mijnheer 1993), and some of the reasons for this are: the multiplicity of energies; the variation of energy with depth; steep dose gradients; the greater inuence of variations in chamber design and phantom materials; and the possibility of larger chamber correction factors, i.e. polarity and recombination. This paper reports a systematic electron dosimetry intercomparison in all UK radiotherapy centres possessing electron facilities. Whilst this work is a scientic study of the currently achieved consistency in megavoltage electron beam dosimetry in the UK, it is intended that as with the photon intercomparison the methodology developed for clinical electron dosimetry audit may be subsequently adopted by regional audits and that the results will provide a baseline data set. A basic photon intercomparison was also carried out as a follow-up to the earlier IPSM study (Thwaites et al 1992) in order to assess its impact and any subsequent changes.

2396 2. Methods

A Nisbet and D I Thwaites

An epoxy resin solid water substitute phantom material, WTe, produced commercially by the Radiation Physics Department at St Bartholomews Hospital, London was employed for all the measurements. WTe is a formulation developed for use in electron beams. The depth ionization curves measured in the material were shown to be in agreement with those measured in water within the limits of measurement uncertainty. Fluence ratios of water to solid water were found to be linearly dependent on the mean electron beam energy at the depth of measurement, with corrections being no more than 0.5% (Nisbet and Thwaites 1997b). The water equivalency of the material for megavoltage photon beam use has also been reported (Allahverdi et al 1997). The WTe solid water phantom consisted of 250 mm 250 mm sheets with thickness varying between 1 mm and 50 mm. The sheets were measured with a micrometer and found to be within 0.1 mm of the nominal thickness. For all the measurements 100 mm of phantom material was positioned behind the ionization chamber to provide backscatter. Maximum sheet thicknesses of 20 mm were used to set the depth of the chamber in the phantom material and no problems with charge storage were observed with this arrangement. A Nuclear Enterprises 2570 dosemeter with an NACP type-02 design parallel plate ionization chamber (marketed by Scanditronix) were employed for the electron beam measurements. An operating voltage of 250 V was employed and ion recombination was evaluated both during calibration in a cobalt-60 gamma-ray beam and also in the subsequent determination of absorbed dose in an electron beam. This ensured that any changes in sensitivity at different voltages were automatically taken into account and is an acceptable alternative approach (IPEMB 1996a) to using an operating voltage of less than 100 V (Burns 1991, 1994). A 10 mm sheet of WTe was machined to hold the NACP chamber with the front surface of its entrance window ush with the sheet surface. The chamber was calibrated against a secondary standard dosemeter (Nuclear Enterprises model 2560/2561) in a cobalt-60 beam according to the HPA (1985) code of practice. The calibration of the NACP chamber was repeated quarterly throughout the course of the study. The standard deviation of the quarterly calibrations was 0.25%. In addition, to further ensure consistency, the chamber response was checked with a strontium-90 check source (available commercially from Gothic Crellon) before and after each visit to a radiotherapy centre. The standard deviation of the strontium-90 consistency checks was 0.2%, with a maximum deviation from the standard response of 0.4%. Ion recombination and polarity effects for the ionization chamber were determined across the range of measurement conditions encountered. Ion recombination corrections were then determined for each beam in a particular centre based on the value of the operating dose per pulse of the linear accelerator for that modality and quality. The operating dose per pulse was estimated from the gun pulse repetition frequency and the doserate as indicated on the accelerator console during beam-on operation of the accelerator. Where such parameters were not available on the console, the gun pulse repetition frequency and doserate as quoted by the host department were employed. Time considerations prevented the actual measurement of recombination in each centre. It should be noted that an error of 20% in the pulse repetition frequency results in an error of no more than 0.1% in the ion recombination correction factor. This may be employed as a very conservative estimate of the uncertainty in the use of the recombination correction. The polarity corrections were determined from the mean electron beam energy at the depth of measurement. The values for the chamber and electrometer employed in the intercomparison have been reported in a technical note (Nisbet and Thwaites 1997a).

