Survey of 5 mm small‐field output factor measurements in Australia

Abstract The Australian Radiation Protection and Nuclear Safety Agency (ARPANSA) held a comparison exercise in April 2016 where participants came to ARPANSA and measured the output factor of a nominal 5 mm cone attached to the ARPANSA Elekta Synergy (Elekta, Crawley, UK) linear accelerator. The goal of the exercise was to compare the consistency and methods used by independent medical physicists in measuring small‐field output factors. ARPANSA provided a three‐dimensional scanning tank for detector setup and positioning, but the participants were required to measure the output factor with their own detectors. No information regarding output factors previously measured was supplied to participants to make each result as independent as possible. Fifteen groups travelled to ARPANSA bringing a wide range of detectors and methods. A total of 30 measurements of the output factor were made. The standard deviation of the measurements (excluding one expected outlier from an uncorrected ionization chamber measurement) was 3.6%. The results provide an insight into the consistency of small‐field dosimetry being performed in Australia and New Zealand at the present time.

With the lack of an accepted protocol, it is currently up to the clinical medical physicist to decide how to perform dosimetry of small fields. The choice of detector and use of associated correction factors are very important for dosimetric accuracy. Das et al. described the challenges of small-field dosimetry and how they could be overcome. 3 Cranmer-Sargison et al. measured output ratios for a variety of diodes on two different accelerators highlighting some of the practical aspects of small-field dosimetry. 5 Kairn et al. have published a practical guide on how to perform small-field output factor measurements with diodes and microchambers. 6 The comparison exercise was initiated to investigate the consistency of small-field measurements in Australia. ARPANSA does not have a primary standard for absorbed dose in small fields which can measure the dose output of the nominal 5 mm cone. Therefore, the results presented here will not be compared to a primary standard measurement of the output factor. ARPANSA participated in the comparison, but its results should not be considered any more valid or accurate than any other result in the comparison.

| ME TH OD
Each participant was given detailed information well before their allocated time to allow appropriate choice of detectors for the measurements needed. All measurements were performed on the ARPANSA Elekta Synergy linear accelerator using the 6 MV (TPR 20,10 = 0.673) photon beam. The output factor to be measured was the ratio of the dose in the 5 mm cone-defined field to the dose in the 10 9 10 cm multileaf collimator (Elekta MLCi2)-defined field for a given number of MU. The reference conditions for the measurement are shown in Fig. 1.
The 5 mm cone was commercially manufactured (Elekta Stereotactic Collimator, product number TRT 0065). It was screwed into a frame (Elekta product number MRT 13541) which was bolted onto the head of the linear accelerator. The frame, when in position, was located approximately 60 cm from the location of the accelerator target. The cone was composed of a combination of Sn, Bi, and Pb.
Its length was 9 cm and it produced a circular field at isocenter of nominal diameter 5 mm. It was not interlocked with the accelerator and care was taken to ensure that the jaws were set to a field size of 3 9 3 cm when the cone was in place. This setting was previously determined to inhibit any radiation leaking around the cone and minimize any effect on the radiation passing through the cone. The reproducibility of the cone placement was tested by removing and remounting the cone and frame nine times. The maximum difference in the field produced by the cone in both lateral directions (i.e., perpendicular to the beam direction) was 0.2 mm. This was measured by scanning the field with a PTW 60017 electron diode (PTW-Freiburg GmbH, Freiburg, Germany) at a source to surface distance (SSD) of 95 cm and depth of 5 cm in inline and crossline orientations to determine the central position of the field. The effect on the output factor of a cone misalignment was measured with a PTW 60017 electron diode and a PTW 60019 microDiamond detector. The cone was deliberately shifted by 0.3 mm in both inline and crossline directions to simulate the worst-case scenario misalignment after removing and then replacing the cone. If the detector was not realigned with the center of the field, the percentage change in the measured output factor was À1.4% and À0.8% for the microDiamond and electron diode, respectively. It was expected that the larger diameter detector will see the biggest difference, and that is what was observed.
However, if the detectors were realigned in the field with the shifted cone, the change in output factor measured for both detectors was less than 0.1%. The dosimetric field width of the 5 mm cone field was defined as the 50% isodose level normalized to central axis. It was measured with a PTW 60017 electron diode and a PTW 60019 microDiamond at 5 cm depth in a water phantom with a 95 cm source to surface distance. Both detectors were oriented parallel to the beam direction. The crossline full width half maximum was 5.8 mm and the inline full width half maximum was 5.9 mm, with both detectors agreeing to within 0.1 mm in both directions.
An isocentric setup of 95 cm SSD and 5 cm depth in water was chosen for the comparison. These values were based on the feedback given by several clinics who had measured small-field output factors for commissioning data for a range of treatment planning systems. As there is no accepted protocol for small-field dosimetry and different treatment planning systems have different SSD and depth setups, the choice was arbitrary. The combination of SSD of 95 cm and depth of 5 cm had the advantage that IAEA TRS-398 could be used to perform reference dosimetry in the 10 9 10 cm field for the 6 MV photon beam.
ARPANSA provided an IBA (IBA Dosimetry GmbH, Schwarzenbruck, Germany) Blue Phantom 3D scanning water tank with associated software, virtual water, adapters, electrometer, triaxial cables, and various mounts for participant detectors in the water phantom.
All detectors were brought to ARPANSA and mounted in the water F I G . 1. Experimental setup for the small-field output factor measurement indicating the geometry for the reference and small fields. In this example, the 10 9 10 cm field was being measured by a Farmer chamber and the small field by a diode; however, the detector types were chosen by each participant. The measurement by both detectors in an intermediate field to relate the two displayed measurements is not shown.

