- Email: [email protected]
Essential considerations for accurate evaluation of photoneutron contamination in Radiotherapy
Applied Radiation and Isotopes 145 (2019) 24–31
Contents lists available at ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
Essential considerations for accurate evaluation of photoneutron contamination in Radiotherapy ⁎
T
⁎
Amir Hossein Karimia, , Hrvoje Brkićb,c, Daryoush Shahbazi-Gahroueid, , Somayeh Biparva Haghighie, Iraj Jabbarif a
Department of Medical Physics, School of Medicine, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran Department of Biophysics and Radiology, Faculty of Medicine, University of J.J. Strossmayer, Osijek, Croatia c Department of Biophysics and Radiology, Faculty of Dental Medicine and Health, University of J.J. Strossmayer, Osijek, Croatia d Department of Medical Physics, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran e Department of General Courses, School of Medicine, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran f Department of Nuclear Engineering, Faculty of Advanced Sciences and Technologies, University of Isfahan, Isfahan, Iran b
H I GH L IG H T S
MCNPX code was used to simulate the photoneutron production in the head of a medical LINAC (Varian Clinac 2100C/D 18 MV). • The ambient dose equivalent and energy spectrum were calculated for different field sizes at the patient table and inside the maze. • Neutron energy changed with field size and also detector location. • Neutron the change in neutron energy lead to major uncertainty in dosimetry. • Neglecting • Neutron ambient dose equivalent at the patient table changed with removing Radiation Therapy room from simulations. ®
A R T I C LE I N FO
A B S T R A C T
Keywords: Radiotherapy LINAC Photoneutron contamination Medical dosimetry Monte Carlo
Nowadays, high-energy X-rays produced by medical linear accelerators (LINACs) are widely used in many Radiation Therapy (RT) centers. High-energy photons (> 8 MeV) produce undesired neutrons in the LINAC head which raise concerns about unwanted neutron dose to the patients and RT personnel. Regarding the significance of radiation protection in RT, it is important to evaluate photoneutron contamination inside the RT room. Unfortunately, neutron dosimeters used for this purpose have limitations that can under the best conditions cause to > 10% uncertainty. In addition to this uncertainty, the present Monte Carlo (MC) study introduces another uncertainty in measurements (nearly up to 20%) when neutron ambient dose equivalent (H*n (10)) is measured at the patient table or inside the maze and the change in neutron energy is ignored. This type of uncertainty can even reach 35% if H*n (10) is measured by dosimeters covered by a layer of 10B as converter. So, in these cases, neglecting the change in neutron energy can threaten the credibility of measured data and one should attend to this energy change in order to reduce measurement uncertainty to the possible minimum. This study also discusses the change in neutron spectra and H*n (10) at the patient table caused by removing a typical RT room from MC simulations. Under such conditions, neutron mean energy (Ēn) overestimated by 0.2–0.4 MeV at the patient table. Neutron fluence (φn) at the isocenter (IC) was underestimated by 23–54% for different field sizes that caused H*n (10) to be miscalculated up to 24%. This finding informs researchers that for accurate evaluation of H*n (10) at the patient table, simulating the RT room is an effective parameter in MC studies.
1. Introduction Nowadays, high-energy X-rays produced by LINACs are widely used in many RT centers. The usage of these beams, instead of low energy photons, has many advantages, such as good skin sparing, less
⁎
scattering, sharp field edges, small penumbra, and uniform spatial dose distribution (Das and Kase, 1992; Vega-Carrillo and Baltazar-Raigosa, 2011). Despite such advantages, high-energy photons (> 8 MeV) produce undesired neutrons in the LINAC head (Khabaz et al., 2018; Khosravi et al., 2013; NCRP60, 1984; Vega-Carrillo et al., 2015, 2012)
Corresponding authors. E-mail addresses: [email protected] (A.H. Karimi), [email protected] (D. Shahbazi-Gahrouei).
https://doi.org/10.1016/j.apradiso.2018.12.007 Received 23 August 2018; Received in revised form 7 November 2018; Accepted 4 December 2018 Available online 05 December 2018 0969-8043/ © 2018 Elsevier Ltd. All rights reserved.
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Fig. 1. The schematic of simulated Varian Clinac 2100 C/D 18 MV. Table 1 Main components of the LINAC head. Part
Shape
Materials & density (g/cm3)
Target
Cylindrical
Primary collimator Flattening filter
Inside:Conical Outside:Cylindrical Conical
W: 19.30 Cu: 8.96 W: 19.30
Ion chamber
Cylindrical
Mirror
Flat
Secondary collimator
Inside:Conical Outside:Cylindrical Cubical Circular
Jaw Upper circle a b
Ta: 16.65 Fe: 7.86 Cu: 8.96 Kaptona: 1.42 Al:2.70 Mylarb: 1.39 W: 19.30 Pb: 11.34 W: 19.30 Fe: 7.86
2.6% H, 69.1% C, 7.3% N and 21.0% O. 4.2% H, 62.6% C and 33.2% O.
