Whole‐body dose equivalent including neutrons is similar for 6 MV and 15 MV IMRT, VMAT, and 3D conformal radiotherapy

Abstract Purpose This study investigates the difference in whole‐body dose equivalent between 6 and 15 MV image‐guided radiotherapy (IGRT) for the treatment of a rhabdomyosarcoma in the prostate. Methods A previously developed model for stray radiation of the primary beam was improved and used to calculate the photon dose and photon energy in the out‐of‐field region for a radiotherapy patient. The dose calculated by the treatment planning system was fused with the model‐calculated out‐of‐field dose, resulting in a whole‐body photon dose distribution. The peripheral neutron dose equivalent was calculated using an analytical model from the literature. A daily cone beam CT dose was added to the neutron and photon dose equivalents. The calculated 3D dose distributions were compared to independent measurements conducted with thermoluminescence dosimeters and an anthropomorphic phantom. The dose contributions from the IGRT treatments of three different techniques applied with two nominal X‐ray energies were compared using dose equivalent volume histograms (DEVHs). Results The calculated and measured out‐of‐field whole‐body dose equivalents for the IGRT treatments agreed within (9 ± 10) % (mean and type A SD). The neutron dose equivalent was a minor contribution to the total out‐of‐field dose up to 50 cm from the isocenter. Further from the isocenter, head leakage was dominating inside the patient body, whereas the neutron dose equivalent contribution was important close to the surface. There were small differences between the whole‐body DEVHs of the 6 and 15 MV treatments applied with the same technique, although the single scatter contributions showed large differences. Independent of the beam energy, the out‐of‐field dose of the volumetric‐modulated arc therapy (VMAT) treatment was significantly lower than the dynamic intensity‐modulated radiation therapy (IMRT) treatment. Conclusion The calculated whole‐body dose helped to understand the importance of the dose contributions in different areas of the patient. Regarding radiation protection of the patient for IGRT treatments, the choice of beam energy is not important, whereas the treatment technique has a large influence on the out‐of‐field dose. If the patient is treated with intensity‐modulated beams, VMAT should be used instead of dynamic IMRT in terms of radiation protection of the patient. In general, the developed models for photon and neutron dose equivalent calculation can be used for any patient geometry, tumor location, and linear accelerator.

