“Characterization of ELEKTA SRS cone collimator using high spatial resolution monolithic silicon detector array”

Abstract Purpose To investigate the accuracy of the dosimetry of radiation fields produced by small ELEKTA cone collimators used for stereotactic radiosurgery treatments (SRS) using commercially available detectors EBT3 GafchromicTM film, IBA Stereotactic diode (SFD), and the recently developed detector DUO, which is a monolithic silicon orthogonal linear diode array detector. Methods These three detectors were used for the measurement of beam profiles, output factors, and percentage depth dose for SRS cone collimators with cone sizes ranging from 5 to 50 mm diameter. The measurements were performed at 10 cm depth and 90 cm SSD. Results The SRS cone beam profiles measured with DUO, EBT3 film, and IBA SFD agreed well, results being in agreement within ±0.5 mm in the FWHM, and ±0.7 mm in the penumbra region. The output factor measured by DUO with 0.5 mm air gap above agrees within ±1% with EBT3. The OF measured by IBA SFD (corrected for the over‐response) agreed with both EBT3 and DUO within ±2%. All three detectors agree within ±2% for PDD measurements for all SRS cones. Conclusions The characteristics of the ELEKTA SRS cone collimator have been evaluated by using a monolithic silicon high spatial resolution detector DUO, EBT3, and IBA SFD diode. The DUO detector is suitable for fast real‐time quality assurance dosimetry in small radiation fields typical for SRS/SRT. This has been demonstrated by its good agreement of measured doses with EBT 3 films.

micro-MLC or a circular cone collimator. Small field dosimetry is known to be complex due to the existence of lateral charge disequilibrium, source occlusion, and volumetric effect if the size of the detector is comparable to the field size. 1 Dosimetry protocols such as IAEA TRS-398 2 and AAPM TG-51 3 provide guidelines for traditional radiation fields (from 3 9 3 cm 2 to 40 9 40 cm 2 ), but are not designed for small field dosimetry. By extending the use of these codes for small field dosimetry, some researchers found a difference in the output factors by up to 30% between different detectors. [4][5][6] For accurate and precise dosimetry of small field techniques, many groups have been working to get new protocols for MV small field dosimetry, which recommends specific requirements of detectors and use of different strategies. [7][8][9][10] Nowadays, there are a number of commercially available detectors for small field dosimetry, each having some characteristic that makes it suitable for some small field measurements. However, there is no single ideal detector that fulfills all the required characteristics for SRS dosimetry until now.
Previously, many researchers suggested the use of different small size detectors and compare between them to overcome the drawbacks of each detector, to get the full characteristics of small beams. 11 Different groups are working these days on silicon array detectors [12][13][14] ; taking advantage of their excellent spatial resolution and small size compared with ion chambers, their real-time measurements compared with EBT3 and TLDs, and their high sensitivity compared with ion chambers and EBT3. In addition, they are less expensive than diamond detectors and have reasonable uniformity and are more practical compared with gel dosimetry. Diodes work without external bias and they provide almost energy independence of mass collision stopping power ratios of silicon to water for electrons for clinical use in the range from 4 to 20 MeV. 15 However, silicon detectors have some limitations, which need to be characterized to derive appropriate correction factors or to minimize these effects. 15 The sensitivity of diodes is increased by increasing the instantaneous dose per pulse. This effect can be significantly minimized by selecting p-type, preirradiation by large electron doses, or using heavy platinum doping or epitaxial guarded silicon diode. [16][17][18] Another factor that can affect silicon diodes is the temperature, which could affect the level of recombination, and hence the sensitivity of the detector in a linear correlation. This variation in the sensitivity can be canceled by the preirradiation of the diode to high dose. 19,20 The energy dependence of a silicon diode at low photon energy (<200 keV) is related to the geometry and material surrounding the diode and diode material, which usually have higher atomic number when compared with water. Therefore, each diode should be selected with caution depending on the energy range it was designed for. Previous studies showed that the energy dependence increased as the thickness of the buildup material increased. 21 However, for small field dosimetry such as SRS, diodes are used without shielding (almost no buildup material). 22 In addition, it has been reported that silicon diodes have angular dependence when they are used in rotational beam measurements. This depends on their design and packaging. Several research groups have studied the angular dependence of diodes and found an over-response up to 30% for 6 MV photon beams at 90°AE 10°and 270°AE 10°. [23][24][25] Using different techniques can eliminate the angular dependence, such as adding half pipe-shaped boluses, 26 filling air gaps with sheets of lucite and pieces of copper, 27 or by applying angular correction factors. 28 Recently, edgeless diodes were developed which, combined with a special drop in packaging in kapton tails, avoided high Z overlayers, and demonstrated angular independence in MV photon field. 29 The silicon diode arrays due to their atomic composition is different to water and leading to beam perturbation effect due to difference of secondary electron energy fluence in water and silicon radiation sensitive volume of the detector, therefore violating the condition for CPE and breaking down Bragg-Gray cavity theory. The low-energy part of the differential electron fluence in silicon is larger than in not perturbed water. This difference is governed by the density term in electron mass stopping power of the radiation sensitive volume of the detector relative to water and by extracameral material (packaging) of the detector and depends on difference of ionizing potentials of Si and water. This is why calibration of silicon detector in a large beam where CPE is established and difference in electron energy fluence in water and silicon radiation sensitive volume is negligible is not valid for small field. For more details reader can references to Andreo (2018). 30 The performance of the detector can be improved by removing the high atomic number and density materials near the sensitive volume or by adding low atomic number and density materials around the sensitive volume to compensate the over-response due to higher Z material than water. 31,32 To compensate the effect of volume averaging due to the large sensitive volume of detector, the deconvolution method could be used. However, this method is complicated if done manually and is not practical in clinical setting as large number of profiles require long postprocessing time. 15 Hence, different scientific papers have focused on finding proper correction factors to minimize the over-response of the different diodes by studying their response for different beam qualities, field sizes, and types of LINACs in comparison with EBT3 and Monte Carlo simulation. 4,[33][34][35][36][37] The Center of Medical Radiation Physics (CMRP) at the University of Wollongong is specializing in development of different types of silicon detectors for radiotherapy dose verification applications.
The purpose of this study is to characterize recently released ELEKTA circular SRS cone collimator by using the high spatial resolution monolithic silicon diode array, DUO, for relative dosimetry of small radiation fields and compare it with some available commercial detectors.