through Giant Dipole Resonance process (γ, n) (Esposito et al., 2008; Mohammadi et al., 2014; Vega-Carrillo et al., 2009). This raises concerns about unwanted neutron dose to the patients (McGinley et al., 1976; Mohammadi et al., 2015; Shahbazi-Gahrouei et al., 2012) and RT personnel (Donadille et al., 2008; Esposito et al., 2008). Regarding the significance of radiation protection in RT, it is important to evaluate photoneutron contamination inside the RT room. Neutron dosimeters used for this purpose are unfortunately saturated by initial or leakage photon flux; also, their response depends strongly on neutron energy (Chibani and Ma, 2003; Zabihinpoor and Hasheminia, 2011). These limitations imply that uncertainty < 10% cannot be achieved in measurements (Chibani and Ma, 2003). In addition to this uncertainty, it has been recently found in a MC study that if the change in neutron energy is neglected when H*n (10) is measured at the patient table, there will be up to 20% uncertainty in measurements (Brkić et al., 2016). Neglecting the neutron energy change can predictably lead to more measurement uncertainty if H*n (10) is measured by dosimeters whose response depends strongly on energy (Brkić et al., 2016). A kind of these dosimeters is nuclear track detectors CR-39 covered by a layer of 10 B as converter. In these detectors, 10B converts neutrons to alpha particles through 10B (n, α) 7Li reaction whose cross-section depends
Fig. 2. The ground plan of simulated bunker geometry. The height of room is 5.5 m. The thicknesses of the floor and ceiling concrete are 1.0 m and 1.7 m, respectively. The position of the detectors was numerated and shown using small circles.
substantially on neutron energy (Khaled et al., 2014; Vukovic et al., 2010). So far, the possible effects of neglecting neutron energy change on measurement uncertainty, when H*n (10) is measured inside the maze, have not yet been studied. Therefore, this work attempted to evaluate the predictions regarding the measurement uncertainties caused by ignoring neutron energy change inside the maze and also at the patient table. 25
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Percentage Depth Dose (PDD) and dose profile, in electron-photon mode for 4 × 4 cm2, 10 × 10 cm2 and 20 × 20 cm2 field sizes (Chegeni et al., 2018). Some of the characteristics of this model are given briefly below: PDDs had been calculated inside a water phantom along the central axis up to 30 cm depth. Here, the depth of maximum dose (dmax) was found 3.4 cm. In addition, the dose profiles had been calculated in offaxis direction for the mentioned field sizes. The calculated data had been benchmarked with measured data using γ-index function, a useful tool for comparing two dose distributions (for more information about γ-index function, please see the Low et al., 1998 paper). In this model, energy and spatial distribution of electrons impinging the target are Gaussian (mean energy = 18.3 MeV, energy FWHM = 1.2 MeV and spatial FWHM = 0.14 cm). In addition, the number of electrons needed to deliver 1 Gy photon dose to dmax, in the standard condition (Field Size (FS): 10 × 10 cm2, SSD = 100 cm), is 1.26 × 1015 electrons per Gy, which is comparable with 1.52 × 1015 electrons per Gy reported in the literature (Martinez-Ovalle et al., 2011). This factor is multiplied by calculated φn and H*n (10) to report these values according to 1 Gy photon dose delivered to dmax (MCNP code normalizes the output to one particle source). The neutron source strength for this model had been found 1.38 × 1012 n Gy−1 (for closed jaws) which is similar to 1.2 × 1012 n Gy−1 and 1.3 × 1012 n Gy−1 reported in other studies (Mao et al., 1997; Mesbahi et al., 2010). These differences seem reasonable as the type of MC code, cross-section libraries for photoneutron production, and simulated geometry of LINAC head differ from model to model. Thus, we were assured that this model is reliable for evaluating photoneutron contamination. The ground plan of the simulated RT room is shown in Fig. 2, where the LINAC has been located (Golestan Hospital, Ahvaz, Iran). The concrete walls had the density of 2.35 g/cm3 which were constructed from: 49.83% oxygen, 31.58% silicon, 8.26% calcium, 4.56% aluminum, 1.92% potassium, 1.71% sodium, 1.22% iron and 0.92% hydrogen (all in terms of their weight percentages). To consider neutron transmission in the simulations, neutron mode was added to electron-photon mode. Also, the fourth parameter of photon physics card was set in the bias state (PHYS:P 40 j j 1 j j) to be focused on photoneutron production. To account all available interactions between different particles into the MC simulation, cross-section libraries of ENDF/B-V2.0, MCPLIB04 and EL03 were used for neutron, photon, and electron, respectively. These interactions included; absorption, dispersion, and capture for neutron, Rayleigh and Compton scattering, pair production, and the photoelectric effect for photon, and finally elastic and inelastic collision, Bremsstrahlung, and annihilation for electron and positron. Additionally, cross-section libraries of LA150u, KAREI01u and CNDC01u were applied to simulate photoneutron production in the LINAC head.
