Machine characterization and central axis depth dose data of a superficial x-ray radiotherapy unit

Objectives. The purpose of this study is to present data from the clinical commissioning of an Xstrahl 150 x-ray unit used for superficial radiotherapy, Methods. Commissioning tasks included vendor acceptance tests, timer reproducibility, linearity and end-effect measurements, half-value layer (HVL) measurements, inverse square law verification, head-leakage measurements, and beam output calibration. In addition, percent depth dose (PDD) curves were determined for different combinations of filter/kV settings and applicators. Automated PDD water phantom scans were performed utilizing four contemporary detectors: a microDiamond detector, a microSilicon detector, an EDGE detector, and a PinPoint ionization chamber. The measured PDD data were compared to the published values in BJR Supplement 25, Results. The x-ray unit’s mechanical, safety, and radiation characteristics were within vendor-stated specifications. Across sixty commissioned x-ray beams, the PDDs determined in water using solid state detectors were in excellent agreement with the BJR 25 data. For the lower (<100 kVp) and medium-energy (≥100 kVp) superficial beams the average agreement was within [−3.6,+0.4]% and [−3.7,+1.4]% range, respectively. For the high-energy superficial (low-energy orthovoltage) x-rays at 150 kVp, the average difference for the largest 20 × 20 cm2 collimator was (−0.7 ± 1.0)%, Conclusions. This study presents machine characterization data collected for clinical use of a superficial x-ray unit. Special focus was placed on utilizing contemporary detectors and techniques for the relative PDD measurements using a motorized water phantom. The results in this study confirm that the aggregate values published in the BJR 25 report still serve as a valid benchmark when comparing data from site-specific measurements, or the reference data for clinical utilization without such measurements, Advances in knowledge. This paper presents comprehensive data from the acceptance and commissioning of a modern kilovoltage superficial x-ray radiotherapy machine. Comparisons between the PDD data measured in this study using different detectors and BJR 25 data are highlighted.

Kilovoltage radiotherapy units are utilized in many radiation oncology departments due to their ease of use, quality of treatment, and relative cost-effectiveness. In many cases kV radiotherapy offers simple, yet very effective treatment options compared to megavoltage photon and electron alternatives. Beyond the radiotherapy horizon, kV x-ray beams are predominantly utilized for radiographic, fluoroscopic and tomographic diagnostic applications and research (Yin et al, 2009, Xiong et al, 2017, Fontenot et al, 2014, Mccullough et al, 2021. A multidisciplinary crossover between diagnostic and therapeutic domain exemplifies the clinical use of on-board kV imaging on linear accelerators for patient setup and target localization. Additionally, kV x-rays are extensively used for the preclinical and basic radiobiology research (Zhong et al, 2020, Gronberg et al, 2020, Desrosiers et al, 2013, Pidikiti et al, 2011. A characteristic feature of superficial kV beams is that dose is at a maximum at the skin surface which then falls off rapidly with depth due to beam attenuation and scattering. Therapeutic kV beams are therefore used predominantly for treatment of skin cancers and management of other dermatological conditions such as keloid scars, mycosis fungoides, and psoriasis. The doses prescribed for superficial radiotherapy can be relatively high and may utilize standard or hypofractionated regimens (Mcgregor et al, 2015). Treatments can cause unwanted acute reactions such as skin erythema, skin depigmentation, and hair loss, and minimization of these toxicities can be at odds with therapeutic goals. It is clear, then, that accurate assessment of dose to the patient is essential. A prerequisite for accurate treatment is the acquisition of treatment planning parameters. The PDD data tables for various filter/kV and applicator combinations and the corresponding dose rates are essential for determining the optimal prescription depth in superficial radiotherapy.
Although kV radiotherapy delivery is relatively simple, the determination of PDDs, beam qualities, and beam output is not straightforward. The measurement of kV PDD curves involves particular considerations that increase uncertainty, especially near the surface. In scenarios for which it is not possible to directly measure the PDDs, Ma et al (Ma et al, 2001) recommended interpolating from the published data in BJR Supplement 25 (Radiology, B. I. O. & Biology, I. O. P. A. E. I. M. A (1996)). The BJR 25 report includes superficial x-ray PDD data for HVLs ranging from 1-8 mm Al, for source-to-surface distances (SSDs) from 10-30 cm, and for equivalent circular field diameters from 0-20 cm. The BJR 25 PDDs represent composite data from eight UK radiotherapy centers using cylindrical or parallel-plate ionization chambers for measurements in water or water-equivalent phantoms from 0-10 cm depth.
A published survey (Palmer et al, 2016) indicated that about 57% of radiotherapy centers in the UK are using data from BJR 25 for treatment planning of kV radiotherapy. It is therefore valuable to reassess and validate the accuracy and reliability of the PDD data in BJR supplement 25 utilizing contemporary dosimeters and equipment. There exist published studies that reported clinically relevant deviations from the BJR 25 PDD data. Johnstone et al (Johnstone et al, 2015) used IBA FC65-G and IBA CC13 ionization chambers to measure kV PDDs and found differences from −14% to +15.7 relative to the BJR 25 data. Aspradakis et al (Aspradakis and Zucchetti, 2015) used PTW TW30013 and Wellhöfer/IBA IC-15 ionization chambers in both water and plastic water to measure PDDs and found that above 100 kVp the agreement with published BJR 25 data was within ±5%, while at lower potentials the differences were greater than 10%.
The key limitations in the aforementioned studies were that the PDDs were measured using either ionization chambers with relatively large volumes or waterequivalent, rather than water, phantoms. In this study automated PDD scans were performed in a water phantom utilizing four contemporary detectors, and the measurements were compared to published BJR 25 data. Additional measurements were performed to address vendor acceptance tests, timer reproducibility, linearity and end-effect tests, half-value layer (HVL) determination, inverse square law verification, headleakage determination, and the beam output calibration.