UK radiotherapy dosimetry intercomparison

2397

For the photon beam measurements the Nuclear Enterprises 2570 dosemeter was used with an NE2571 graphite-walled cylindrical ionization chamber. For the photon beam intercomparison the chamber was calibrated against the secondary standard dosemeter in a cobalt-60 gamma-ray beam and also in x-ray beams of nominal energy 4 MV, 6 MV, 9 MV and 16 MV, with quality indices (TPR200 100 ) of 0.63, 0.68, 0.72 and 0.765 respectively. Calibration was carried out according to the IPSM (1990) code of practice and is in terms of absorbed dose to water, traceable to UK (NPL) national dosimetry standards. Similar consistency checks as described above were also employed for this chamber. The standard deviation of the quarterly calibrations was 0.1%. The standard deviation of the strontium90 consistency checks was 0.2%, with a maximum deviation from the standard response of 0.4%. A dedicated barometer and thermometer were used throughout the study to make pressure and temperature corrections to the ionization chamber readings. An aneroid barometer (Negretti and Zambra type M2236) was used, whilst the thermometer was supplied by Nuclear Enterprises. The temperature and pressure were compared with those measured locally to help quantify any differences in dosimetry. The HPA (1985) code of practice with the IPSM (1992) addendum were used to determine absorbed dose to water for the electron beams, and the IPSM (1990) code of practice was employed to determine absorbed dose to water for the photon beams. The electron beam intercomparison was carried out on one unit in each radiotherapy centre visited and for three energies across the range of available energies. For each nominal energy the depth of maximum dose for a standard eld size, the beam calibration and the electron beam energy were independently determined. The Ce conversion factor necessary to convert the ionization chamber reading into a measure of absorbed dose to water (as dened and used in the HPA (1985) electron dosimetry code of practice) is energy dependent. Therefore to carry out an independent measurement of the beam calibration of an electron beam means that a beam energy determination is also required. Depth ionization curves were used to estimate the electron beam energy. They were measured with a 200 mm 200 mm eld size and a constant focus to surface distance (FSD) of around 1 m (dependent on the make of accelerator). Strictly speaking, a constant focus to chamber distance should be employed. However, given that the phantom dimensions were only 250 mm 250 mm, there may have been insufcient sidescatter if a large stand-off was used to provide sufcient space beneath the applicator end to add the required thickness of phantom material and this problem would have increased for higher-energy beams. The depth ionization curve measured was therefore converted to xed focus to chamber distance by applying an inverse square law correction, i.e. by multiplying each reading by (f + d)2 /d 2 , where f is the distance of the front surface of the phantom from the source, and d is the depth of the effective point of measurement of the chamber beneath the surface. Strictly speaking f should be the distance from the chamber to the virtual electron beam source; however, it was not practicable to determine the virtual electron beam source during the intercomparison and so the nominal focus to surface distance was used. The difference in the energy determined by using this in the inverse square law correction is insignicant. The energy relationships given in HPA (1985) were used to determine the mean incident electron beam energy and the mean electron beam energy at the depths of measurement. The depth of maximum dose for the standard eld size was determined by taking readings at a number of depths 1 mm apart, around the quoted depth of maximum dose (three readings were taken at each depth). The chamber reading multiplied by the appropriate air kerma to dose conversion factor for the mean electron beam energy at that depth gives the absorbed dose to water, and hence the depth of maximum dose could be determined.