the developed models for photon and neutron dose equivalent calculation can be used for any patient geometry, tumor location, and linear accelerator. Ten percent of these second tumors are induced by the radiation treatment the patient received. 1 Most second cancers occur at the peripheral region where the dose is greater than 3.0 Gy. 1 However, Diallo et al. 2 identified a peak frequency in second malignant neoplasm (including spontaneous cancers) for volumes that received a dose smaller than 2.5 Gy. In external radiation beam therapy, the treated volume receives a high dose while the remaining body is exposed to an unwanted low dose of radiation. Usually, the dose is calculated around the target volume and the out-of-field dose is not accurately considered, if at all. 3 Therefore, whole-body dose distributions are needed for accurate cancer risk estimates and for optimizing treatment plans by minimizing the cancer risk.
Another motivation for whole-body dose calculation is the radiation protection of the fetus. Negative effects for a fetus can be substantially minimized if the dose to it is reduced to 100 mGy. 4 However, practical models to estimate the fetal exposure for intensity-modulated treatments of pregnant patients do not yet exist. 1 Takam et al. 5 presented the current status of out-of-field neutron and photon leakage dose in radiotherapy and the associated risk for the patient. Most of their results were based on patient treatments which occurred decades ago. Therefore, studies including novel treatment machines and techniques are urgently needed. 5 For the same technique applied with different nominal X-ray energies, the target coverage, conformity, and homogeneity of the treatments are similar. 6 The choice of nominal X-ray energy should be based on normal tissue complication probability and on radiation protection issues of the patient. Many studies investigated the difference in the peripheral dose between high (≥10 MV) and low nominal X-ray energy (<10 MV). [7][8][9][10][11] A Monte Carlo (MC) study conducted by Kry et al., 7 showed a similar photon out-of-field dose for 6 MV compared to 18 MV intensity-modulated radiation therapy (IMRT) treatments. For the nine organ locations investigated, the simulated neutron doses were typically much lower than the corresponding photon dose. Nevertheless, they warranted an improved neutron dosimetry in order to achieve superior estimates. Ruben et al. 8 measured the components of the out-of-field photon dose for 6 and 18 MV treatments. The neutron dose contribution was obtained from published data. They reported that X-ray energy does not affect the total photon scatter for the same treatment technique.
However, the additional neutron dose for 18 MV may have increased total body cancer risk compared to 6 MV IMRT treatments. However, they were not able to draw a firm conclusion. Hälg et al. 12  With increasing number of treatments using volumetric-modulated arc therapy (VMAT) and similar dose distributions of VMAT compared to IMRT treatments, the question arises about the difference in the out-of-field dose between the two techniques. To our knowledge, there is no study published comparing the peripheral dose (including neutrons) of high-energy VMAT treatments with IMRT treatments. In the current study, the difference between the dose equivalent of 6 and 15 MV treatments was examined. It is per se not clear that for 15 MV X-ray nominal beam energy the out-of-field dose will be smaller in comparison to 18 MV because of the reduced photoneutron production. Compared to photons, neutrons are a minor part of the total out-offield dose equivalent. 1 Howell et al. 9 reported a higher effective dose for 15 MV compared to 18 MV 3D-conformal radiation therapy (3DCRT) treatments.
Also the use of X-ray imaging modalities can give a substantial dose to the patient. 13 The choice of treatment technique and indication determines the image modality and therefore, the additional amount of dose to the patient. For patient positioning, the imaging dose is justified by the reduction of the margins around the target. A smaller planning target volume will lead to a sparing of the organs at risk during irradiation. If image-guided radiotherapy (IGRT) is used, the patient receives an additional dose from X-ray imaging. By including all contributions of the whole-body dose for an IGRT treatment, a better understanding in radiation protection of the patient can be achieved.
In the current study, we investigated the whole-body dose equivalent for 6 and 15 MV IGRT treatments of a rhabdomyosarcoma in the prostate applied with three different techniques (3DCRT, IMRT, and VMAT). The analytically calculated dose distributions were verified with whole-body dose measurements. The results from the calculation were used to identify differences in the whole-body dose between the investigated treatments. HAURI AND SCHNEIDER | 57 2 | ME TH ODS AND MATERIALS In this manuscript, the indexes m, c, and s describe quantities which were derived either from measurements, calculations, or MC simulations, respectively. The abbreviation n stands for neutrons and γ for photons.

2.
A | Whole-body photon and neutron dose calculation Dose calculations were performed for an anthropomorphic phantom (Alderson-Rando, RSD Radiology Support Devices, Long Beach, CA, USA) using a whole-body grid with a voxel dimension of 0.2 × 0.2 × 0.5 cm 3 (see Fig. 1). However, the radiotherapy photon and neutron dose models used in this work are generally applicable to any 3D-patient data set.

2.A.1 | Photon imaging dose
For each treatment fraction, the patient was assumed to be positioned with a full trajectory kV cone beam CT (CBCT) of the pelvis.
Hence, we assigned a relatively high imaging dose for the IGRT treatments. The mean absorbed CT dose per Alderson slice was calculated using the average of the thermoluminescence detector (TLD) dose measurements in the corresponding slice. The TLD measurements of the full trajectory pelvis CBCT are reported in Hauri et al. 14 The absorbed dose per voxel of a CBCT scan was calculated by interpolating the average CBCT dose per Alderson slab along the medial patient axis (MPAX). Hence, the dose was the same for all voxels in a transversal dose-grid slice [see Fig. 1(a)]. According to Schneider et al., 15 the dose of a full rotation CBCT is in a first approximation homogeneous in a transversal slice.