Methods and materials
The Xstrahl 150 x-ray system The superficial treatment unit used in this study was an Xstrahl 150 x-ray system (Gulmay Medical Inc., Surrey, United Kingdom), shown in figure 1(a). The unit is equipped with an x-ray generator of 3 kW maximum power capable of producing x-rays from 10-150 kVp and tube currents up to 30 mA. The unit is available as either a flooror ceiling-mounted system and offers a range of superficial treatment options for radiation oncology and dermatology departments. The tungsten (W) target angled at 30°has a focal spot size of 7.5 mm. The tube has inherent filtration of 0.8 ± 0.1 mm of beryllium (Be) and minimum and maximum HVLs of 0.2 mm Al and 0.5 mm Cu, respectively. Six of the nine available filter and tube potential settings were commissioned for clinical use. A built-in system interlock necessitates that one filter, which determines kVp/mA tube setting, is properly inserted in the gantry while the remaining filters are docked in a wall-mounted storage unit. The properties of each clinically used filter corresponding to fixed kVp/mA tube setting, added filtration and nominal HVL, are presented in table 1.
The unit has ten open-ended applicators made of steel, copper and Perspex: nine circular applicators ranging from 1.5 cm to 18 cm in diameter, and one 20 × 20 cm 2 square applicator. The focal spot distance (FSD) from the source to the applicator opening varies with applicator but is limited to 15 cm, 25 cm, 30 cm, or 50 cm; see table 2. All six clinically used filters, ID #s 3-7 and 9, and all ten applicators, IDs A-J, were used in this study. For all PDD measurements, the applicator openings were set at the water surface to mimic the standard clinical setup in which the applicator is flush against the patient's skin.  Figure 1(b) HVL measurement setup: ionization chamber at a focal spot distance of 75 cm from the x-ray source with gantry rotated 180°to facilitate insertion of additional filtration sheets. Figure 1(c) PDD measurement setup. The opening of each applicator was in contact with the water surface and the detector was placed in the center of the opening. Mechanical and safety tests The performed system's checks included tube stand movements and brakes functionality, tube roll, pitch and yaw rotations and locking capability, smooth rotation and locking of applicators in the gantry, wallmounted storage unit and gantry filter docking as well as generator water coolant circulation and cooling fans spinning. Furthermore, power on key, x-ray on button, x-ray off button, emergency beam off switch, controlled area light illumination and audible x-ray on warning functionality was inspected. Lastly, the functionality of built-in system interlocks was tested, i.e., treatment room door interlock, filter-in-use and docked filters interlocks and timer interlock.
X-ray tube leakage measurements Head leakage measurements were made using a Fluke 451P Survey Meter (Fluke Corporation, Everett, Washington). Circular applicator A, i.e., the smallest applicator with 1.5 cm diameter opening, and filter 9 corresponding to the highest 150 kVp and 10 mAs tube setting, was used. The survey meter was placed at a fixed distance 1 m away and at 90°relative to the machine focal spot. The measurements were recorded for the gantry oriented at angles 0°and 180°, i.e., perpendicular to beam direction, and opposite to beam direction for the gantry angle at 270°.