2398

A Nisbet and D I Thwaites

The dose per monitor unit at the depth of maximum dose for the standard eld size was determined from the readings taken. The temperature in the phantom cavity was used to apply a temperature correction. The ion recombination correction was calculated from the dose per pulse, and uence and polarity corrections were calculated from the mean electron beam energy at the measurement depth. It should be noted that the host department was asked to measure the beam calibrations of the appropriate electron and photon beams immediately prior to the intercomparison. The determination of the depth of maximum dose and hence the beam calibration was the rst set of electron beam measurements to be carried out. This ensured that there was no difference in dosimetry between the intercomparison and the locally measured beam calibration due to drift in output caused by prolonged use of the accelerator. Data obtained from the questionnaire (see below) enabled estimates of the Ce values around the quoted depth of maximum dose to be calculated, and these values were initially employed in the above determination of the depth of maximum dose. Subsequent use of the derived Ce values from the intercomparison measure of beam energy ensured an independent check of the beam calibration and depth of maximum dose. Finally the ionization chamber used by the host department for electron beam dosimetry was cross-calibrated against the intercomparison NE2571 graphite-walled cylindrical ionization chamber in the photon beam normally used for the calibration of the host departments electron chamber. This provided two outcomes. Firstly it gave a quantitative measure of any differences in the calibration chains between the intercomparison and the host department. In addition it enabled a measurement of the photon beam calibration to be carried out. The quality index as quoted by the host department was used to determine the calibration factor to convert the corrected instrument reading (corrected for temperature, pressure and ion recombination) to absorbed dose to water. It should be noted that in the host departments measurements of beam calibration their own previously determined chamber calibration factor was employed. The result from the cross-calibration with the intercomparison NE2571 was only used to quantify the deviation in chamber calibration factor to the total discrepancy in dose between locally measured beam calibration and intercomparison measured beam calibration. The local centre was only given the result of this chamber cross-calibration with the subsequent report. A degree of procedural audit was incorporated at each visit. To achieve this aim a questionnaire was designed which dealt with the quality control procedures in operation and the techniques used to determine dosimetric and related parameters. This questionnaire was sent out before the visit and was then used as a basis for discussion with local staff. The results from this part of the audit are not analysed here, but it should be noted that recommendations on appropriate quality control procedures and frequency of checks are given in IPSM Report 54 (IPSM 1988). The questionnaire also helped in identifying reasons for any differences observed in dosimetry between the intercomparison and that measured locally. A full analysis of the reasons for any differences in dosimetry were given with the subsequent report. The report quantied differences in dosimetry due to the different calibration chains employed; the use of different thermometers and barometers; differences in dosimetry due to differences in energy determined; differences in dosimetry due to the difference in the depth of maximum dose determined; and differences due to recombination and polarity corrections. In addition any deviation from the correct use of the HPA (1985) code of practice and IPSM (1992) addendum were highlighted. The intercomparison visits started in May 1995 and were completed in May 1996. The intercomparison was carried out at 52 radiotherapy centres in the UK, of which 50 were NHS Trust hospitals and two were private. The results were communicated back to the

UK radiotherapy dosimetry intercomparison

2399

individual centres as soon as possible following a visit. Any discrepancies outwith a pre-set tolerance value of 5% (3% for photon beam calibrations) were followed up to conrm that the causes of the difference in dosimetry as outlined in the reports were addressed. The condentiality of the results of each centre was maintained 3. Results 3.1. Electron beam calibration Figure 1 shows the distribution of nominal electron beam energies included in the study. The nominal electron beam energies range in value from 3 MeV up to 22 MeV.

Figure 1. Distribution of nominal electron beam energies in the intercomparison.

The results, expressed as the ratio of the intercomparison measured dose to the locally measured dose, are presented in gure 2. The mean value of this ratio is 0.994 (standard deviation 1.8%) with minimum and maximum values of 0.949 and 1.046, i.e. a spread in dose of 9.7% across the 52 centres. It should be noted that a value close to unity for the mean ratio suggests that there are no signicant systematic errors in the intercomparison dosimetry. A difference in dosimetry greater than 5% was observed for only one of the results. This was due to the local centre employing a cylindrical ionization chamber in the low-energy region for which the use of cylindrical chambers is not recommended. Subsequent follow-up measurements conrmed agreement to within 1%. The results have been subdivided into three energy regions: (i) a mean electron d < 5 MeV, the energy region for which beam energy at the depth of measurement, E d < 10 MeV; and cylindrical ionization chambers are not recommended; (ii) 5 MeV E d 10 MeV. A summary of the results in the three energy regions is given in table 3. (iii) E There are no signicant differences in the precision of dosimetry between these three energy regions. The results have also been subdivided according to the type of ionization chamber employed locally for electron beam dosimetry. The results are shown in table 4. It can

2400

A Nisbet and D I Thwaites

Figure 2. Results for the electron beam calibration measurements showing the distribution of the ratio of intercomparison measured dose to locally measured dose.