2.A.2 | Therapy dose photons and neutrons
A previously developed photon stray dose model for static and intensity-modulated 6 MV treatments 16 was improved and adapted for 15 MV (see Appendix 1). The algorithm calculated the wholebody out-of-field dose of the coplanar treatments starting 4 cm longitudinal from the treatment volume (∼3 cm from the field edge).
According to Kry et al., 1 the differences between treatment planning system (TPS) and measurements exceed 30% of the local dose as close as 3 cm from the field edge, and differences increase by orders of magnitude at greater distances. At 4 cm longitudinal from the treatment volume, the dose of the TPS (Varian Eclipse, AAA-algorithm version 13.6.23) was fused with the model-calculated 3D outof-field dose resulting in a whole-body photon dose [see Fig. 1 The peripheral neutron dose was calculated using an analytical model from the literature. 17 This model was commissioned for True-Beam linear accelerators (linacs) (Varian Medical Systems, Palo Alto, CA, USA), operated at 15 MV [see Fig. 1(c)]. The model assumed a point source of neutrons in the X-ray producing target to predict the neutron fluence in the Alderson phantom. The fluence was converted to a neutron dose equivalent according to Sibert and Schumacher. 18 Only the peripheral neutron dose was calculated since inside the primary X-ray beam, the dose from neutrons can be neglected when compared to the photon dose. 12

2.B | Whole-body TLD measurements
The whole-body photon and peripheral neutron dose measurements served as verification of the photon stray dose and neutron dose calculation.
LiF TLD-chips (4.5 mm diameter, 0.6 mm thickness, Harshaw, Thermo Fisher Scientific, Waltham, MA, USA) were used to measure the in-and out-of-field dose of external therapy. For TLD100 (LiF: Mg,Ti) and TLD100H (LiF:Mg,Cu,P), the same thermal treatment, calibration procedure, and readout were used as described by Hauri and Schneider. 21 The thermal treatment, calibration procedure, and F I G . 1. The whole-body dose equivalent for the 15 MV IMRT treatment with a daily CBCT. The dose equivalent is shown for (a) 23 times a CBCT, (b) photon scatter radiation fused with the treatment planning system calculation, (c) neutrons and (d) the summation of (a)-(c). The Fractionation scheme is presented in Table 1. Furthermore, the outline of the rhabdomyosarcoma in the prostate can be seen. readout for TLD600/700 (LiF:Mg,Ti) and TLD700H (LiF:Mg,Cu,P) were the same as applied to TLD100 and TLD100H, respectively. TLD100 contains the natural abundance of 6 Li and 7 Li, while TLD600 contains primarily 6 Li. According to Schwahofer et al.,22 TLD600 and TLD100 show the same response to photon radiation since the number of neutrons in Li does not affect the energy bands of the TLD crystal. For the same reason, it was assumed in this manuscript that there is no difference in the response with photon radiation energy of TLD100H and TLD700H.
For each TLD, an individual photon dose-to-water calibration factor (in mGy/count) was determined using 6 MV nominal X-ray energy applied with a TrueBeam linac. All absolute photon dose measurements were correlated to a Farmer Chamber 30013 (PTW, Freiburg, Germany). The irradiations and detector readouts were performed according to a strict protocol 21 to ensure consistency of the measurements.

2.B.1 | Treatment intention, planning, and irradiation
The target volume of this study was a rhabdomyosarcoma in the prostate of an adolescent patient. The planning CT of the anthropomorphic Alderson phantom as well as the contouring of the target volume and organs were performed at one hospital.
The treatment planning of the 6 and 15 MV 3DCRT (four field box), IMRT (five fields with dynamic multileaf collimator (MLC)), and VMAT (one arc) treatments was done using the Eclipse TPS. All treatments were planned by an experienced worker. The 3DCRT treatments included a sequential boost and the intensity-modulated treatments an integrated boost. The motivation regarding the treatment intention and the fractionation scheme, and a detailed description of the treatments and the strict planning guidelines can be found in Hälg et al. 20 The diameter of the pelvis CBCT was 46.5 cm in a transversal slice and a field-of-view of ±8.75 cm from the isocenter in the longitudinal direction. The CBCT protocol (version 2.5.28.0, half-fan type, full trajectory, 125 kV p , 1080 mAs) was given by the vendor.
Using a conventional linac equipped with an on-board imaging system (TrueBeam), the six treatments and the CBCT were irradiated onto the Alderson phantom, each time loaded with new TLDs. The phantom was positioned head first supine. For the 6 MV treatments and the CBCT scan, each measurement location in and on the phantom was equipped with a TLD100H stacked on top of a TLD100.
For the 15 MV treatments, each measurement location was loaded with a TLD700H stacked on top of a TLD600. Confetti (made out of normal paper) were placed between all (TLD600, TLD700H)-pairs to avoid an α-particle contribution to the TLD700H signal originating from the 6 Li(n, α) capture. The measurement locations were distributed in the Alderson phantom according to Hälg et al. 20 Additionally, for the 15 MV treatments, the out-of-field photon dose of the skin was measured along a line from the pelvis to the nose of the phantom in steps of 10 cm. For this, the TLD700H were loaded in empty pill casings made of PMMA simulating the thickness of the skin. All absolute photon dose measurements were correlated to a Farmer Chamber 30013 since the chamber was used to determine the in-field TLD-calibration dose.