Machine timer end effect
The end effect, Δt, is generally defined as the amount of beam delivery time not accounted for by a machine timing mechanism. It represents the time difference between the start of timing and the moment when the desired mA and kVp is achieved (Ma et al, 2001). Even a relatively small end effect (0.5-3 s) may lead to significant dosimetric errors due to the high dose rates and consequent short treatment times (1 min or less). The end effect for an x-ray unit can be determined using the graphical extrapolation method (Attix, 1986). The graphical solution of zero exposure, i.e., the intercept of the regression line on the time axis, yields the end effect. This method was used to determine the end effect for the entire range of clinical tube voltages for two different applicator sets using a PTW 31010 ionization chamber.

Timer reproducibility and linearity
The timer reproducibility and linearity tests were performed via in-air measurements using applicator F and a PTW 31010 ionization chamber placed at 15 cm distance from the focal spot. The reproducibility R was tested via a series of five consecutive 1-minute measurements at 50 kVp and 150 kVp tube potentials using equation where s was the standard deviation and M was the mean value of the charge collected in five measurements. The tolerance for R was set to ±0.1%. The reproducibility test was repeated by stopping and resuming the beam during 1-minute exposures at 150 kVp. The effect was quantified using where R i was the reproducibility with interruptions, M int was the mean value of the collected charge for five consecutive 1-minute measurements each with one interruption, and M was the mean value of the collected charge for five consecutive 1-minute measurements without any interruption. The linearity test, performed for filters 3 and 9, included measurements of 0.5-, 3-, and 5-minute beam-on times. The linearity was calculated using the following expression: here M 3 was the mean charge collected for a 3-minute exposure time, M t was the mean charge collected for an exposure time of t . min The tolerance for machine timer linearity L was set to ±1%.
The output constancy, for filters 3 and 9, was tested at gantry angles at 0°, 135°, and 225°for 1-minute beam-on times. The average measurements for gantry positions at 135°and 225°were compared to gantry 0°v ia equation where OC was the relative output difference, M G was the mean value of the charge collected during four 1-minute measurements at 135°and 225°gantry angles, and M 0 was the mean value of the charge collected during four consecutive 1-minute measurements at gantry angle 0°. The tolerance for the output constancy OC was set to ±1%.

Inverse square law verification
For reference clinical dosimetry, the calibration point for each applicator was set at the center of the circular opening, i.e., at the applicator central axis at FSD listed in table 2. However, it was impossible to measure dose rate at this exact location for every setup due to the intrinsic dimensions of ionization chamber and the small applicator opening. In such situations the chamber was placed just downstream of the opening so that the chamber touched the applicator's end, and an inverse-square correction was applied to determine the dose rate at midplane of the opening. Specifically, the measurement point and the intended calibration point differ by half the ionization chamber 6.9 mm outer diameter. The inverse square verification measurements were made using a PTW 31010 ionization chamber and a PTW Unidose E electrometer. Applicators A, C, and G and filters 3 and 9 were used in the measurements. The inverse square law was verified by taking two ionization chamber measurements. The first measurement position was at a point of contact with the applicator opening while the second one was at a point shifted 10 cm downstream along the applicator central axis. The measurements were compared with the expected values calculated using the inverse square law.