Table 3. Variation of beam calibration with electron beam energy. Energy d < 5 MeV E No of Beams Mean SD (%) Max. negative deviation Max. positive deviation Spread (%) 65 0.996 1.8 0.949 1.044 9.5 5 MeV 60 0.995 1.6 0.959 1.046 8.7 d < 10 MeV E 10 MeV 31 0.995 2.0 0.951 1.026 7.5 d E

Table 4. Variation of beam calibration with ionization chamber used locally. Ionization chamber NACP No of Beams Mean SD (%) Max. negative deviation Max. positive deviation Spread (%) 93 0.991 1.6 0.951 1.046 9.5 Markus 36 1.006 1.7 0.969 1.044 7.5 NE2571 21 0.995 2.0 0.949 1.024 7.5 Others 6 0.984 1.6 0.965 1.004 3.9

be seen that there are differences in the mean ratio of intercomparison measured dose to locally measured dose for each chamber type. It should be noted that the calibration procedure for parallel plate ionization chambers described in the IPEMB code of practice for electron dosimetry for radiotherapy (IPEMB 1996a) is expected to improve the consistency in dosimetry between different types of ionization chamber.

UK radiotherapy dosimetry intercomparison

2401

Table 5. Variation of beam calibration with ionization chamber used locally and with electron beam energy. (i) NACP chamber d < 5 MeV E No of beams Mean SD (%) Max. negative deviation Max. positive deviation Spread (%) (ii) Markus chamber d < 5 MeV E No of beams Mean SD (%) Max. negative deviation Max. positive deviation Spread (%) 13 1.012 1.7 0.990 1.044 5.4 5 MeV 12 1.002 1.7 0.965 1.033 6.8 41 0.992 1.5 0.957 1.030 6.3 5 MeV 37 0.993 1.5 0.966 1.046 8.0 Energy d < 10 MeV E 10 MeV 11 1.003 1.5 0.979 1.025 4.6 d E Energy d < 10 MeV E 10 MeV 15 0.985 2.0 0.951 1.026 7.5 d E

The results according to the type of ionization chamber have been further subdivided into the three energy regions. The results are presented in table 5. It can be seen that the differences in the use of different ionization chambers are reected across the entire energy range. 3.2. Electron beam energy The results, expressed as the difference in energy between that determined during the intercomparison and that quoted locally, are presented in gure 3. With 56 beams the agreement was within 0.1 MeV; with 94 the agreement was within 0.2 MeV; 116 were within 0.3 MeV; and 142 were within 0.5 MeV. The maximum positive difference in nominal electron beam energy was 1.04 MeV and the maximum negative difference was 0.72 MeV. In the latter case a depthdose curve had been used by the local centre to determine the mean incident electron beam energy. The constant linking the mean incident energy and the depth of 50% ionization is taken to be 2.4 in the HPA (1985) code of practice; however, the constant 2.33 MeV cm1 is a more representative value of the available data for a derivation of the mean incident electron beam energy from depthdose data (ICRU 1984, Brahme and Svensson 1976). If this constant had been employed by the local centre then the difference in energy determination would be 0.33 MeV. The difference of 0.72 MeV occurred for a nominal 14 MeV electron beam, and the subsequent difference in dose is estimated to be 0.4%. For the former case the difference in energy determination of 1.04 MeV occurred for a nominal electron beam energy of 20 MeV, and the subsequent difference in dose is estimated to be 0.4%. This difference occurred due to the fact that the depth ionization curves used locally to determine the mean incident electron beam energy (measured with a constant source to surface distance) did not apply an inverse square law correction to convert to depth ionization curves effectively measured with a constant source to detector distance. It should be noted that the mean ratio of intercomparison measured energy to locally quoted energy was 1.000.

2402

A Nisbet and D I Thwaites

Figure 3. Results for the mean incident electron beam energy determination showing the deviation between intercomparison measured energy and locally quoted energy.