Table 2 Position of detectors located at the patient table (detectors 1–8) and inside the maze (detectors 9–12). Detector number
X (cm)
Y (cm)
Z (cm)
1 2 3 4 5 6 7 8 9 10 11 12
0 0 0 0 0 0 0 0 360 120 − 120 − 360
0 − 10 − 20 − 30 − 40 − 50 − 60 − 70 − 620 − 620 − 620 − 620
0 0 0 0 0 0 0 0 0 0 0 0
Regarding the mentioned limitations for neutron dosimeters, MC studies are more accurate for the evaluation of neutron contamination in RT, especially for neutron dosimetry in the patient's organs (Khabaz et al., 2018; Mohammadi et al., 2014). Some of the MC studies overlooked RT room in the simulations (Brkić et al., 2016; Frankl and Macián-Juan, 2015; Rezaian et al., 2017) while concrete walls of RT room can scatter the neutrons toward the patient table. Recently, the change in neutron spectrum, when neutrons are scattered from walls of vault, has been introduced as an interesting subject for further investigation (Brkić et al., 2016). However, the effect of neglecting the RT room in MC simulations on neutron spectra and H*n (10) has not yet been thoroughly studied. Accordingly, the next aim of this work is to evaluate the change in Ēn , φn and H*n (10) at the patient table when a typical RT room (including floor, ceiling, and walls made of concrete) is removed from MC simulations. 2. Materials and methods This study aims to present key points for MC and experimental studies on photoneutron contamination in RT. For this purpose, the photoneutron production in the head of Varian Clinac 2100 C/D 18 MV (California, Varian Medical Systems, USA) was simulated using MCNPX® (2.6.0) code (Pelowitz, 2008). The main parts of this LINAC has been modeled according to the available manufacture data and literature (Alem-Bezoubiri et al., 2014). Cross-section of simulated LINAC head is shown in Fig. 1. The model consists of target, beryllium window, primary collimator, flattening filter, ion chamber, mirror, secondary collimator, jaws, and upper circle (Table 1). The model had already been verified with measured data,
Table 3 Neutron mean energy (Ēn) at different positions and field sizes with considering the concrete walls in MC simulations (CW) and without (No CW). Position
Ē n (MeV) 0 × 0 cm2
1 2 3 4 5 6 7 8 9 10 11 12
4 × 4 cm2
6 × 6 cm2
10 × 10 cm2
20 × 20 cm2
40 × 40 cm2
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
0.283 0.286 0.287 0.297 0.320 0.343 0.318 0.307 0.172 0.061 0.036 0.028
0.496 0.498 0.510 0.540 0.591 0.640 0.632 0.637 – – – –
0.542 0.368 0.361 0.345 0.341 0.355 0.327 0.313 0.170 0.061 0.036 0.028
0.781 0.576 0.580 0.575 0.591 0.635 0.623 0.629 – – – –
0.597 0.398 0.386 0.366 0.353 0.360 0.332 0.317 0.173 0.061 0.037 0.028
0.820 0.607 0.604 0.594 0.600 0.635 0.624 0.629 – – – –
0.647 0.454 0.419 0.400 0.377 0.371 0.343 0.323 0.171 0.061 0.035 0.028
0.846 0.659 0.639 0.630 0.628 0.646 0.634 0.634 – – – –
0.687 0.643 0.473 0.440 0.416 0.398 0.367 0.339 0.176 0.062 0.037 0.029
0.859 0.839 0.703 0.687 0.681 0.684 0.670 0.661 – – – –
0.738 0.750 0.738 0.433 0.360 0.331 0.308 0.292 0.184 0.065 0.037 0.031
0.938 0.951 0.980 0.768 0.696 0.677 0.670 0.666 – – – –
26
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Table 4 Neutron fluence (φn ) at different positions and field sizes with considering the concrete walls in MC simulations (CW) and without (No CW). φn (107 n cm−2 Gy−1)
Position
0 × 0 cm2
1 2 3 4 5 6 7 8 9 10 11 12
4 × 4 cm2
6 × 6 cm2
10 × 10 cm2
20 × 20 cm2
40 × 40 cm2
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
0.834 0.833 0.821 0.812 0.814 0.816 0.765 0.729 0.211 0.090 0.038 0.024
0.386 0.382 0.370 0.362 0.362 0.362 0.311 0.277 – – – –
1.275 1.011 0.974 0.923 0.883 0.864 0.800 0.755 0.213 0.090 0.038 0.023
0.824 0.559 0.522 0.469 0.426 0.405 0.342 0.298 – – – –
1.452 1.087 1.032 0.970 0.913 0.879 0.813 0.763 0.210 0.090 0.038 0.023
0.999 0.634 0.578 0.514 0.455 0.421 0.355 0.307 – – – –
1.681 1.216 1.099 1.029 0.954 0.900 0.830 0.772 0.211 0.089 0.038 0.023
1.227 0.762 0.645 0.573 0.497 0.441 0.371 0.315 – – – –
1.923 1.640 1.141 1.035 0.955 0.890 0.819 0.757 0.204 0.087 0.036 0.023
1.478 1.196 0.693 0.591 0.508 0.444 0.373 0.314 – – – –
1.572 1.586 1.369 0.747 0.661 0.622 0.589 0.562 0.178 0.076 0.032 0.020
1.183 1.197 0.982 0.359 0.272 0.232 0.199 0.175 – – – –
Table 5 Neutron ambient dose equivalent (Hn* (10)) at different positions and field sizes with considering the concrete walls in MC simulations (CW) and without (No CW). [H*n (10)] (mSv Gy−1)
Position
0 × 0 cm2
1 2 3 4 5 6 7 8 9 10 11 12
4 × 4 cm2
6 × 6 cm2
10 × 10 cm2