| RESULTS
The departments that participated in the comparison are listed in alphabetical order in Table 1. Table 2 lists the types and dimensions of the active detectors used in the comparison. All results including some experimental details are shown in Tables 3 and 4. Table 3 shows the active detector results while Table 4 shows Gafchromic EBT3 film results. The "Uncorrected output factor" in Table 3 is the ratio of the small-field detector reading in the 5 mm field to the small-field detector reading in the 10 9 10 cm field if it was used in both the fields. Otherwise, it is the ratio of the reading of the small-field detector in the 5 mm field and an intermediate These results are presented to show the influence of the subsequent corrections applied and to allow these results to be compared using different correction factors. The correction factors used by the participants came from a variety of sources and these are shown in the table. For the active detectors, each participant was also asked if this detector was used clinically and if so, what was the smallest field measured with this detector. This is also shown in Table 3. Table 4 displays the Gafchromic (Ashland Inc., KY, USA) EBT3 film measured output factors and gives some information about the measurements and subsequent analysis. Some groups measured a dose linearity curve at ARPANSA although this was challenging considering the limited time. We believe that they did this to ensure that the same batch of film was used for both output and linearity The results for each detector type are summarized below.

Germany) Razor detector
There were four measurements performed with the IBA Razor diode.
In one case, the diode was used in the 10 9 10 cm and 5 mm fields; however, in the other three cases, the response of the diode in the 5 mm field was "daisy chained" to the response of an ionization chamber in the 10 9 10 cm field through an intermediate field. 19 For one result, a correction factor was estimated from Cranmer-Sargison et al., 7 and for another result, a volume averaging correction was applied calculated using the 5 mm field profile. The remaining two results did not apply a correction factor. The standard deviation of the results was 1.1% and the average value output factor reported with this detector was 0.642.

3.B | Sun Nuclear (Sun Nuclear, FL, USA) Edge detector
Two measurements were made with the Sun Nuclear Edge detector.
In both the cases, the detectors were used in the 5 mm cone fields and 10 9 10 cm fields. In one case, a correction factor was applied from Bassinet et al., 8 and in the other case, no correction was applied. The standard deviation of the two results was 3.9% and the average was 0.649. was used to measure the 5 mm field and the 10 9 10 cm field.
No correction was applied to the A1SL result, and it is an obvious outlier being significantly lower than the other ionization chamber reported output factors. The correction factors for the three corrected ionization chamber measurements were large and ranged from 12% to 22%. Two were from the literature 9,17 and one was based on a volume averaging correction from a measured profile.
The average of these four results was 0.558 with a standard deviation of 10.6%. The correction factors applied were from a range of sources and one was a weighted average from five different papers. 10,11,[13][14][15] The average of the reported output factors was 0.613 with a standard deviation of 2.5%.

3.E | IBA EFD electron
There were two instances of output factor measurements with the IBA EFD Electron diode. Both measurements in the 5 mm field were daisy chained back to IBA CC13 measurements in the 10 9 10 cm  18 We do not believe that the use of Virtual Water TM will significantly affect the film-measured output factors. We have measured percentage depth dose curves in this beam in Virtual Water TM and a water phantom and observed no local differences down to a depth of 20 cm. The average of the output factor measurements made by film was 0.608 with a standard deviation of 3.3%.

3.G | PTW 60019 microDiamond
Three measurements were made with PTW 60019 microDiamond detectors. In all three cases, the microDiamond was used in both the 5 mm and 10 9 10 cm fields. All results were corrected with data  | 333 from the literature. 10,16 Interestingly, the published correction factors do not agree on whether the microDiamond will under respond or over respond in the 5 mm field. The average of the microDiamond output factor results was 0.620 with a standard deviation of 0.4%.

3.H | PTW 60018 SRS diode
One measurement was performed with a PTW 60018 SRS diode. Its response in the 5 mm field was daisy chained to the response of an IBA CC04 ionization chamber in the 10 9 10 cm chamber through an intermediate field of size 3 9 3 cm. A correction factor from the literature was applied to the result. 12 The output factor reported was 0.612.

3.I | PTW 60017 electron diode
One measurement was made with the PTW 60017 electron diode. It was used to measure the output in both the 5 mm and 10 9 10 cm fields. A correction factor from the literature 11

3.K | Air core plastic scintillation detector
A noncommercial air core plastic scintillation detector was also used to measure the output factor. It was used in both the 5 mm and  Each participant was asked to provide the uncertainty in their output factor measurements including both random and systematic components. The uncertainty for each output factor measurement shown in Fig. 2  | 335 in the clinic should include Type B uncertainties as outlined in the paper by Hill. 22 The standard deviation of all reported output factors was 5.6%. If the outlier ionization chamber result is excluded, the standard deviation reduces to 3.6%. This is still larger than the general recommendation on accuracy in radiotherapy of 3%. 23 The results were also analyzed by dividing by the correction factors listed in Table 3 to investigate the influence of the stated correction factors. The standard deviation was then 8.4% (or 7.2% excluding the ionization chamber outlier result). The large increase is mainly due to the ion chamber results which require large correction factors. If these are excluded, the standard deviation is reduced to 3.8%. For comparison, the IAEA TRS-398 estimate of the standard uncertainty in the calibration of a high-energy photon beam in reference conditions is 1.5%. 1 Nine of the 30 reported output factors were from detectors which had not been used to measure clinically employed small fields.
If these output factors were excluded, then the standard deviation of the results increased from 3.6% to 3.9% indicating that these results did not contribute to an increase in the overall variability in the results. If the results only included those detectors which had been used to measure clinical fields equal to or smaller than that measured in this comparison, then the standard deviation remained unchanged at 3.6% also indicating no bias when only including those results.