2.B.2 | Photon dose and mean photon energy of the CBCT and 6 MV treatments
The in-and out-of-field photon dose of the 6 MV treatments and the CBCT were measured separately using a combination of TLD100 and TLD100H chips. The two TLD types show a difference in response with photon radiation energy. 22,23 If calibrated with 6 MV nominal beam energy, TLD100 show an over-and TLD100H an under-response toward lower energy (down to 0.1 MeV). 21 By building the ratio of the TLD100 and the TLD100H measured doses, the mean photon energy can be determined. Using the photon energy at a specific measurement location, the TLD correction factors for the response with radiation energy can be determined. A comprehensive description of photon dose and the mean energy measurements for the CBCT and the 6 MV treatments is given in Hauri and Schneider. 21 Furthermore, a detailed description of the uncertainties in photon dose and mean photon energy is presented.

2.B.3 | Photon dose of the 15 MV treatments
The whole-body photon dose of the 15 MV treatments was determined for 189 locations in the Alderson phantom. The detected outof-field photon dose by each TLD700H was corrected for the response with photon radiation energy. The individual correction factors were estimated by using the photon scatter contribution at each measurement location in the phantom.
The total out-of-field photon dose consists mainly of three contributions: patient scatter, collimator scatter, and head leakage. 8,16,24 In the middle of a 30 × 30 × 30 cm 3 water-slab phantom (source surface distance = 85 cm), the mean energy of the three scatter contributions was measured using a combination of TLD700 and TLD700H. 21 For a 10 × 10 cm 2 field (defined by the MLC), the mean energy of patient scatter was measured at 15 cm distance to the field edge. For the same field size, the mean energy of collimator scatter was determined at 15 and 35 cm distance from the field edge. At the T A B L E 1 The total treatment dose, total MUs, and MUs per treatment Gy. The treatments were planned by an experienced worker according to a strict protocol. 20  same locations, the mean energy of head leakage was measured for closed jaws and MLC. The separation of a field measurement into the three scatter contributions is described by Hauri et al. 16 A previously developed 6 MV out-of-field dose model 16 was improved and adapted for 15 MV (see Appendix 1). Using the adapted model, the doses of patient scatter, collimator scatter, and head leakage were calculated for each measurement location l in the phantom. The final out-of-field mean energy E γ l;c was determined by,

Plan
with i = {patient scatter, collimator scatter, head leakage}. D γ i;l;c is the calculated dose at the measurement location l and E γ i;m ; is the mean energy of the scatter contribution i.
The calculated out-of-field mean energies were used to deter- As a consistency check, the TLD correction factors for the response with photon radiation energy were calculated for the 6 MV treatments (3DCRT, IMRT, and VMAT). In distinction to the 15 MV measurements, the photon energies for the 6 MV treatments were explicitly measured. 21 Using Eq. 1, the mean photon energy was calculated for each measurement location of the 6 MV treatments. The calculated and measured mean energies were converted to correction factors for the TLD100H response with photon radiation energy. The calculated and measured correction factors for the 6 MV treatments were compared to estimate the uncertainty of the TLD700H photon dose measurement.