3.3. Depth of maximum dose The results, expressed as the difference between that determined during the intercomparison and that quoted locally, are presented in gure 4. With 97 electron beams the determined depth of maximum dose was within 1 mm; in 140 beams the difference was within 2 mm; and 150 were within 3 mm. The maximum positive difference was 10.3 mm and the maximum negative difference 9.5 mm. The latter case occurred for a nominal 20 MeV electron beam, where the depthdose has a broad peak, and the subsequent difference in dose is estimated to be 0.6%. The former case occurred for a nominal 9 MeV electron energy and the subsequent difference in dose is estimated to be 7.8%; fortuitously, other areas of disagreement were observed which acted in an opposite direction and the total difference in dose was under 5%. The mean difference was 0.3 mm, this highlights the fact that a large percentage of centres take the effective measuring position of a parallel plate chamber to be on the front face of the chamber rather than at the inside of the front face. 3.4. Air kerma calibration factor Fifty ve ionization chambers (27 NACP design parallel plate ionization chambers, 20 NE2571 graphite-walled cylindrical ionization chambers and 8 Markus chambers) were cross-calibrated against the intercomparison Farmer chamber. The intercomparison was carried out in the WTe phantom material. The mean ratios of intercomparison determined calibration factor to locally quoted calibration factor was 1.006 (standard deviation 1.2%) for the NACP chambers, 0.999 (standard deviation 1.3%) for the NE2571 ionization chambers and 1.008 (standard deviation 1.2%) for the Markus ionization chambers. Comparative calibration measurements carried out in Perspex and WTe conrm that the calibration factor determined in WTe is 1.006 greater than that determined in Perspex for the NACP and Markus chambers. This is in agreement, within measurement uncertainty, with reported values of pwall (the perturbation factor to correct chamber reading for deviations from

UK radiotherapy dosimetry intercomparison

2403

Figure 4. Results for the depth of maximum dose determination showing the deviation between intercomparison measured depth and locally quoted depth.

Figure 5. Results for the air kerma calibration showing the deviation between intercomparison determined calibration factor and locally quoted calibration factor for the NACP chamber.

perfect BraggGray behaviour due to the non-medium equivalence of the chamber wall) for these chambers in these phantom materials (Laitano et al 1993). The spread in the ratio of the intercomparison determined calibration factor to locally quoted calibration factor are presented in gures 5, 6 and 7 for the NACP chamber, the NE2571 chamber and the Markus chamber respectively.

2404

A Nisbet and D I Thwaites

Figure 6. Results for the air kerma calibration showing the deviation between intercomparison determined calibration factor and locally quoted calibration factor for the NE2571 chamber.

Figure 7. Results for the air kerma calibration showing the deviation between intercomparison determined calibration factor and locally quoted calibration factor for the Markus chamber.

3.5. Photon beam calibration Figure 8 shows the distribution of nominal photon beam qualities included in the study. These are the photon beam energies which different departments are employing to calibrate their parallel plate ionization chambers.

UK radiotherapy dosimetry intercomparison

2405

Figure 8. Distribution of nominal photon beam qualities in the intercomparison.

Figure 9. Results for the photon beam calibration measurements showing the distribution of the ratio of intercomparison measured dose to locally measured dose.

The results, expressed as the ratio of the intercomparison measured dose to the locally measured dose, are presented in gure 9. The mean value of this ratio is 1.003 (standard deviation 1.0%) with minimum and maximum values of 1.026 and 0.983, i.e. a spread in dose of 4.3% across the 52 centres.

2406

A Nisbet and D I Thwaites

Table 6. Comparison of results from 1992 photon intercomparison with 1997 photon intercomparison. Number in study Co-60 (1992) (1997) MV x-rays (1992) (1997) Number within 3% (1992) (1997) 61 16 100 36 97% 100% Mean ratio 1.002 0.998 1.003 1.005 SD (%) 1.4 1.1 1.5 0.9