HVL measurements
Central-axis HVL measurements were acquired using a narrow-beam geometry following the recommendations of the AAPM TG-61 protocol (Ma et al, 2001). The setup utilized an in-house ionization chamber mount which provided minimal scatter of the primary beam while the chamber was placed at a 75 cm FSD. As shown in figure 1(b), the gantry was rotated 180°to facilitate placement of aluminum (Al) and copper (Cu) sheets atop the applicator opening. Applicator E, with circular opening of 4 cm diameter, and an ADCLcalibrated PTW 31010 ionization chamber were used for the measurements. The measured HVLs were compared to the nominal vendor-provided values. In addition, the nominal and measured HVLs were compared to values obtained from the SpekCalc and SpekPy software (Poludniowski and Evans, 2007, Poludniowski, 2007, Poludniowski et al, 2021, Bujila et al, 2020, Healy and Hill, 2022.

Output dose rate calibration measurements
The in-air method of the AAPM TG-61 protocol (Ma et al, 2001) was used to determine the absorbed dose rate to water at the water surface, with the applicatorspecific FSD listed in  is the dose rate to water at water surface; M raw is the uncorrected in-air ionization chamber charge reading; P , TP P , ion P , pol and P elec are the temperature and pressure, ion recombination, polarity effect, and electrometer correction factors, respectively; N k is the air-kerma calibration coefficient for a given beam quality; B w is the backscatter factor; P stem air , is the chamber stem correction factor, which was taken as 1; is the water-to-air ratio of the mean mass energy absorption coefficients averaged over the incident photon spectrum, for an inair measurement; and ISF is the inverse square factor needed if the measurement point is not at the calibration point, i.e., not at the applicator opening which indeed was the case here. Temperature and pressure were measured using a CNMC traceable digital thermometer and a Druck DPI 705 digital barometer. Output measurements were made for six filters and ten applicators, resulting in 60 clinical techniques. The dose rate for each filter and applicator combination was reported as dose rate in air D air  in Gy min −1 , i.e., the backscatter factor B w was factored out from D . w  The reason for this is the requisite to calculate treatment beam-on time for clinical setups different than the reference calibration conditions. In clinical practice, a beam-on time, t, needed to deliver the prescribed therapeutic dose, D , R X is calculated using the following formula: is the calibrated dose rate in air for the reference FSD , ref gap is the separation between applicator and the patient if the applicator is not flush )is the backscatter factor for a given treatment FSD and field diameter d , )is the percent depth dose at the prescription depth z for the reference FSD ref measured during commissioning, and F M is the Mayneord's factor representing the ratio of the inverse square component of PDDs from the reference FSD ref to treatment FSD also as needed when gap is nonzero.

Percent depth dose measurements
The PDD water tank measurements were carried out using an IBA Blue Phantom 2 (IBA Dosimetry, Schwarzenbruck, Germany) for combinations involving six filters and ten applicators, i.e., for 60 different clinical x-ray beams. The opening of each applicator was positioned at the water surface and the detectors were placed in the center of the applicator, see figure 1(c). The detectors used for PDD measurements were a synthetic diamond detector (microDiamond 60019, PTW, Freiburg, Germany), a diode detector (microSilicon diode 60023, PTW, Freiburg, Germany), another diode detector (EDGE, Sun Nuclear Corp., Melbourne, Florida) and an ultra-smallvolume ionization chamber (PinPoint 3D 31016, PTW, Freiburg, Germany). The ionization chamber was positioned with its axis perpendicular to the beam central axis, while the diamond and diode detectors were positioned with their axes parallel to the beam central axis. Applicator-specific profile scans were used to determine the central axis of each field. Zero depth for the ionization chamber was established as the central chamber axis coincided with the water surface. Two assumptions were implicit here, first, the effective point of measurement coincided with the ionization chamber's central axis, and second, the depth doses were directly proportional to the ionization readings without applying any corrections. A bias voltage of −300 V was used for the ionization chamber scans. For the solid state detectors, zero depth was established as the line on the housing, indicating the reference point, was aligned with the water surface, and zero voltage bias was used for the scans. The detectors were pre-irradiated with vendorrecommended doses to stabilize the detector response. The PDDs were measured from 20 cm depth to water surface along the beams' central axes, with the detectors moved upward at constant speed of 0.3 cm s −1 to minimize disturbances to the water surface. The detector readings were recorded in approximately 0.1 mm intervals, i.e., for each PDD scan the raw data had over 2000 data points. For efficiency and to minimize setup uncertainty, all PDDs for an applicator were measured in sequence before changing the setup for the next applicator.
The ISO methodology(ISO, 2011) provides a general basis for determining uncertainties associated with experimental measurements. A few exemplary uncertainty budget considerations relevant for kV x-rays can be found in the literature (Andreo et al, 2000, Ma et al, 2001, Hill et al, 2009, Gronberg et al, 2020. The contributing sources of uncertainty for PDD measurements in this study included setup repeatability, applicators FSD specifications and air gap (if necessary), detectors placement at the water surface, motors positional accuracy during scanning, detectors directional response in water, variation in the mass energy absorption coefficient ratios with beam size and depth in medium, machine output fluctuations during long exposure times necessary for data acquisition, and drift of the measuring equipment.
After obtaining the scanned data, the measured PDD points were fitted in Matlab(Mathworks, 2021) using the least-squares spline approximation by minimizing expression where w i represents the weight factors with default values equal to 1, y i represents the measured data points and f x i ( ) is the spline function of polynomial order k with the knot sequence for which for all i. A 9th, 11th and 13th order polynomial were used for data fitting. The goodness of fit was judged by comparing the adjusted R 2 values with a higher value closer to 1 signifying better fit. The spline technique was selected for fitting since it facilitated consistent data fitting for all detectors.
The measured PDD data were compared to the BJR 25 data by simply computing the differences, Δ PDD , which were reported as percentages because PDD values are intrinsically relative numbers. The relative difference, or the ratio, of published and measured depth doses is not an adequate metric since the ratio of two small numbers with a small difference yields a large relative difference that creates a false perception of a large dose discrepancy.