40 × 40 cm2
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
1.187 1.184 1.170 1.170 1.214 1.263 1.115 1.025 0.171 0.037 0.011 0.006
0.907 0.898 0.882 0.884 0.925 0.974 0.827 0.738 – – – –
2.623 1.694 1.605 1.475 1.385 1.374 1.201 1.085 0.171 0.037 0.011 0.006
2.331 1.403 1.316 1.187 1.098 1.085 0.911 0.797 – – – –
3.192 1.919 1.774 1.608 1.467 1.417 1.237 1.109 0.171 0.037 0.011 0.006
2.902 1.629 1.486 1.320 1.179 1.128 0.948 0.820 – – – –
3.933 2.329 1.994 1.804 1.607 1.484 1.293 1.141 0.170 0.037 0.011 0.006
3.637 2.037 1.703 1.514 1.316 1.192 1.001 0.849 – – – –
4.751 3.850 2.212 1.920 1.691 1.522 1.323 1.153 0.168 0.036 0.011 0.006
4.459 3.558 1.920 1.627 1.398 1.230 1.032 0.865 – – – –
3.989 4.051 3.384 1.299 1.026 0.910 0.816 0.747 0.150 0.032 0.010 0.005
3.720 3.782 3.117 1.034 0.762 0.645 0.552 0.485 – – – –
neglecting the neutron energy change can predictably lead to more measurement uncertainty (Brkić et al., 2016). To investigate the credibility of this prediction, the expected number of alpha particles, recorded by this type of dosimeters, was estimated in various points inside the maze (detectors 9–12). These calculations were performed by multiplying the cross-section of 10B (n, α) 7Li reaction by the neutron spectra, arbitrary in 20 × 20 cm2 FS. In fact, the cross-section of 10B (n, α) 7Li was divided to several energy segments. These parts were chosen to fit curves by the least square method (each segment had its own equation). The midpoint of each logarithmic energy bin was inserted in the obtained equations and was then multiplied by the corresponding neutron fluence. Next, the expected number of alpha particles and H*n (10) were normalized to their maximum values. Subsequently, in each point, the H*n (10) ratio was compared to alpha particles ratio. These comparisons estimate the possible uncertainty in dosimetry (for dosimeters that use 10B) when the change in neutron energy inside the maze is ignored. Finally, by removing concrete walls from simulations, Ēn , φn and H*n (10) were calculated at the patient table (detectors 1–8). As a final point, to achieve valid confidence intervals for MC calculations, at least 5 × 108 electron histories were traced in each simulation. Relative statistical uncertainty for total φn (also H*n (10)) was kept < 1% at the patient table and < 2% inside the maze. Regarding the neutron spectra, scaled by % of total fluence, the average absolute uncertainty for normalized fluence was < 0.1%. In all calculations, to short the run time of programs, energy cut-offs for electron and photon were chosen as 6 MeV (less than threshold energy of photonuclear
Using F5 tally (point detector), neutron spectra were calculated in different field sizes (0 × 0, 4 × 4, 6 × 6, 10 × 10, 20 × 20 and 40 × 40, all in cm2) for several points at the patient table and inside the maze (Table 2 and Fig. 2). Besides, using F5 tally, DF card and ICRP 74 conversion factors (1996), H*n (10) was calculated for all the detectors. Using obtained spectra and according to Eq. (1), Ēn was computed for each detector. 100
Ēn =
20 × 20 cm2
∑a = 1 Ea.ϕ(Ea) 100
∑a = 1 ϕ(Ea)
(1) −9
where energy bin of [10 , 10] MeV is divided into 100 bins logarithmically. Φ (Ea) is the calculated neutron fluence related to the ath energy bin and Ea is the middle point in the ath energy bin. To investigate the amount of uncertainty expected during neutron dosimetry at the IC, only due to neglecting the change in Ēn with FS, the calculated values for φn and H*n (10) at the IC were normalized to their related maximum value (i.e. in 20 ×20 cm2 FS). A similar analysis was performed also along the patient table to determine the amount of possible measurement uncertainty caused only by neglecting the change in Ēn with detector position. In this regard, the calculated values for φn and H*n (10) along the patient table (arbitrary in 20 × 20 cm2 FS) were normalized to their related maximum value (i.e. to IC). Finally, in each case, the φn ratio was compared with the H*n (10) ratio and introduced as expected uncertainty in measurements. Detectors that covered by a layer of 10B as converter are commonly used for neutron dosimetry in different fields, including RT. As mentioned before, if H*n (10) is measured by these types of detectors, 27
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Fig. 4. Neutron fluence (φn ) and ambient dose equivalent (H*n (10)) normalized to their maximum values: (A) for different field sizes at the IC, (B) for different points at the patient table.
