British Journal of Radiology 75 (2002),754-762 © 2002 The British Institute of Radiology
A pragmatic approach to dosimetric audit in radiotherapy
S W Blake, MSc
1 and
M P Casebow, MSc
2
1 Medical Physics and Clinical Engineering, Royal Devon and Exeter Healthcare NHS Trust, Barrack Road, Exeter EX2 5DW and 2 Department of Medical Physics, Derriford Hospital, Derriford Road, Plymouth PL6 8DH, UK
 |
Abstract
|
|---|
In 1993 a dosimetric audit programme was initiated by physicists from a group of hospitals in the South West UK with the aim of detecting dosimetric errors greater than 5%. The scope of this programme was developed to cover megavoltage and kilovoltage photons as well as electrons. Procedures operated within each department were also audited in situations where they could influence dosimetry. No dose discrepancies greater than 5% were detected in the course of the audits, though differences did correlate with complexity in some situations. These measurements provide an ongoing assurance that dosimetry within the participating centres is well controlled and provides a firm foundation for clinical practice.
|
Introduction
|
|---|
The need for an external audit of radiotherapy dosimetry was demonstrated by two incidents in the UK in 1988 and 1991. The first involved the incorrect calibration of a cobalt unit [1] and the second a misunderstanding regarding the use of a treatment planning system [2]. A national megavoltage photon audit was then undertaken in 1991 to address the question of quality control in radiotherapy dosimetry [3]. Each centre calculated a treatment plan for a specially designed phantom, which was then exposed and the delivered dose assessed.
Ongoing dosimetry audit had been recommended to maintain and improve consistency [3], but sequential visits to each radiotherapy centre in turn had taken 4 years to complete. Because of this, eight separate regional audit groups were formed across the UK [4]. Membership of such a group is now a requirement of the National Cancer Services Standards [5]. This paper documents the approach and experience of the Southwest group, whose participating centres are listed in Table 1
.
Audits were initially designed to detect discrepancies in absorbed dose at the 5% level, using the UK code of practice for dosimetry in force at the time of audit for megavoltage photons [6]. Over a period of time the scope of the audits was extended to cover kilovoltage photons and electrons using their corresponding codes of practice [7, 8]. Procedures have evolved in time but it was felt that some measurements should form part of every audit in order to make long-term assessments of any changes. Results are considered confidential and regarded as the property of the centre being audited. Each centre has agreed to the presentation of results here but no attempt is made to identify centres with specific data.
Because of the extended geography of the region, a system of "peer-to-peer" audits was planned in which each centre was checked by the next centre on the list. This was easier to resource than a system in which a central auditor visited each hospital. In this way each centre becomes the auditor (visiting centre) and auditee (host centre) in turn. Audits were designed to take up less than 2 h of treatment machine time. Procedures were also audited using a questionnaire that covered the documentation that had a direct bearing on dosimetry. Physicists from each centre meet twice yearly to review audits and plan further developments.
An external validation of the dosimetry within the group has been provided by the National Physical Laboratory (NPL), who separately audited three centres within the group. This formed part of a series of audits in which the NPL visited several centres from each group within the UK.
An audit of local barometer performance was also carried out by sending an independent instrument around the participating centres.
|
Method
|
|---|
Before the audit, both host and visiting centres performed strontium check source measurements on their dosemeters. Audit measurements were made by the visiting physicist using an ionization chamber calibrated at the visitor's centre. Normalization measurements were also made by the host centre, using the local dosemeter. Prior to any measurements, phantoms and dosemeters were left in the treatment room to reach ambient temperature. Ionization chamber readings were corrected for temperature and pressure.
Calibration of the visiting centre's dosemeter was previously determined by intercomparison measurements with a secondary standard dosemeter in a radiation beam at the visiting centre and was therefore traceable to national standards of absorbed dose and air kerma, using current protocols. The host centre beam energy could, however, be significantly different for photon beams, so a calibration correction factor was required. Arbitrary reference energies were therefore chosen (half value layer of 4 mm Al for kilovoltage, Quality Index 0.7 for megavoltage beams) and sensitivities of those chambers whose calibrations spanned reference energies could then be given relative to the interpolated reference energy. Data on chamber sensitivities at different energies were pooled within the group and relative energy correction curves determined for kilovoltage and megavoltage photons (Figures 1 and 2). These curves fit the points to within ±1%. Different curves were needed for graphite and nylon capped chambers.