Results
Mechanical and safety tests The x-ray unit's mechanical and safety features were all functional.

X-ray tube leakage
The survey meter readings of the x-ray treatment head leakage for gantry angles 0°, 180°, and 270°were 0.41, 0.03, and 0.19 mGy h −1 , respectively. The measured values were much less than 1/1000 or 0.1% of the primary beam output, 262 Gy h −1 , and thus in accord with the International Electrotechnical Commission IEC 60601-2-8 standard (IEC, 2010) for medical electrical equipment, which limit is 1 mGy h −1 at 1 m distance.

Machine timer end effect
The intrinsic operation of the Xstrahl 150 timer is such that for a given kVp and mA setting, the kV first rises to 90% of the required value before the mA begins ramping up to its required value. Once the mA rampup is 50% complete, the treatment timer starts. This approach enables the kV and mA to stabilize adequately and deliver a reproducible dose. The largest measured timer end effect, Δt, was 0.19 s for filter 3, which has the lowest tube potential of 50 kVp, with a dosimetric impact of less than 1% for an exposure time of 20 s. For filters 4 and 5, Δt was 0.01 s, and for filters 6, 7 and 9, Δt was 0.06 s. In general, the measured end effects for all filters were determined to be negligible for clinically relevant beam-on times which are on the order of minutes.

Timer reproducibility and linearity
The measured timer reproducibility, R, defined by equation (1), was 0.1% and 0.03% for filters 3 and 9, respectively. The reproducibility with intentional interruptions, R , i as defined by equation (2), was determined to be 0.04% for filter 9. This indicated that clinical treatment interruptions would have negligible dosimetric effect. The measured timer linearity, L, defined by equation (3), was 0.2% and 0% for filters 3 and 9, respectively. The machine output constancy, OC, defined by equation (4), was less than 0.2% over the range of gantry angles tested, demonstrating that gantry orientation has negligible dosimetric impact on clinical treatments.

Inverse square law verification
In general, for two measurement points 10 cm apart on the beam central axis, the relative difference between measured and calculated inverse square factors, Δ ISF , decreased with increasing applicator size. For instance, for applicator A, Δ ISF was 1.8% and 1.3% for filters 9 and 3, respectively. For applicator C, Δ ISF was 1.2% for both filters. For applicator G, Δ ISF decreased to 0.2% and 0.7% for filters 9 and 3, respectively. Notably, a smaller applicator opening combined with a smaller FSD contributed to relatively larger in-scattering for applicator A compared to applicator G. Since the inverse square law for photons is universally valid, the main contribution to the observed variances was due to relative difference of scattered electrons from the applicator walls. Consequently, the ISF component for smaller collimators has an intrinsically larger relative systematic error relative to the larger collimators. This confirms the recommendation to use the smallest applicator for inverse square law validation, since a 4% difference was reported for a 2.5 cm diameter applicator at 10 cm FSD (Aukett et al, 2005).