2.A | Source and LINAC system
All measurements were performed in the radiation oncology department of the Nelune Comprehensive Cancer Centre, Prince of Wales SHUKAILI ET AL.

| 115
Hospital (Randwick, NSW, Australia) by using 6 MV flattened photon beam from an ELEKTA Axesse TM linear accelerator with a retrofitted Agility head (Elekta AB, Stockholm, Sweden), adapted for stereotactic treatment by using an additional gantry mounted ELEKTA cone collimators system. The cone collimators diameter varied from 5 to 50 mm, every 2.5 mm up to 20 mm, and then every 5 mm. The X (MLC) and Y jaw positions were set as in Table 1, for all measurements with these circular cones.

2.B | Detectors
The IBA stereotactic field diode (SFD) (IBA dosimetry, Nuremberg, Germany) is a p-type unshielded low-resistivity silicon diode. It has active volume of 0.017 mm 3 with 0.06 mm thickness and 0.6 mm diameter. Its sensitivity is 6 nC/Gy. 38 Morales et al. 39 had studied the dose rate dependence of SFD by measuring the PDD for field size 10 9 10 cm 2 and compared it with the ionization chamber.
The results showed an agreement within 0.5%. It was confirmed that IBA SFD is almost dose rate independent. However, many authors have published correction factors for SFD detectors to correct the over-response due to the density, nonwater equivalency, and the volume averaging effects for different LINACs and different depths in water. 4,10,[33][34][35][36] In this study, we have used the correction factors provided by IAEA-AAPM TRS-483 10 to overcome the over-response of IBA SFD detector in the output factors (OF) measurements.

2.B.2 | Gafchromic TM Films (EBT3)
The EBT3 film is comprised of a 27 lm active layer, sandwiched between two 120 lm matte polyester layers, which make it more robust in principle for use in water. It is used in conjunction with an Epson 10000 XL film scanner, which enables RGB multichannel analysis. The dose range of the EBT3 film is up to 10 Gy with the red color channel. The EBT3 film is dose rate independent, near tissue equivalent, and can be used in water phantoms. 40 The spatial resolution of EBT3 is determined mostly by the scanner resolution.