View larger version (14K):
[in this window]
[in a new window]
|
Figure 1. Energy correction curve for low kilovoltage (kV) photons as a function of half value layer (HVL) in aluminium (Al).
|
|
For electrons, energy dependent factors were determined separately for each beam by additional measurements within the audit.
Megavoltage photons
Before the audit the host centre calculated three-field treatment plans for both isocentric and fixed focus-to-skin distance (FSD), irradiations of a phantom, intending to deliver a dose of 2 Gy at a central point. The phantom was then irradiated according to the plan and direct dose measurements made within the phantom using a Farmer dosemeter. These measurements were then compared with the predictions of the planning system.
The phantom used was similar in design to that used in the national 1991 audit, being made of water-equivalent plastic (ScanplasTM) with one central and four off-axis measurement points, an optional lung insert and one sloping surface (Figure 3). A hard copy of the phantom cross section was used to digitize the plan outline. The density of the lung insert relative to water was determined as 0.19 from physical measurements of its mass and volume. As plans were based on digitized outlines, the electron densities for CT inhomogeneity corrections were not required.
The phantom was irradiated with three rectangular orthogonal fields, one of which was wedged. A homogeneous dose distribution over the central five measurement points was produced for a planned dose to the central point of 2 Gy. Plans were produced for both isocentric and fixed FSD techniques, and monitor units (MUs) calculated by the host centre according to their normal method. Staff were asked to process the plans as they would for a normal patient.
Visiting centre and host centre chambers were also compared at a depth of 5 cm in the phantom so that the linac output calibration could be eliminated as a source of any discrepancy. The visiting centre chamber was then placed at the central point in the phantom, which was positioned by the host centre's radiographer. The phantom was then irradiated using the previously calculated MUs and three consistent readings were obtained for each beam.
For the first two rounds of the audit, only central doses were measured in each of the four configurations, i.e. isocentric or fixed FSD, with or without the lung insert. For later rounds, measurements without the lung were discontinued and additional readings were taken at the four off-axis points for the isocentric plan.
Electrons
Measurements were made using the visiting centre's dosemeter and phantom. Ionization chambers were either a Farmer or parallel plate ionization chamber (PPIC) such as the Nordic Association of Clinical Physics chamber (Scanditronix), Marcus (PTW type 23343; PTW-Freiburg Physikalisch-Technische Werkstätten, Freiberg, Germany) or Roos (PTW type 34001; PTW-Freiburg Physikalisch-Technische Werkstätten, Freiberg, Germany). Phantoms were photon or electron formulation ScanplasTM slabs, and chamber depths were adjusted to account for the effective point of measurement of the chamber. Electron audits were not constrained to be performed as a sequential series as for photons, because phantoms from each auditor's centre were used.
The response of a chamber in an electron beam needs to be corrected for the mean energy at depth, so the following simple procedure was adopted:
- Measurements were made at the stated dose maximum depth together with points close to 60% and 30% ionization depths (Figure 4). These positions were estimated at 0.95 and 1.1 times the stated depth of the 50% isodose on the central axis of the beam.
- The measured 50% ionization depth was then determined from these readings by interpolation, assuming linearity of this section of the depth dose curve.
- The practical range was calculated by extrapolating to zero dose (Rp' in Figure 4), ignoring the bremsstrahlung background for the purposes of these measurements.
- The mean electron energy at the surface and maximum dose positions were then calculated using the values for the 50% ionization depth and practical range. This enabled the appropriate stopping power and perturbation corrections to be applied for individual chambers.
Kilovoltage photons
For this energy range both the visiting and host centres made measurements for comparison with the host centre stated values. The Farmer dosemeter timer was used to integrate dose for a fixed period. Separate readings based on the machine timer were also made to determine the significance of any "ramp up" effects. Measurements were made using the host department's standard applicator (usually 100 mm square/circle) and also a second applicator with field dimensions not less than 40 mm.
For low energy X-rays (50–150 kVp), measurements were made in air at the end of the applicator. An inverse square correction was made to account for chamber displacement from the end of the applicator. Backscatter factors were applied to convert measurements to dose in-phantom.
For medium energy X-rays (150–300 kVp), measurements were made at a depth of 20 mm in a water phantom supplied by the audit group.
Procedural audit
Audit of the procedures affecting patient dose was performed by assessing the following four areas:
- Calibration of the dosemeter.
- Output calibration of the treatment unit.
- Quality control (QC) programme for the treatment unit.
- QC programme for the treatment planning system.