Fig. 3. Neutron spectra: (A) for different field sizes at the isocenter, (B) for different positions at the patient table.
reactions in the LINAC head). The energy cut-off for neutron remained the MCNP default (0 MeV).
table. In addition, it is perceptible that φn inside the maze does not depend on FS and depends only on the detector position. By changing the position from 9 to 12, φn decreases nearly 0.9 times (neutrons were attenuated by concrete walls and fewer neutrons reached the entrance). According to Table 5, H*n (10) for different field sizes at the IC ranged from 1.187 to 4.751 mSv·Gy−1 which was at least 6 times higher than H*n (10) inside the maze. This value becomes more significant (83–333 mSv) when 70 Gy photon dose is delivered to a tumor in a complete treatment. The issue will be more critical for modern RT techniques which use more monitor units for treatment, in turn might increase the risk of secondary cancer (Brkić et al., 2016). So regarding the significance of radiation protection in RT, it is important to evaluate photoneutron dose accurately inside the RT room. It is also evident that with moving off axis from IC to 70 cm distant point, H*n (10) drops to 1.043 mSv·Gy−1 averagely on all field sizes. Table 5 also shows that H*n (10) inside the maze is independent of FS and decreases only by moving away from the end of maze (detector 9), a phenomenon that was previously reported using a simplified MC model (Mohammadi et al., 2014; Zabihzadeh et al., 2009). Dependence of neutron energy spectra to FS (Fig. 3A) and off-axis distance (Fig. 3B) was analyzed at the patient table with the presence of RT room in the simulations. Accordingly, one can observe that the neutron spectra are characterized by two peaks. The first one is located in the thermal energy region, i.e. [10−9, 10−7] MeV, and the second one in the fast energy region, i.e. [10−1, 10] MeV. A flat intermediate
3. Results and discussion At the first step, Ēn was calculated for several field sizes at the patient table (with/without considering vault walls) and inside the maze (Table 3). Table 3 shows that Ēn for different field sizes at the IC ranges from 0.283 MeV to 0.738 MeV in the presence of concrete walls. These results are in fairly good agreement with recent research (Alem-Bezoubiri et al., 2014). After removing concrete walls from MC simulations, Ēn shifted toward the higher energies averagely about 0.2 MeV (0.496 – 0.938 MeV). This effect was more evident for detectors located far away from IC, so as for detector 8 (70 cm of IC), Ēn shifted toward the higher energy up to 0.4 MeV in some cases (FS: 40 × 40 cm2). Next, φn and H*n (10) were analyzed for different field sizes at the patient table and inside the maze. Results are summarized respectively in Tables 4 and 5 which comply well with the relevant data reported in other studies (Alem-Bezoubiri et al., 2014; Kralik et al., 2008). These tables also present the data related to the circumstances where the RT room was removed from MC simulations. In consort with the recent research (Brkić et al., 2016), Table 4 indicates that φn at the IC increases with FS and reaches maximum in 20 × 20 cm2 while it decreases with moving off axis at the patient 28
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
2008). Regarding Fig. 3A, it is evident that neutron energy spectra and consequently Ēn at the IC strongly depends on FS. When FS changes from 0 × 0 cm2 to 40 × 40 cm2, Ēn at the IC will be 2.6 times (Table 3). Moreover, as Fig. 3B illustrates if only the position at the patient table changes, using the same FS, Ēn will shift to lower energies. Herein for detector 8 (70 cm of IC), Table 3 shows nearly 50% reduction (related to IC) in Ēn . Awareness of neutron mean energy is necessary to choose flux-todose conversion factors correctly in neutron dosimetry (ICRP, 1996). If the change in neutron energy with FS is not taken into account when H*n (10) is measured at the IC, up to 19% measurement uncertainty can be expected (Fig. 4A). This finding is well confirmed by Brkić et al.’s work (2016) which introduced 20% measurement uncertainty in this case. This type of measurement uncertainty can be expected when fluxes are converted to doses using constant factors and overlooking the neutron energy change (i.e. with changing FS at IC, φn ratio was assumed as H*n (10) ratio). Besides, when H*n (10) is measured at different points on the patient table (using the same FS) and the shift in neutron energy is not taken into account, the measurement uncertainty can be even > 10% (Fig. 4B) in most detectors, as it was already reported by Brkić et al. (2016). The above findings interested us to perform similar analysis for detectors located inside the maze. The dependence of neutron energy spectra on FS and distance inside the maze were summarized in Figs. 5A and 5B, respectively. According to Fig. 5A, it can be seen that for an arbitrary point inside the maze (detector 11), the change in FS does not have significant effects on neutron energy spectra. Therefore, if the change in neutron