2.B.3 | DUO detector
The DUO detector was designed and developed at the CMRP. DUO is made of two orthogonal axial monolithic silicon diode arrays of 505 pixels with a total size of 52 9 52 mm 2 . Each diode measures 0.04 mm in the direction of linear array axis, and 0.8 mm in the orthogonal direction, with a pitch of 0.2 mm to provide the required spatial resolution for measurements of the sharp fall off penumbra region. An air gap of 0.5 mm was introduced above the DUO detector to compensate the over-response that results from the high density of silicon and extracameral components. 41 The diodes are operating in a passive mode (no bias). The DUO detector is placed on a thin printed circuit board (PCB-500 lm thick), and connected to the fast data acquisition system (DAQ). The DAQ is based on a commercial analog front-end ASIC named AFE0064 (Texas Instruments), which consists of 64 channels with each providing an analog differential output proportional to the charge accumulated in a capacitor during a determined time frame. The DAQ system contains eight AFE0064 chips, for 512 channels in total. The acquisition of the data is synchronized with the LINAC by using the pulse-by-pulse scope trigger available on the service panel. These are connected to a fully programmable gate array (FPGA) that used for a signal processing. The digital data then sent to the computer via USB cable for analysis.

4.A | Beam profiles
Both cross-plane and in-plane beam profiles of all cone sizes were measured using DUO, SFD, and EBT3 films. The cross-plane profiles of the radiation fields from SRS cones of different sizes were compared as shown in Fig. 3. The beam profiles were normalized to 1 at the center. In general, there is an agreement between the three detectors in both X and Y profiles.
The beam profile parameters (FWHM and penumbra width 20%-80%) were calculated for each cone size in the X and Y profiles for the three detectors as shown in Tables 2 and 3. The 20%-80% penumbra width is calculated as the average between the 20%-80% ascending and descending parts on the beam profiles. There is a good agreement among DUO, EBT3 films, and IBA SFD within AE0.5 mm for FWHM.
The 20%-80% penumbra widths showed a good agreement within AE0.7 mm for all cone sizes. By comparing the profile parameters between X profiles and Y profiles, there were differences up to 0.3, 0.6, and 0.8 mm in FWHM measured by DUO, EBT3, and IBA SFD, respectively. In terms of the penumbra width, the differences between X profiles and Y profiles were high, up to 0.5, 0.6, and 1.0 mm, measured by DUO, EBT3, and IBA SFD, respectively.   The comparison of PDD measured by the three detectors for all cone sizes showed overall average agreement of AE0.5%; with maximum differences within AE2% for DUO/EBT3 and SFD/EBT3.

| DISCUSSION
In this study, three detectors were used to measure the beam profiles, output factors, and PDD for ELEKTA SRS cone collimators.
One of these detectors is water equivalent and has sufficient spatial resolution for small field dosimetry (EBT3) and two are diodes (DUO 2D high spatial resolution monolithic diode array and a single diode IBA SFD), which required correction for the nonwater equivalence and/or volume averaging effects.

5.A | Beam profiles
By comparing the beam profiles of the SRS cone collimators, DUO shows good agreement with the EBT3 films in the "in-field" area, but slightly lower dose in the "out-field" area for all the cone sizes. This could be due to the dose rate dependence of DUO for very low dose rate.
The normalized response measured by SFD shows higher drop off in the penumbra region than DUO and EBT3 for cone sizes smaller than 20 mm, as shown in Fig. 7. Taylor et al. 42 also found this.
F I G 2 . 2D dose map for 5 mm cone diameter at 10 cm depth, (dose in cGy).
The same behavior was found in the comparison of Y profiles. This could be due to the large sensitive volume and high Z extracameral material that has been explained in more detail by Benmakhlouf and Andreo. 43 It could be also improved by using the deconvolution method, which has been developed to obtain beam profiles independent of the detector size. 44 The results depend on both the pitch size (and spatial sampling in case of SFD) and the size of the sensitive volume of the detector. DUO shows good agreement in comparison with EBT3 and IBA SFD diode in both FWHM and penumbra width, for both X and Y profiles.
It was observed that penumbra measured in X/Y directions for cones is less than for Varian linac equivalent square fields as presented in Ref. [41]. Smaller penumbra for cones in comparison with penumbra for equivalent square filed is partially due to cones closer to the phantom surface and different geometry and scattering in cones in comparison with jaws. This relative difference is decreasing with field increasing as expected.
However, all three detectors show differences between X and Y profiles in terms of FWHM and 20%-80% penumbra widths (Tables 2 and 3). X-profile parameters were higher than Y-profile parameters, which could be due to the elliptical-shaped focal spot of the ELEKTA Axesse linac source. Different groups have studied the shape and size of the x-ray source and found that it depends on the linac model, with mostly elliptical shape. [45][46][47] The observed eliptical shape focal spot for Elekta was not observed for Varian Linac. 41