For each of these categories a number of questions were asked. First, a clear verbal explanation of the process was sought. Following this, documented evidence of work instructions, tolerances and frequencies of checks, responsibilities, corrective action, test equipment calibration and, finally, the results of QC checks were required. Tolerances and frequencies were compared with national standards.
National Physical Laboratory audit
A water filled Perspex phantom was used to make measurements of the beam quality index (TPR20/10), output and calibration of the field instrument. These measurements were performed by NPL staff using their own instrumentation. Results were compared with departmental beam data and local measurements.
Barometer survey
A portable barometer (Oregon Scientific (UK) Ltd, Maidenhead, UK) was used to audit atmospheric pressure as determined by local instruments. The common barometer was passed around all participating centres and readings were taken on a number of separate days to sample a range of pressures.
|
Results and discussion
|
|---|
Megavoltage photons
The maximum, minimum, mean and standard deviation of measurements are shown in Table 2 for the four audit rounds currently completed.
View this table:
[in this window]
[in a new window]
|
Table 2. Megavoltage intercomparison results. (a) Dose normalization measurements. Ratio of visiting centre dose measurements to host centre dose measurements at the normalization point (5 cm depth) for a 10 x 10 cm2 field. (b) Central point dose measurements from the three-field irradiation. Visiting centre dose measurement at the central point relative to the planned 2 Gy, assuming that the visiting dosemeter was correctly calibrated. (c) Off-axis dose measurements from the three-field irradiation relative to calculated (plan) dose, assuming a correct dose at the central point
|
|
Table 2a shows a direct comparison between the host and visiting centres' dosemeter readings at the normalization point.
Table 2b gives the ratio of measured doses at the centre of the phantom for the three-field plan, relative to the intended dose of 2 Gy. These measurements were made using the visiting centre's dosemeter, which is assumed to be correctly calibrated.
Table 2c summarises the off-axis dose error factors relative to the central dose. The dose error factors tabulated here are the ratio of the measured/calculated relative off-axis doses.
Mean dose ratios for all four techniques did not deviate by more than 0.7% from unity. However, this only reflects the average position and the maximum, minimum and standard deviation (SD) are more indicative of outlying results. Analysed in these terms, results for the first and second rounds indicate a trend in which isocentric plans without an inhomogeneity give the best results and fixed FSD plans with inhomogeneity give the worst results. This was expected in light of the fact that, for isocentric treatments using orthogonal fields, a deficit in dose from one field owing to a set up error may be compensated by an increase in dose from the opposing field, whereas for fixed FSD each field is individually set up. It is also to be expected that treatment planning systems may not adequately account for inhomogeneity, especially when simpler algorithms are used that do not separate primary and scattered radiation.
The main assessed criterion was the total measured dose at the central point relative to the planned dose of 2 Gy. All audits were within the 5% specification. Dose discrepancies could be owing to errors in a number of areas including dosemeter calibration, machine output and monitor unit calculation. These results are comparable with those from recent photon dosimetry intercomparisons, falling between the national photon dosimetry intercomparison [3] and the national electron dosimetry intercomparison (photon reference measurements) [9] in terms of standard deviation.
The off-axis dose error factors (Table 2c) indicate errors in addition to those at the central point. These indicate a difference between the calculated and measured dose distribution that may be undetectable by a measurement at the central point alone. Thus, even if the central point was correctly irradiated, the dose calculated at an off-axis point may be in error. The off-axis errors in these audit measurements were less than 3%.
Electrons
The results for the electron intercomparisons are shown in Figure 5. The results for the ratio of host and visitor determined doses were well within tolerance with a maximum spread of ±2.5%, a mean value of 1.004 and a standard deviation of 1.4%. These values compare well with the recent national electron dosimetry intercomparison [9], in which a mean value of 0.994 and a standard deviation of 1.8% was reported.
View larger version (15K):
[in this window]
[in a new window]
|
Figure 5. Electron intercomparison results as a function of nominal electron beam energy. DV, visiting centre dose measurement; DH, host centre dose measurement.
|
|
Larger differences for electrons relative to megavoltage photons may be due in part to the range of chambers used. This was the only audit where different types of ionization chambers are in routine use. In principle the chambers are exposed to much larger dose gradients for high energy electrons than for megavoltage photons, so differences in chamber design and construction may be more significant. There was no obvious link between dose discrepancy and chamber type or beam energy. However, there was a small increase in the discrepancy between the stated and measured values of the 50% dose depth and the practical range with increasing energy. The 50% dose depths were within 2 mm of the stated values and the practical range was within 3 mm. This was probably due to the zero bremsstrahlung approximation.