The results from this study are compared with those obtained in the initial photon intercomparison of 1992 in table 6. It can be seen that in this repeated intercomparison both the standard deviations on the observed distributions and also the incidence of major deviations have decreased. 4. Discussion 4.1. Uncertainties All uncertainties are quoted as one standard deviation. The type A uncertainties (random uncertainties) have been estimated as follows. The uncertainty in dosemeter response has been estimated from the strontium check source measurements to be 0.2%. In order to ensure an accurate temperature measurement the phantom material was generally placed in the treatment room approximately two to three hours before the measurements began; however, this is still probably an insufcient time for the phantom to reach equilibrium and so the temperature in the chamber insert was continuously monitored throughout the course of the measurements. It may be argued, however, that this may not have been an accurate value for the temperature in the air of the ionization chamber, and so a relatively high uncertainty in the temperature correction of 0.1% (equivalent to 0.3 C) has been estimated. The uncertainty in the barometer reading is estimated at 0.01% (equivalent to 0.1 mbar). The uncertainty in monitor unit uctuations has been estimated from the variation in chamber readings over repeated sets of measurements; an uncertainty of 0.3% has been taken for cobalt-60 units, and 0.5% for the megavoltage x-ray and electron measurements. As regards uncertainties in the focus to surface distance an uncertainty of 0.2% (equivalent to 1 mm in 1 m) has been assumed for accelerators and 0.25% (equivalent to 1 mm in 80 cm) for cobalt units. The uncertainty in the method of estimating the recombination correction factor is taken as 0.1% This leads to an overall type A uncertainty of 0.6% for the megavoltage x-ray and electron measurements and an overall type A uncertainty of 0.5% for the cobalt units. Type B uncertainties (systematic uncertainties) may arise from the calibration of the ionization chambers and the accuracy of the thermometer and barometer. In addition the use of the WTe phantom material may increase the uncertainties. It is estimated, from the results in sections 3.1 and 3.5, that these factors contribute approximately 0.6% to the

UK radiotherapy dosimetry intercomparison

2407

uncertainty with electron beams and 0.3% uncertainty to the photon beams. The total uncertainties are therefore estimated at 0.8% for electron beams, 0.7% for megavoltage x-rays and 0.6% for cobalt-60 gamma-ray beams. 4.2. Sources of differences in dosimetry The intercomparison has been designed to identify the reasons for the differences in doses determined by the intercomparison and those measured locally, and a number of common causes have been identied. Firstly the use of ion recombination corrections (IPEMB 1996a) is not considered in a large number of centres. This effect depends upon both the dose per pulse and type of ionization chamber employed, and can be a signicant effect. Likewise polarity effects are not considered in a large number of centres. Secondly the use of epoxy resin solid water phantom materials is becoming increasingly common, and depth ionization curves and Ce values identical to water are generally assumed. It has been reported that an assumption of unity for the uence ratios of epoxy resin solid water phantoms may introduce a systematic error of the order of 1% in electron beam dosimetry (Thwaites 1985, Nisbet and Thwaites 1997b). Specic factors for these materials are included in the recent IPEMB (1996a) code of practice. Thirdly, signicant differences in chamber calibration factors were observed (see section 3.4). One possible area of disagreement arises from the use of megavoltage x-ray beams to calibrate parallel plate ionization chambers. The photon beams included in the current photon intercomparison were generally those which the local centres employed for the calibration of these chambers (see gure 5). The HPA (1985) code of practice and its IPSM (1992) Addendum recommends that the calibration be carried out in a cobalt-60 gamma-ray beam and minimal provision is given for calibration in other beams, due to lack of data on the wall perturbation correction factors for parallel plate chambers in these situations. The magnitude of this effect may be of the order of 1% to 2% (Nisbet and Thwaites 1995) and there is evidence that the correction can vary between chambers of the same nominal design due to small differences in construction (IPEMB 1996a). The recent IPEMB (1996a) electron code of practice recommends that parallel plate chambers are calibrated in higher-energy electron beams because of these problems. 5. Conclusions A total of 156 electron beams were included in the intercomparison. The mean ratio of intercomparison measured dose to locally measured dose was 0.994 with a standard deviation of 1.8%. The maximum positive deviation was 4.6% and the maximum negative deviation was 5.1%. One electron beam lay outside the intercomparison tolerance level of 5%. The reason for this discrepancy has been determined and subsequent follow-up of the dosimetry has conrmed agreement to within 1%. In addition 52 photon beams were included in the intercomparison and no beams were outside the intercomparison tolerance level of 3%. The mean ratio of intercomparison measured dose to locally measured dose was 1.003 with a standard deviation of 1.0%. The maximum positive deviation was 2.6% and the maximum negative deviation 1.7%. In conclusion the study has demonstrated generally consistent radiotherapy dosimetry for electron beam dosimetry at the level of beam calibration. The methodology described can identify problems at the selected tolerance limits, allowing them to be investigated and rectied. This has been demonstrated in the study. It has also been shown that the standard deviations and the incidence of discrepancies have decreased for megavoltage photon beams since the earlier national exercise. This is in part due to the use of the intercomparisons