HVL measurements
The measured HVLs were compared to the nominal vendor-stated values as well as to the HVLs calculated by SpekCalc and SpekPy software (Poludniowski, 2007, Poludniowski and Evans, 2007, Bujila et al, 2020, Healy and Hill, 2022, Poludniowski et al, 2021; see table 3. For the six clinically used filters, the SpekCalc and SpekPy HVLs were within 0.1 mm relative to the measured values. The agreement between all measured and nominal HVLs was within 10%. The observed HVL differences are common and acceptable and typically translate to less than 1.5% dosimetric change for tube potentials above 80 kVp. This is related to the choice of the beam quality points offered by the Accredited Dosimetry Calibration Laboratories (ADCL) for an ionization chamber calibration in terms of air kerma. Since the ADCL beam qualities are not only stratified based on the HVL values (mm Al or Cu) but are also based on the inherent filtration (mm Be), the additional filtration (mm Al or Cu), and the tube voltage (kVp), the choice of calibration beam quality may not be straightforward. In the event the ADCL calibration conditions do not match the user's machine specifications a compromise of selecting the 'nearest match' point is one option, alternatively, an interpolation between two nearest neighbor calibration points can be utilized.

Output dose rate calibration
The in-air calibration was performed in accordance with the AAPM TG-61 protocol with the estimated combined standard uncertainty (1σ) of ±4.7%. Ten applicators and six filters yielded 60 clinical beam combinations whose dose rates are listed in table 4. As expected, the measured dose rates at a given FSD gradually increased with an increase of applicator size due to increased backscatter. At the same time, the measured dose rates were considerably reduced for applicators G to J owing to the increased FSDs and resulting inverse square law losses.