Kilovoltage photons
Results of the kilovoltage photon intercomparisons are given in Table 3 and Table 4 for low and medium energies, respectively. The results were all within tolerance except for one centre where an inappropriate inverse square correction was made for the dosemeter position in the audit. However, for both energy ranges the mean ratio of visitor to locally measured dose was within 0.8% of unity for the large applicators, with a SD of approximately 2%. For the smaller applicators the variation was greater, with a larger mean difference and SD. Results show a greater spread for the smaller applicators and lower energies and so the worst results are seen for a combination of both. Similar results have been reported elsewhere [10]. The ratio of visitor measured dose (DV) to the expected dose (DE) (based on host centre output tables), DV/DE, showed a larger variation than the ratio to the host centre dose (DH), DV/DH, measured the same day. This was probably because a kilovoltage treatment machine is frequently controlled by a timer rather than a dose integrating monitor chamber. With a machine timer, day-to-day variations in dose rate will contribute to the observed discrepancies.
Variation in dose is greater than that for megavoltage photons and reflects the greater uncertainties involved in using ionization chambers in the kilovoltage region, as discussed below. For smaller applicators, measurements with an ionization chamber of similar dimensions to the radiation field clearly introduce additional uncertainties.
Measurements made using the machine timer were up to 6% different from those made using the Farmer timer, indicating significant ramp up effects. These will be less significant at the higher dose levels used for patients. Local arrangements for dealing with this problem included the use of monitor chambers on medium energy beams and the use of timer corrections when calculating machine settings.
Procedural audit
Specific procedural audit non-conformities were not collected centrally and were sometimes omitted if time was short. Centres reported that this section of the audit was useful and showed an improving trend in the standard of documentation. All participating centres were in the process of implementing or operating ISO 9000 Quality Systems in line with Department of Health recommendations [11].
National Physical Laboratory audit
Results of three separate NPL audits performed at different centres are shown in Table 5 in terms of local value relative to NPL result. Excellent agreement was obtained with all parameters agreeing to within 1%. The possibility of the whole group being subject to a systematic error in the same direction is unlikely but possible for a peer-to-peer audit. However, the use of the NPL as a totally independent auditing body has effectively ruled this out.
Barometer survey
The barometer survey involved 11 centres and 12 barometers. Six of these were Fortin (mercury in glass), three were aneroid and three were digital. Only one centre applied a temperature correction to their Fortin barometer. The mean difference between the common and local instruments was -0.1±0.3% (1 SD). Only two barometers showed a variation of greater than 0.4% and the maximum deviation was 0.8%.
Common and host barometers were compared over a number of days to sample a wide range of pressures. A high level of consistency was seen for each barometer type, giving confidence in the common instrument. The common barometer was also compared with a local instrument at the beginning and end of the measurements and the results were found to agree within 0.2%. Systematic (type B) uncertainty in using this instrument was estimated at ±0.1%.
|
Uncertainties
|
|---|
The aim of this audit programme is to ensure that there are no significant dosimetric discrepancies arising from the equipment, protocols and methods used by a particular radiotherapy centre. This has been implemented by intercomparison between centres, which themselves are also subject to audit. However, in order to determine whether a difference is significant, the uncertainties associated with the measurement process must be understood. The audit involves comparing two identical or similar dosemeters with calibrations traceable back to a common standard using the same dosimetric protocols. In this situation most systematic (type B) uncertainties involved in the calibration will cancel, and the differences between the two systems will then be owing to random (type A) uncertainties and any unaccounted for systematic differences. This is why a Farmer dosemeter calibration factor will differ by only fractions of a percent when measured on different occasions despite the total uncertainty involved (including systematic) being several percent.
Estimates of uncertainties for the different modalities are shown in Table 6. Each calibration step can be seen to introduce up to 0.5% uncertainty, and for electrons this is increased by the use of a PPIC that must be calibrated against a Farmer field instrument. The positioning uncertainty has been estimated for a nominal 1 mm displacement, which has the smallest effect for megavoltage photons because of the low dose gradient. The use of the energy correction curve for megavoltage and kilovoltage photons introduces approximately 1% uncertainty, whereas for electrons the energy correction is dependent on the stopping power ratio, which varies by appraximately 1% per MeV and 0.5% per mm. Adding the uncertainties in quadrature gives a maximum uncertainty of approximately 2%. Setting the threshold for investigation at 5% gives an unambiguous indication of systematic differences.