2408

A Nisbet and D I Thwaites

themselves and in part due to the recent implementation of quality systems and regular quality audit via the dosimetry audit network. This study has given condence in the basis of clinical delivery of radiation dose in radiotherapy treatment and in the consistency (precision) of dosimetry between different centres. Acknowledgments The study was funded by a grant from the UK Department of Health (NHS Management Executive Audit Group). The members of the IPSM Radiotherapy Dosimetry Intercomparison Working Party, responsible for the submission of the grant application, are gratefully acknowledged (D I Thwaites (Chairman), E G Aird, T Jordan, S C Klevenhagen and A McKenzie). The cooperation of the physicists in the radiotherapy departments which participated in this dosimetry intercomparison is also gratefully acknowledged. References
Allahverdi M, Nisbet A and Thwaites D I 1997 An evaluation of epoxy resin phantom materials for photon dosimetry Phys. Med. Biol. submitted AAPM 1994 Comprehensive QA for radiation oncology: report of AAPM Task Group 40 Med. Phys. 12 581618 Bleehan N (Chairman) 1991 Quality Assurance in Radiotherapy, Report of Standing Committee on Cancer Bonnett D E, Mills J A, Aukett R J and Martin-Smith P 1994 The development of an interdepartmental quality assurance programme for external beam therapy Br. J. Radiol. 67 27582 Boutillon M, Coursey B M, Hohleld K, Owen B and Rogers D W O 1994 Comparison of primary water absorbed dose standards Measurement Assurance in Dosimetry (Vienna: IAEA) pp 95112 Brahme A 1984 Dosimetric precision requirements in radiation therapy Acta Radiol. Oncol. 23 37991 Brahme A and Svensson H 1976 Specication of electron beam quality from the central axis depth absorbed dose distribution Med. Phys. 3 95 Burns D T 1991 Measurement of the Ce factor for the NACP ionisation chamber in water relative to the farmer HPA values NPL Report RSA(EXT) 27 (Teddington: National Physical Laboratory) 1994 Potential errors in recombination corrections for electron dosimetry IPSM Radiotherapy Newsheet no 10 (July) Derreumaux S, Chavaudra J, Bridier A, Rossetti V and Dutreix A 1995 A European quality assurance network for radiotherapy: dose measurement procedure Phys. Med. Biol. 40 1191208 Hanson W F, Stovall M and Kennedy P 1994 Review of dose intercomparison at a reference point Radiation Dose in Radiotherapy from Prescription to Delivery (Vienna: IAEA) pp 12130 HPA 1983 Revised code of practice for the dosimetry of 2 to 35 MV x-ray, and of caesium-137 and cobalt-60 gamma ray beams Phys. Med. Biol. 28 1097104 1985 Code of practice for electron beam dosimetry in radiotherapy Phys. Med. Biol. 30 116994 Hoornaert M-Th, Van Dam J, Vynckier S and Bouiller A 1993 A dosimetric quality audit of photon beams by the Belgian Hospital Physicists Association Radiother. Oncol. 28 3743 ICRU 1976 Determination of absorbed dose in a patient irradiated by beams of X- or gamma rays in radiotherapy procedures ICRU Report 24 (Bethesda, MD: ICRU) 1984 Radiation dosimetry: electron beams with energies between 1 and 50 MeV ICRU Report 35 (Bethesda, MD: ICRU) IPEMB 1996a The IPEMB code of practice for electron dosimetry for radiotherapy beams of initial energy from 2 to 50 MeV based on an air kerma calibration Phys. Med. Biol. 41 2557 604 1996b The IPEMB code of practice for the determination of absorbed dose for x-rays below 300 kV generating potential (0.035 mm Al4 mm Cu HVL; 10300 kV generating potential) Phys. Med. Biol. 41 260525 IPSM 1988 Commissioning and quality assurance of linear accelerators IPSM Report 54 (York: IPSM) 1990 Code of practice for high energy photon therapy dosimetry based on the NPL absorbed dose calibration service Phys. Med. Biol. 35 135560 1992 Addendum to the code of practice for electron beam dosimetry in radiotherapy (1985): interim additional recommendations Phys. Med. Biol. 37 147783