energy with FS is ignored during neutron dosimetry inside the maze, the measurement uncertainty will be less pronounced (up to 7%, Fig. 6). On the other hand, Fig. 5B shows that the change in detector position from 9 to 12, using the same FS, lead to a significant change in neutron energy (also see Table 3). So if this energy shift is not considered when H*n (10) is measured at different positions inside the maze, there will be nearly 21% uncertainty in measurements (Fig. 7, difference between H*n (10) and φn ). This type of uncertainty can even approach 35% (Fig. 7, difference between H*n (10) and the expected number of alpha particles) if measurement is performed by dosimeters use 10B as converter. As Brkić et al. (2016) had predicted, the large energy dependence of 10B (n, α) 7Li cross-section can lead to more measurement uncertainty when the change in neutron energy is neglected. Finally, with/without considering the RT room in MC simulations, neutron energy spectra were compared together at the IC in 0 × 0 cm2 FS, illustrated in Fig. 8. Accordingly, one can observe major changes in the neutron spectra with removing the RT room from MC simulations. The figure also demonstrates when the concrete walls were not accounted for MC simulations, there were only fast neutrons in the spectrum. This means that all the neutrons produced in the LINAC head lie in the fast region, an occurrence that was also encountered in an earlier study (Mohammadi et al., 2015). In consort with Pena et al. (2005) study, Fig. 8 illustrates that with considering the RT room in MC simulations, the fluence of fast neutrons increases at the IC. It seems rational since some of the fast neutrons which were produced in the LINAC head can be scattered elastically from the concrete walls toward the patient table and consequently can increase the fluence of fast neutrons at the IC. Removing concrete walls from MC simulations had another effect on the shape of neutron spectrum. Neutrons could also be scattered inelastically for several times from concrete walls toward the patient table and subsequently appeared as thermal and epithermal neutrons in the spectrum (Fig. 8). It is noteworthy that with increasing distance from IC, the amount of thermal and epithermal neutrons remain practically the same while the fast neutrons are reduced (Table 4), a phenomenon which was also reported previously by Vega-Carrillo and Baltazar-
Fig. 5. Neutron spectra inside the maze: (A) for different field sizes, (B) for different positions.
Fig. 6. Neutron fluence (φn ) and ambient dose equivalent (H*n (10)) at detector 11 for different field sizes normalized to their maximum values (i.e. to 0 × 0 cm2).
energy region, i.e. [10−7, 10−1] MeV, related to epithermal neutrons is located in the middle. The shapes of the calculated spectra correspond well with those measured before (Esposito et al., 2008; Kralik et al.,
29
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Fig. 7. Expected number of alpha particles, neutron fluence (φn ) and neutron ambient dose equivalent (Hn* (10)) inside the maze normalized to their maximum values (i.e. to detector 9).
Fig. 9. Change in neutron ambient dose equivalent (H*n (10)) with removing concrete walls from MC simulations: (A) at the IC, (B) at 70 cm of IC.
Fig. 8. Neuron spectra at the IC with and without considering the concrete walls in MC simulations.
photoneutron dose in RT. At first, when H*n (10) is measured at the patient table or inside the maze and the change in neutron energy (with FS or detector location) is ignored, there will be nearly up to 20% measurement uncertainty. This type of uncertainty can approach 35% if H*n (10) is measured by dosimeters whose response depends strongly on energy (e.g. CR-39 covered by layer of 10B as converter). So, in these cases, neglecting the change in neutron energy can threaten the credibility of measured data and one should attend to this energy change in order to reduce measurement uncertainty to the possible minimum. For this purpose, MC calculations can be used as a correction for measured data where neglecting the neutron energy change lead to large differences between φn ratio and H*n (10) ratio (Figs. 4, 6 and 7). Secondly, it was found when the RT room is not considered in MC simulations, Ēn is overestimated by 0.2–0.4 MeV at the patient table. φn at the IC will be underestimated by 23–54% for different field sizes that cause H*n (10) to be miscalculated up to 24%. This finding informs researchers that for accurate evaluation of H*n (10) at the patient table, simulating RT room is an effective parameter in MC studies. Although the calculated values for H*n (10) in the air (at the patient table and inside the maze) can be used as a good parameter for radiation protection aspects, it is far better to consider a more realistic situation. With considering this point and using an anthropomorphic phantom in MC simulations, our future study will concentrate on unwanted H*n (10) in the patient's tissues/organs (risk of secondary cancer) for different RT techniques.