Percent depth dose
The four detectors used for PDD measurements were a PTW microDiamond 60019, a PTW microSilicon diode 60023, a Sun Nuclear EDGE diode and a spherical PTW ionization chamber PinPoint 3D 31016. Six filter and ten applicator combinations, i.e.,  1,2,3,4,5,6,7,8,9 and 10 cm as tabulated in the BJR 25 report. A total of 101 discrete points, corresponding to 0-10 cm depth in 1 mm increments, were used for the cross comparison between four detectors utilized for the PDD measurements. In line with general textbook knowledge, it was observed that the measured PDDs increased with an increase in applicator size or beam energy. The relative uncertainty of the performed percent depth dose measurements was estimated to be ±1.5% at the one standard deviation level; see table 5.
For the medium-energy superficial x-rays, i.e., 100 kVp and 120 kVp, corresponding to HVLs of 3.0 mm, 4.0 mm and 5.0 mm Al in this study, on average all measured data for the solid state detectors were within [−3.7, +1.4]% range relative to the published BJR 25 data. The largest outlier −3.7% was again observed for applicator I (18 cm diameter) for which the comparison was based on interpolated BJR 25 values. For all applicators smaller than applicator I the average agreement with BJR 25 data improved across the board, for all PDDs measured with solid state detectors. The Pin-Point ionization chamber data was within [−1.3, +3.7]% range relative to the published BJR 25 data. However, this only applies to applicators A to H, while for the two largest applicators I and J, the recorded data for PinPoint detector became very noisy rendering it unsuitable for obtaining reliable PDD curves. The average difference, Δ PDD , compared to twelve BJR 25 PDD points, is plotted for applicators E and D in figure 3. The average differences for applicator E and filter 5 in figure 3(a) were (−0.6 ± 1.2)%, (−0.9 ± 1.3)%, (−1.7 ± 1.0)% and (1.2 ± 1.2)% for micro-Diamond, microSilicon, Edge and PinPoint 3D detectors, respectively. Likewise, for applicator D and filter 6 in figure 3(b) the results were (0.7 ± 1.4)%, (0.1 ± 1.4)%, (−0.6 ± 0.7)% and (2.8 ± 1.7)% for micro-Diamond, microSilicon, Edge and PinPoint 3D detectors, respectively.
For the high-energy superficial x-rays, i.e., the low-energy orthovoltage x-rays at 150 kVp, corresponding to HVL of 0.5 mm Cu, there are no published BJR 25 data for circular applicators. The only comparison was possible for the largest 20 × 20 cm 2 square applicator J at 50 cm FSD. In absence of BJR 25 data, figure 4(a) shows the intercomparison of measured PDDs for applicator C and filter 9 relative to micro-Diamond detector. The average differences, Δ PDD , were (−0.2 ± 0.4)%, (−1.6 ± 0.7)% and (1.2 ± 0.6)% for microSilicon, Edge and PinPoint 3D detectors, respectively. Figure 4(b) shows a comparison of composite PDD curve. i.e., an average PDD of all three solid state detectors (microDiamond, microSilicon and Edge) for applicator J and filter 9, relative to Detector directional response in water 0.5 1.0 Drift of the measuring equipment 0.5 0.5 X-ray unit output fluctuations 0.5 0.5 Combined standard uncertainty (1σ) 1.5 1.5 twenty-one BRJ 25 data points available for this energy. The average difference, Δ PDD , for the composite data was (−0.7 ± 1.0)%. The PinPoint 3D data was not considered here as the scan was too noisy to obtain clinically acceptable PDD curve. The composite PDD values measured by three solid state detectors are summarized in table 6. The data shown are for filter 7, HVL 5.0 mm, 120 kVp, for all ten clinically used applicators, A-J. The appendix includes tabulated composite PDD data for all combinations of available filters and applicators, i.e., for 60 different clinical x-ray beams. The composite PDD values do not contain the PinPoint measurements since the ionization chamber data was not reliable for the two largest applicators I and J and because the ionization chamber scans exhibited small but systemic overestimation of measured PDD curves compared to solid state detectors.