5.B | Output factors
A number of studies reported that diode over-response in small fields is due to the density of silicon and the extracameral materials  surrounding the detector. However, it has been recently clarified that electron density, rather than density as a fundamental characteristic of material, is driving the diode response through density effect in mass stopping power of electrons and ionization potential of silicon. 42 In this study, the DUO silicon detector was corrected by using a 0.5 mm air gap above detector as studied earlier. 41 This correction provides an overall agreement in the output factors for cone sizes from 5 to 50 mm, between DUO and EBT3 within AE0.7%.
IBA SFD diode was corrected by using the correction factors provided by TRS-483. 10 This led to better agreement in the OF measurements between IBA SFD and EBT3, where the difference reduced from 5.7% to 2% for the 5 mm cone size. The average agreement in the OF for all cone sizes is about AE0.8% after applying the correction factors for the IBA SFD.
Recently published AAPM practice guidelines recommend SRS-SBRT annual QA for the OF and the tolerance is AE2% from the T A B L E 2 FWHM measured by DUO, EBT3, and SFD in both X profiles and Y profiles.  baseline for >1.0 cm apertures, and AE5% from the baseline for ≤1.0 cm apertures. 48 Our results showed a good agreement between the three detectors indicating that DUO could be a suitable candidate for stereotactic cones regular QA. Its practicality and online reading would make it the preferred option over the other two detectors in clinical settings.

5.C | Percentage depth dose
In the PDD measurements, the depth of maximum dose (d max ) was changed as a function of cone diameter and it was difficult to detect the exact d max as it depends on the available phantom thickness and the resolution of the detector. Therefore, all PDD measurements were normalized to 100 mm depth.
The comparison of PDDs between DUO, EBT3, and IBA SFD shows an agreement within AE2% for small cone sizes up to 20 mm, and then the agreement increased to AE1.5% for the larger cones.

| CONCLUSION
It was recommended in the IPEM report 103 to avoid the use of ion chambers for the small beam profile measurements due to various F I G 7 . Penumbra comparison between DUO, SFD, and EBT3 for SRS cone diameter: 5 mm.
issues such as their large volume, which causes volume averaging effects, therefore artificial broadening in the penumbra region, and increasing the FWHM of the measured beam profile. 9 In addition, the use of small air-filled ionization chambers causes under-response to the dose due to their low mass density of air. 49 Thus, the use of Gafchromic TM films and diodes was recommended. To facilitate small field dosimetry and propose an alternative to time-consuming films and single diode water tank measurements, CMRP designed a monolithic silicon detector DUO with 0.2 mm spatial resolution for SRS dosimetry.
The effect of the high density of silicon on the output factor measurements was successfully compensated by introducing an air gap of 0.5 mm above it, as proposed by others for single diodes. 50,51 The circular SRS cones used with the ELEKTA LINAC were characterized in terms of the beam profiles, output factors, and PDD for cone diameters ranging from 5 to 50 mm and by using high spatial resolution detectors DUO, EBT3 films, and IBA SFD. The results showed that DUO agrees with EBT3 in terms of beam profiles and output factors.
The good agreement between DUO, EBT3, and IBA SFD in the profiles shows a difference within AE0.5 mm in the FWHM and AE0.7 mm in the 20%-80% in the penumbra width. The output factors show very good agreement between DUO and EBT3 for all cone sizes within AE0.7%. IBA SFD detector agrees with the EBT3 and DUO measurements of output factors after applying the volume averaging correction factors, which shows an average agreement of AE0.8%, with maximum difference about 2% for 5 mm cone. In the percentage depth dose curves, there is a good agreement among DUO, SFD, and EBT3 for all depths of all measured cone diameters; with average difference within AE0.5% and maximum difference within AE2%.
In conclusion, ELEKTA SRS cones have been characterized by using three high spatial resolution detectors, two are commercially available and one is designed by CMRP at UOW. DUO is a suitable detector for fast SRS/SRT dosimetry as it has excellent resolution (0.2 mm) in a direction of steepest dose gradient, on line data analysis, and it provides both X and Y profiles. The good agreement with EBT3 films measurements confirms its accurate and precise data for SRS/ SRT measurements, which is the treatment modality where small-field dosimetry is paramount, and DUO can be applied successfully.

This project is funded by the National Health and Medical Research
Council of Australia (Grant ID1030159). Al Shukaili would like to thank the government of Sultanate of Oman for their continuous support and for providing the scholarship funding.

CONF LICTS OF INTEREST
The authors have no relevant conflicts of interest to disclose.