Raigosa (2011). So, if concrete walls are overlooked in MC simulations, the contribution of the scattered neutrons in spectra is ignored and consequently φn at the IC is underestimated by 23–54% for different field sizes (Table 4). In another MC study, Alem-Bezoubiri et al. (2014) showed for 10 × 10 cm2 FS, the contributions of the scattered neutrons in total φn at the IC and 40 cm of IC were 27% and 58% respectively which are similar to the corresponded values of 27% and 48% in this project (Table 4, reduction in φn with removing the RT room). This underestimation will be even higher when moving off axis at the patient table, so as for detector located at 70 cm from IC, it approaches 69% in some field sizes (40 × 40 cm2). It seems reasonable as by taking distance from IC, the contribution of thermal and epithermal neutrons in the spectrum will increase (Fig. 3B). As this reduction occurs mainly in thermal and epithermal energy region (Fig. 8), where flux-to-dose conversion factors have low values (compared to fast region), H*n (10) is miscalculated on average 0.286 mSv Gy−1 at the patient table (Table 5). This miscalculation cause to maximum 24% underestimation in the evaluation of H*n (10) at the IC (Fig. 9A) that can reach even 35% for detector located at 70 cm of IC (Fig. 9B). 4. Conclusion This study discussed critical issues (via MC simulation) and suggested precautions to be considered for the accurate evaluation of 30
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Conflict of interest
Phys. 72, 524–529. Martinez-Ovalle, S.A., Barquero, R., Gomez-Ros, J.M., Lallena, A.M., 2011. Neutron dose equivalent and neutron spectra in tissue for clinical linacs operating at 15, 18 and 20 MV. Radiat. Prot. Dosim. 147, 498–511. McGinley, P., Wood, M., Mills, M., Rodriguez, R., 1976. Dose levels due to neutrons in the vicinity of high-energy medical accelerators. Med. Phys. 3, 397–402. Mesbahi, A., Ghiasi, H., Mahdavi, S.R., 2010. Photoneutron and capture gamma dose equivalent for different room and maze layouts in radiation therapy. Radiat. Prot. Dosim. 140, 242–249. Mohammadi, N., Miri-Hakimabad, H., Rafat-Motavlli, L., Akbari, F., Abdollahi, S., 2015. Neutron spectrometry and determination of neutron contamination around the 15 MV Siemens Primus LINAC. J. Radioanal. Nucl. Chem. 304, 1001–1008. Mohammadi, N., Miri-Hakimabad, S.H., Rafat-Motavalli, L., 2014. A Monte Carlo study for photoneutron dose estimations around the high-energy linacs. J. Biomed. Phys. Eng. 4, 127–140. NCRP60, 1984. Neutron Contamination from Medical Electron Accelerators: Recommendations of the National Council on Radiation Protection and Measurements, Ncrp Reprt 60, Bethesda, MD, USA. Pelowitz, D.B., 2008. MCNPX User's Manual, Version 2.6.0. Los Alamos National Laboratory, Los Alamos. Pena, J., Franco, L., Gomez, F., Iglesias, A., Pardo, J., Pombar, M., 2005. Monte Carlo study of siemens PRIMUS photoneutron production. Phys. Med. Biol. 50, 5921–5933. Rezaian, A., Nedaie, H.A., Banaee, N., 2017. Measurement of neutron dose in the compensator IMRT treatment. Appl. Radiat. Isot. 128, 136–141. Shahbazi-Gahrouei, D., Khosravi, M., Jabbari, K., Gheisari, R., 2012. Measurement of photoneutron dose in the linear accelerator at the radiation therapy section of SeyedAl-Shohada hospital, Isfahan, Iran. J. Isfahan Med. Sch. 29. Vega-Carrillo, H.R., Baltazar-Raigosa, A., 2011. Photoneutron spectra around an 18 MV linac. J. Radioanal. Nucl. Chem. 287, 323–327. Vega-Carrillo, H.R., de Leon-Martinez, H.A., Rivera-Perez, E., Luis Benites-Rengifo, J., Gallego, E., Lorente, A., 2015. Induced radioisotopes in a linac treatment hall. Appl. Radiat. Isot. 102, 103–108. Vega-Carrillo, H.R., Hernández-Almaraz, B., Hernández-Dávila, V., Ortíz-Hernández, A., 2009. Neutron spectrum and doses in a 18 MV LINAC. J. Radioanal. Nucl. Chem. 283, 261–265. Vega-Carrillo, H.R., Martínez-Ovalle, S.A., Lallena, A.M., Mercado, G.A., Benites-Rengifo, J.L., 2012. Neutron and photon spectra in LINACs. Appl. Radiat. Isot. 71, 75–80. Vukovic, B., Faj, D., Poje, M., Varga, M., Radolic, V., Miklavcic, I., Ivkovic, A., Planinic, J., 2010. A neutron track etch detector for electron linear accelerators in radiotherapy. Radiol. Oncol. 44, 62–66. Zabihinpoor, S., Hasheminia, M., 2011. Calculation of neutron contamination from medical linear accelerator in treatment room. Adv. Stud. Theor. Phys. 5, 421–428. Zabihzadeh, M., Ay, M.R., Allahverdi, M., Mesbahi, A., Mahdavi, S.R., Shahriari, M., 2009. Monte Carlo estimation of photoneutrons contamination from high-energy Xray medical accelerators in treatment room and maze: a simplified model. Radiat. Prot. Dosim. 135, 21–32.