2.B.4 | Neutron dose equivalent of the 15 MV treatments
With a combination of TLD600 and TLD700H, the whole-body neutron dose equivalent of the 15 MV treatments was determined.
TLD700H is not affected by neutrons in the energy range of interest. 25 TLD600 register photons and neutrons. Using the mean photon energy (Eq. 1), the neutron signal detected by TLD600 in the phantom was corrected for the photon contamination measured by TLD700H. The measured neutron signal of a TLD600 was transformed to neutron dose equivalent (including fast neutrons) with a depth dependent conversion factor. 17 Each TLD600-specific depth in the phantom was calculated by using a straight line connecting the X-ray producing target and the measurement location. The Alderson phantom was assumed to be a soft tissue-equivalent (ICRU), with the exception of the lungs. For the lungs, a mass density of 0.25 × ρ soft tissue was assumed (relative hydrogen content in lungs compared to soft tissue = 25% 26 ). For the 3DCRT and the IMRT treatments, the calculation of the depth in the phantom was straight-forward since there was no gantry rotation during the beam-on time. For the VMAT treatment, the control points of the one arc were grouped to six fields with different gantry angles and corresponding MUs per field. A more detailed description of the approximation of the VMAT plan by discrete fields can be found in Hauri et al. 16 Compared to the TLD600-registered signal from neutrons, the signal from photons is orders of magnitude higher in the target volume. 27 Therefore, the measurement of neutron dose equivalents was only possible outside the treatment volume.

| RESULTS
Unless otherwise stated, the mean and one standard deviation (σ) are presented. Consistent with the IAEA report, 28 type A stands for the measured σ and the type B for the estimated σ.

3.A.1 | Photons
The deviation between the calculated whole-body dose and the 183 point-dose measurements of the CBCT scan was 0% ± 14% (type A).
The measured mean photon energies of the three stray dose contributions can be seen in Table 2

| DISCUSSION
Averaged over all treatment techniques and nominal X-ray energies, the calculated whole-body dose equivalent agreed within (9 ± 10) % (type A) compared to the measured dose equivalent. This agreement was sufficient to determine differences in the whole-body dose between the investigated treatments.

4.A | Photon dose
The small deviation between the predicted and measured CBCT dose justified the presented method to calculate the whole-body imaging dose.
Close to the target volume, the CBCT dose was a substantial contribution to the out-of-field dose resulting from a treatment (Fig. 5). In the field-of-view of the CBCT, the dose was almost constant. 29 Furthermore, the field-of-view extended around 4 cm over the border of the target volume. In this area, the dose caused by scatter radiation of the primary beam dropped rapidly with increasing distance to the target volume. The CBCT dose decreased exponentially with increasing distance to the field-of-view. Hence, a smaller field-of-view is beneficial regarding radiation protection of the patient. The contribution of imaging to the total dose equivalent was similar for all investigated treatments since the number of sessions and the fraction doses were comparable for all treatments (see Table 1) Compared to the primary beam, the X-ray spectrum in the peripheral region is softer such that an increase in organ-specific relative biological effectiveness (RBE) for carcinogenesis is expected. 34 With the presented method, the dose and corresponding mean photon energy can be calculated separately for patient scatter, collimator scatter, and head leakage. Hence, for every scatter contribution, a separate RBE for cancer induction can be determined.
Close to the field edge, where patient scatter and collimator scatter dominated (see Fig. 2), the 6 MV treatments showed a higher dose than the 15 MV counterparts. This is in agreement with a MC study from the literature. 35 For a standard field, patient scatter was increased by a factor of two for 6 MV compared to 15 MV (Fig. 8), whereas collimator scatter was reduced just by a factor of 1.5 for 6 MV compared to 15 MV (Fig. 9). This factor was reduced further because collimator scatter scales with the applied MUs 16    H¨alg et al. 12