View this table:
[in this window]
[in a new window]
|
Table 6. Uncertainties. These figures are estimates of the percentage random (type A) uncertainty (2 standard deviations from the mean or 95% confidence level)
|
|
Uncertainties associated with delivery of radiation dose to a patient have been estimated at approximately 2.5% for standard conditions [12], which is in good agreement with this study.
|
Conclusions
|
|---|
Dosimetric accuracy is a vital foundation for the transfer of clinical experience between centres. These results show that this is obtained for the categories covered by the audit. There were no overall dosimetry errors above the 5% action level, though errors for individual megavoltage photon fields involving the lung inhomogeneity approached this level. Although small, dose discrepancies generally increased with complexity, showing that vigilance is necessary for some techniques.
The fact that several centres within the group have implemented quality management systems such as ISO 9000 was reflected in the high quality of documentation as determined by the procedural audits. The introduction of such systems requires rigorous calibration arrangements, not only for dosemeters but also for other instruments that have a bearing on the absorbed dose. It is clearly possible for an external audit to determine absorbed dose with an uncertainty below the 5% action level. However, dosimetric audit as reported in this paper is seen as a complementary and not an alternative form of audit as it helps determine the effectiveness of quality measures. This simple, pragmatic approach to dosimetric audit in radiotherapy should be achievable by most radiotherapy physics departments. Work in progress includes the design of audits to cover brachytherapy and simulators.
|
Acknowledgments
|
|---|
The medical physics, radiography and medical staff involved in this audit programme are thanked for their continued contributions, interest and support. The National Physical Laboratory are also thanked for their provision of an audit service.
Received for publication November 12, 2001.
Revision received March 7, 2002.
Accepted for publication April 4, 2002.
|
References
|
|---|
- Thwaites B, Burlin TE, Joslin CAF. The report of the Committee of Enquiry into the overdoses administered in the department of radiotherapy. Royal Devon and Exeter Hospital (Wonford) (The Exeter District Health Authority). 1988.
- Baldwin P, Joslin CAF, Williams PC. Report of the independent enquiry commissioned by the West Midlands Regional Health Authority into the conduct of isocentric radiotherapy at the North Staffordshire Royal Infirmary between 1982 and 1991. 1992.
- Thwaites DI, Williams JR, Aird EG, Klevenhagen SC, Williams PC. A dosimetric intercomparison of megavoltage photon beams in UK radiotherapy centres. Phys Med Biol 1992;37:445–61.[Medline]
- Bonnett DE, Mills JA, Aukett RJ, Martin-Smith P. The development of an interdepartmental audit as part of a physics quality assurance programme for external beam therapy. Br J Radiol 1994;67:275–82.[Abstract/Free Full Text]
- UK Department of Health. Manual of Cancer Services Standards. London: Department of Health, 2000. www.doh.gov.uk
- Institute of Physical Sciences in Medicine. Code of practice for high-energy photon therapy dosimetry based on the NPL absorbed dose calibration service. Phys Med Biol 1990;35:1355–60.
- Institution of Physics and Engineering in Medicine and Biology. The IPEMB code of practice for the determination of absorbed dose for x-rays below 300 kV generating potential (0.035 mm Al-4 mm Cu HVL; 10-300 kV generating potential). Phys Med Biol 1996;41:2605–25.[Medline]
- Institution of Physics and Engineering in Medicine and Biology. 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 1996;41:2557–603.[Medline]
- Nisbet A, Thwaites D. A dosimetric intercomparison of electron beams in UK radiotherapy centres. Phys Med Biol 1997;42:2393–409.[Medline]
- Nisbet A, Thwaites D, Sheridan M. A dosimetric intercomparison of kilovoltage X-rays, megavoltage photons and electrons in the Republic of Ireland. Radiother Oncol 1998;48:95–101.[Medline]
- UK Department of Health. Quality Assurance in Radiotherapy: A Quality Management System for Radiotherapy. London: Department of Health, 1994.
- International Commission on Radiation Units and Measurements. Report 24: Determination of Absorbed Dose in a Patient Irradiated by beams of X or Gamma Rays in Radiotherapeutic Procedures. Bethesda, MD: ICRU, 1976.
This article has been cited by other articles:
|
|
|
|
 
N L A BURTON, J BRIMELOW, and A D WELSH
A regional audit of kilovoltage X-rays -- a single centre approach
Br. J. Radiol.,
May 1, 2008;
81(965):
422 - 426.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