UK radiotherapy dosimetry intercomparison

2409

Johansson K A 1987 Dosimetry audits of radiotherapy institutions in Europe Proc. 6th Annual Meeting of ESTRO (Lisbon, 1987) Johansson K A, Horiot J C, Van Dam J, Jepinoy D, Sentenac I and Sernbo G 1986 Quality assurance control in the EORTC co-operative group of radiotherapy, 2. Dosimetric intercomparison Radiother. Oncol. 7 26979 Johansson K A, Mattsson L O and Svensson H 1982 Dosimetric intercomparison at the Scandinavian radiation therapy centres Acta Radiol. Ther. Phys. Biol. 21 110 Laitano R F, Guerra A S, Pimpinella M, Nystrom H, Karlsson M and Svensson H 1993 Correction factors for calibration of plane parallel ionisation chambers at a cobalt-60 gamma ray beam Phys. Med. Biol. 38 3954 Mijnheer B, Battermann J J and Wambersie A 1987 What degree of accuracy is required and can be achieved in photon and neutron therapy? Radiother. Oncol. 8 23752 Niatel M T, Loftus T P and Oetzmann W 1975 Comparison of exposure standards for 60-Co gamma rays Metologia 11 17 Nisbet A and Thwaites D I 1995 In-phantom calibration of NACP design parallel plate ionisation chambers in photon and electron beams Radiother. Oncol. 37 (suppl 1, paper 11) (Abstracts of the 3rd ESTRO Biennial Meeting on Physics in Clinical Radiotherapy, Gardone Riviera, Italy) 1997a Polarity correction factors and ion recombination effects for ionization chambers employed in electron beam dosimetry Phys. Med. Biol. submitted 1997b An evaluation of epoxy resin phantom materials for electron dosimetry Phys. Med. Biol. submitted Svensson H, Zsdanszky K and Nette P 1994 Dissemination, transfer and intercomparison in radiotherapy dosimetry: the IAEA concept Measurement Assurance in Dosimetry (Vienna: IAEA) pp 16575 Thwaites D I 1985 Measurements of ionisation in water, polystyrene and a solid water phantom material for electron beams Phys. Med. Biol. 30 4153 1992 The role of quality audit in clinical dosimetry (Proc. Quality in Radiotherapy, York, July 1992) Scope 1 14 1994 Uncertainties at the end point of the basic dosimetry chain Measurement Assurance in Dosimetry (Vienna: IAEA) pp 23955 1996 External audit in radiotherapy dosimetry Radiation Incidents ed K Faulkner and R M Harrison (London: British Institute of Radiology) pp 218 Thwaites D I and Allahverdi M 1995 The development of interdepartmental audit methods Radiother. Oncol. 37 (suppl 1, abstracts 52) (Abstracts of the Third Biennial ESTRO Meeting on Physics in Clinical Radiotherapy, Gardone Riviera, Italy, 1995) Thwaites D I, Scalliet P, Leer J W and Overgaard J 1995 Quality assurance in radiotherapy (European Society for Therapeutic Radiology and Oncology advisory report to the Commission of the European Union for the Europe against Cancer Programme) Radiother. Oncol. 35 6173 Thwaites D I and Williams J R 1994 Radiotherapy dosimetry intercomparison Radiation Dose in Radiotherapy from Prescription to Delivery (Vienna: IAEA) pp 13142 Thwaites D I, Williams J R, Aird E G, Klevenhagen S C and Williams P C 1992 A dosimetric intercomparison of megavoltage photon beams in UK radiotherapy centres Phys. Med. Biol. 37 44561 Wittkamper F W and Mijnheer B J 1993 Dose intercomparison at the radiotherapy centres in The Netherlands. 3. Characteristics of electron beams Radiother. Oncol. 27 15663 Wittkamper F W, Mijnheer B J and van Kleffens H J 1987 Dose intercomparison at the radiotherapy centres in the Netherlands. 1. Photon beams under reference conditions and for prostatic cancer treatment Radiother. Oncol. 9 3344 Worsnop B R 1968 Phantom thermoluminescent dosimeter comparison for a co-operative radiotherapy trial Radiology 91 54553