4.C | Total dose equivalent
The DEVHs of the calculated photon stray dose were in good agreement with the DEVHs of the measurement (Fig. 5). Hence, the measurement locations represented a whole-body photon dose well. In comparison, DEVH of the calculated neutron dose equivalents showed more dose per volume than the DEVHs from the measurements. This can be explained by the fact that most of the TLD measurement locations were deeper than 1 cm in the phantom. 20 The locations were chosen such that they cover all ICRP-recommended organs. 37 The neutron dose decreases rapidly with increasing depth in the phantom [ Fig. 1(a)]. In terms of radiation protection, the high neutron dose contribution down to 1 cm in the patient is of less importance since most ICRP organs are located deeper in the body. and are in a first approximation independent of the field shape. 1 The relative difference in MUs for the 6 MV compared to the 15 MV treatment was smaller for the intensity-modulated treatments when compared to the 3DCRT treatments (see Table 1). This explained the crossing of the 6 and 15 MV DEVHs for the intensity-modulated treatments at 0.2 Sv (see Fig. 5c). For the 3DCRT treatments, the DEVH of the 15 MV plan was equal or below the 6 MV DEVH.
Hence, regarding radiation protection of the patient, the 3DCRT 15 MV treatment was superior compared to the 6 MV treatment.
Head leakage and neutron DEVHs can be seen in Fig. 7.
Head leakage was almost constant in the phantom, whereas neutron dose was inhomogeneous. For all techniques, the minimum dose was higher for 6 MV than for 15 MV. This was caused by a smaller leakage dose for 15 MV than for 6 MV (Fig. 9) and the low neutron dose in the center of the body [ Fig. 1 (c)]. Head leakage is assumed to be reduced because of more forward-directed photons in the X-ray producing target for 15 MV than for 6 MV. Primus). 40 Hence, regarding radiation protection of the patient for high-energy treatments, the choice of treatment machine should not only be based on neutron production but rather on the total dose equivalent including photon scatter.
Multiple studies reported an increased cancer risk based on an increased dose equivalent for high energy compared to low-energy radiotherapy. 35,39 The increased dose equivalent for high compared to low-energy X-ray therapy was reported to be caused by the additional neutron dose. However, the neutron energies in these publications were overestimated resulting in an overestimation of neutron dose. An extensive discussion of the overestimation in neutron dose reported by the literature can be found in Kry et al. 7 For the same treatment technique, we did not notice an increase in the whole-body dose equivalent for increasing nominal X-ray energy (up to 15 MV). For IMRT treatments, the findings are in agreement with an MC study made by Kry et al. 7 They found that, the calculated 6 and 18 MV out-of-field doses were similar for IMRT.
A shortcoming of this study is that the investigation was focused only to one treatment location (rhabdomyosarcoma in the prostate).
In addition, the whole-body doses were calculated using an anthropomorphic phantom and not a patient CT. This had the advantage that the calculation could be directly compared to the measurements. Nevertheless, the dose models used in this manuscript are applicable to other treatment locations and patient geometries. The photon dose calculated using the stray dose model 16  These investigations are urgently needed. 5 The calculated CBCT dose distribution was based on whole-body measurement. It is time consuming and not practical to measure the dose of various CBCT protocols. Furthermore, the choice of the protocol influences the dose distribution in the field-of-view region. 29 Analytical models to calculate the CBCT dose are available but they lack the ability to calculate the dose outside the field-of-view. 42 We did not include the out-of-field dose caused by electron contamination. Outside the primary beam, the dose close to the surface can be increased by a factor of 4 compared to inside the body (>2 cm). 10  For the calculation of the total whole-body dose of a real patient, the planning CT can be fused with a phantom containing the contours of critical tissues. 43 Such a feature is not yet clinically available.
Using the application programming interface of the Eclipse TPS, it is planned to fuse the limited patient CT with a computational human phantom from a library to generate a whole-body representation of the patient.