Discussion
The aim of this study was to present commissioning data from an Xstrahl 150 radiotherapy system. Extra effort was expended to acquire PDDs in water for ten applicator and six filter combinations, i.e., sixty beams of unique quality, using four different detectors. Altogether 240 PDD curves were acquired using a high-accuracy motorized scanning water phantom. For x-ray energies in kV range, liquid water is the only medium recommended for reference dosimetry as well as for percent depth dose measurements (Ma et al, 2001, Aukett et al, 2005. This recommendation is notably supported by findings reported by Hill et al (Hill et al, 2010a) which showed the PDD differences compared to liquid water were −21.7% for Plastic Water and +17.6% for polystyrene, for 50 kVp photons collimated by an 8 cm diameter applicator. In general, the water equivalency of plastic or solid water phantoms is energy dependent. As the energy of kV beams increase, the dosimetric variations between water and water-equivalent phantoms decrease. Most, but not all, commercially available solid phantoms are water-equivalent within ±2% in kV range (Hill et al, 2010a).
For absolute and relative measurements in the kV range, reference dosimetry protocols (Ma et al, 2001, Aukett et al, 2005, Klevenhagen et al, 1996 recommend air-filled cylindrical ionization chambers. Welldesigned cylindrical ionization chambers are considered the gold standard due to their nearly constant energy response for tube potentials between 40-300 kV. For relative dosimetry measurements such as PDD curves and relative output factors, small-volume cylindrical or plane-parallel ionization chambers are widely used. For instance, Sheu et al (Sheu et al, 2015) Figure 4. The comparison between measured PDD and BJR 25 data for high-energy superficial x-rays (150 kVp), i.e., low-energy orthovoltage x-rays. (a) PDD curve comparison relative to microDiamond detector for filter 9 (HVL 0.5 mm Cu, 150 kVp), applicator C (2.5 cm diameter, 15 cm FSD). No BJR 25 data for this field size and energy. (b) Composite PDD curve of three solid state detectors (microDiamond, microSilicon and Edge) for filter 9 (HVL 0.5 mm Cu, 150 kVp), applicator J (20 × 20 cm 2 square, 50 cm FSD). used a PTW plane-parallel chamber N23342 and a solid water phantom to generate the PDDs for a Sensus SRT-100 x-ray unit. Johnstone et al (Johnstone et al, 2015) used IBA FC65-G and IBA CC13 ionization chambers for PDD measurements in water, and Aspradakis et al (Aspradakis and Zucchetti, 2015) used a Wellhöfer/IBA IC-15 ionization chamber and a PTW Advanced Markus plane-parallel ionization chamber to determine PDDs both in water and waterequivalent plastic phantoms.
It is worth mentioning that ionization chambers are not ideal for measuring at depths close to the surface. Measurement depth is limited to no less than the outer radius of the chamber, due to the perturbation at the surface when the chamber is sitting partially out of the water and because of the difference in relative detector response in air and in water (Hill et al 2009).
As a consequence, ionization chamber PDDs are flattened out near the surface, resulting in overestimation of relative dose by several percent. In this study, the PDD scans utilizing an ultra-small size PinPoint 3D ionization chamber with a nominal sensitive volume of 0.016 cm 3 and radius of 1.45 mm agreed with the BJR 25 data within [−2.6, +3.7]% range for all filters and applicators A to H. However, due to the ultra-small sensitive volume for two largest applicators I and J at 30 and 50 cm FSD, respectively, the signal was too noisy for a reliable fit compared to the solid state detectors.
The utilization and accuracy of solid state detectors for kV x-ray beam dosimetry has been the subject of a limited number of investigations to date. The advantages of diamond and silicon detectors include very small active volumes, small angular dependence, and minimal energy dependence. The PTW microDiamond detector was proven suitable for PDD measurements across a wide range of kV x-ray energies (50-280 kVp) and field sizes (Damodar et al, 2018, Khan et al, 2020, Daniel et al, 2022, Butler et al, 2018. For all kV energies in this study, the intercomparison between microSilicon and Edge detectors relative to microDiamond was within [−0.6, +2.4]% and [−3.3, +0.7]% range, respectively. The composite PDD values, representing the average of measured data by all three solid state detectors are included in the appendix. In clinical practice, if a suitable detector is not available for relative dosimetry measurements then the BJR supplement 25 data should be used (Ma et al, 2001). There have been several studies that compared PDD measurements against BJR supplement 25 data. Johnstone et al (Johnstone et al, 2015) found differences of −6% to 6% for an 80 kVp beam and −14% to 15.7% for medium-beam energies (>100 kVp). However, these results are at odds with a study by Aspradakis et al (Aspradakis and Zucchetti, 2015) which found PDD differences within ±5% for medium-energy beams and differences of more than 10% for beam energies between 50 kVp and 100 kVp. These results support the conclusions of Hill et al (Hill et al, 2010b) stating that discrepancies between measured PDDs and BJR 25 data could be attributed to differences in machines, detectors, phantoms, and measurement methods across various institutions. The results in this study were better matched with the published BJR 25 data. For the lower (<100 kVp) and medium-energy (100 kVp) superficial beams the average agreement was within [−3.6,+0.4]% and [−3.7,+1.4]% range, respectively. For the high-energy superficial (lowenergy orthovoltage) x-rays at 150 kVp, the average difference for the largest 20 × 20 cm 2 collimator was (−0.7 ± 1.0)%. In summary it is very difficult to make direct study-to-study comparisons of PDD measurements, especially considering the BJR 25 data are an aggregate of measurements contributed by eight UK radiotherapy centers. As the BJR 25 report points out, direct kV measurements of clinical PDDs must be carried out if higher accuracy is needed.

Conclusions
This study presents commissioning data from an Xstrahl 150 x-ray system with a special focus on PDD measurements in water utilizing different detectors. The mechanical, safety, and dosimetric machine performance agreed favorably with specifications. Overall, the measured PDD curves exhibited excellent agreement with the BJR 25 data. The appendix provides tabulated PDD values with more datapoints than the BJR 25 for clinical intercomparison. We conclude that the BJR 25 PDD data still represent a reliable benchmark for measurements, as well as a valid alternative for clinics without resources to perform machine-specific relative dosimetry measurements.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).