| CONCLUSION
The calculated whole-body dose equivalent for IGRT treatments helped Patient scatter D ps is the dose mainly produced by Compton scatter photons of the treatment field penetrating the patient. Photons of the primary field scattered at the jaws and MLC are described by collimator scatter D cs . Head leakage is the out-of-field dose contribution from photons which originate from the X-ray producing target and leaking through the gantry head shielding D hl . In Fig. 8, we see the comparison between patient scatter resulting from 15 to 6 MV nominal X-ray energy. The normalized patient scatter as a function of field widths and lengths showed no difference between 15 and 6 MV [ Fig. 8(a)]. The absolute patient scatter dose 15 cm from the field edge along a line parallel to the central field axis (R-dependence 16 ) showed a higher dose for 6 MV than for 15 MV [ Fig. 8(b)]. HAURI AND SCHNEIDER | 67 For 6 MV, Chofor et al. 33 simulated an almost twofold increase in patient scatter compared to 15 MV. Ruben et al. 8 measured a 1/3 higher patient scatter dose for 6 MV than for 18 MV. However, the difference in patient scatter between two nominal X-ray energies is dependent on the location in the patient, as can be seen in Fig. 8(b). Patient scatter is independent from the beam head design and is an unavoidable result of external radiotherapy. 33 Table III shows that primarily the normalization constant of the patient scatter model has to be adjusted when using different beam energies. The normalization constant can be determined by one measurement set-up. However, for flattening filter free beams, all model parameters could be different compared to flattening filter beams.

Collimator scatter and head leakage models
The empirical models of collimator scatter and head leakage 16  F I G . 8. In (a) the measured and model-predicted field length dependence of patient scatter. Furthermore, the fit of the field width dependence. In (b) the measured and model-predicted dose along a line parallel to the central field axis 15 cm from the field edge (R-dependence 16 ). With the results, a scaling matrix was calculated which corrected for the change of collimator scatter with changing MLC width opening. No such effect was noticed for a variation in the MLC length opening. 16 To calculate collimator scatter and head leakage for a treatment, the dose contributions along the MPAX were scaled with the exponential decrease in depth. To determine the depth, the patient was assumed to be water equivalent with the exception of the lungs (ρ lungs = 0.25 × ρ water ). A more detailed description how the depth in the patient was calculated for each treatment field can be found in Hauri et al. 16 In Fig. 9, we see collimator scatter and head leakage per 100 MUs along the MPAX in the center of the RW3 phantom. Collimator scatter is shown for a fixed jaw field size of 10 × 10 cm 2 in combination with an MLC field size of 9 × 10 cm 2 , representing a 3DCRT field, and for an MLC field size of 3 × 10 cm 2 , representing an IMRT field. The field sizes were similar to the field sizes of the 3DCRT and IMRT treatments measured in the presented work. For calculation of the dose, the IMRT treatment with dynamic MLC was approximated by static fields. For the 3DCRT fields, the ratio of collimator scatter between 15 and 6 MV decreased from 1.5 to 0.88 for increasing distance to the isocenter. This is in agreement with other publications, reporting increased collimator scatter close to the field edge with increasing energy. 1 Close to the field edge, collimator scatter of the 15 MV IMRT field was 0.7 times smaller than the 15 MV 3DCRT field. We believe that this difference was caused by the shielding effect of the MLC for photons of the primary beam scattered in the gantry head. The ratio of collimator scatter between the IMRT and the 3DCRT field rose to 1.3 (25 cm from isocenter) with increasing distance to the field edge. As a potential explanation serves the additional photon scatter from the MLC, which outweighed the shielding effect of the MLC. At distances ≥40 cm from the isocenter, there was no difference in collimator scatter per MU between the 3DCRT and IMRT field.
Along the MPAX, head leakage for 6 MV was around 1.6 times higher than head leakage for 15 MV. For both nominal X-ray energies, head leakage showed the same increase for increasing distance to the isocenter. A possible explanation could be the change in photon attenuation for different paths through the gantry head's shielding.
In Table 4, we see the measured attenuation coefficient for collimator scatter and head leakage. In contrast to the expectation, the attenuation of the scatter contribution was higher for 15 MV than for 6 MV. The attenuation for collimator scatter was calculated using a straight line connecting the rear jaw and the point of interest in the patient. For head leakage, the attenuation was calculated using a straight line connecting the X-ray producing target with the point of interest located in the patient.

R E F E R E N C E S
T A B L E 4 The attenuation coefficients in water for collimator scatter and head leakage.