The authors have no conflicts of interest. References Alem-Bezoubiri, A., Bezoubiri, F., Badreddine, A., Mazrou, H., Lounis-Mokrani, Z., 2014. Monte Carlo estimation of photoneutrons spectra and dose equivalent around an 18 MV medical linear accelerator. Radiat. Phys. Chem. 97, 381–392. Brkić, H., Ivković, A., Kasabašić, M., Poje Sovilj, M., Jurković, S., Štimac, D., Rubin, O., Faj, D., 2016. The influence of field size and off-axis distance on photoneutron spectra of the 18 MV Siemens Oncor linear accelerator beam. Radiat. Meas. 93, 28–34. Chegeni, N., Karimi, A.H., Jabbari, I., Arvandi, S., 2018. Photoneutron dose estimation in grid therapy using an anthropomorphic phantom: a Monte Carlo study. J. Med. Signal. Sens. 8, 175–183. Chibani, O., Ma, C.M.C., 2003. Photonuclear dose calculations for high‐energy photon beams from Siemens and Varian linacs. Med. Phys. 30, 1990–2000. Das, I.J., Kase, K.R., 1992. Higher energy: is it necessary, is it worth the cost for radiation oncology? Med. Phys. 19, 917–925. Donadille, L., Trompier, F., Robbes, I., Derreumaux, S., Mantione, J., Asselineau, B., Amgarou, K., Martin, A., Bottollier-Depois, J.F., Queinnec, F., Aubert, B., Clairand, I., 2008. Radiation protection of workers associated with secondary neutrons produced by medical linear accelerators. Radiat. Meas. 43, 939–943. Esposito, A., Bedogni, R., Lembo, L., Morelli, M., 2008. Determination of the neutron spectra around an 18MV medical LINAC with a passive Bonner sphere spectrometer based on gold foils and TLD pairs. Radiat. Meas. 43, 1038–1043. Frankl, M., Macián-Juan, R., 2015. Monte Carlo simulation of secondary radiation exposure from high-energy photon therapy using an anthropomorphic phantom. Radiat. Prot. Dosim. 168, 537–545. ICRP, 1996. Conversion coefficients for use in radiological protection against external radiation. Ann. ICRP 26, 3–4. Khabaz, R., Boodaghi, R., Benam, M.R., Zanganeh, V., 2018. Estimation of photoneutron dosimetric characteristics in tissues/organs using an improved simple model of linac head. Appl. Radiat. Isot. 133, 88–94. Khaled, N., Ghanim, E., Shinashin, K., El-Sersy, A., 2014. Effect of X-ray energies on induced photo-neutron doses. Radiat. Eff. Defects Solids 169, 239–248. Khosravi, M., Shahbazi-Gahrouei, D., Jabbari, K., Nasri-Nasrabadi, M., BaradaranGhahfarokhi, M., Siavashpour, Z., Gheisari, R., Amiri, B., 2013. Photoneutron contamination from an 18 MV saturne medical linear accelerator in the treatment room. Radiat. Prot. Dosim. 156, 356–363. Kralik, M., Turek, K., Vondracek, V., 2008. Spectra of photoneutrons produced by highenergy X-ray radiotherapy linacs. Radiat. Prot. Dosim. 132, 13–17. Low, D.A., Harms, W.B., Mutic, S., Purdy, J.A., 1998. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25, 656–661. Mao, X., Kase, K., Liu, J., Nelson, W., Kleck, J., Johnsen, S., 1997. Neutron sources in the Varian Clinac 2100C/2300C medical accelerator calculated by the EGS4 code. Health
31
Contents lists available at ScienceDirect
Applied Radiation and Isotopes journal homepage: www.elsevier.com/locate/apradiso
Essential considerations for accurate evaluation of photoneutron contamination in Radiotherapy ⁎
T
⁎
Amir Hossein Karimia, , Hrvoje Brkićb,c, Daryoush Shahbazi-Gahroueid, , Somayeh Biparva Haghighie, Iraj Jabbarif a
Department of Medical Physics, School of Medicine, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran Department of Biophysics and Radiology, Faculty of Medicine, University of J.J. Strossmayer, Osijek, Croatia c Department of Biophysics and Radiology, Faculty of Dental Medicine and Health, University of J.J. Strossmayer, Osijek, Croatia d Department of Medical Physics, School of Medicine, Isfahan University of Medical Sciences, Isfahan, Iran e Department of General Courses, School of Medicine, Ahvaz Jundishapur University of Medical Sciences, Ahvaz, Iran f Department of Nuclear Engineering, Faculty of Advanced Sciences and Technologies, University of Isfahan, Isfahan, Iran b
H I GH L IG H T S
MCNPX code was used to simulate the photoneutron production in the head of a medical LINAC (Varian Clinac 2100C/D 18 MV). • The ambient dose equivalent and energy spectrum were calculated for different field sizes at the patient table and inside the maze. • Neutron energy changed with field size and also detector location. • Neutron the change in neutron energy lead to major uncertainty in dosimetry. • Neglecting • Neutron ambient dose equivalent at the patient table changed with removing Radiation Therapy room from simulations. ®
A R T I C LE I N FO
A B S T R A C T
Keywords: Radiotherapy LINAC Photoneutron contamination Medical dosimetry Monte Carlo
Nowadays, high-energy X-rays produced by medical linear accelerators (LINACs) are widely used in many Radiation Therapy (RT) centers. High-energy photons (> 8 MeV) produce undesired neutrons in the LINAC head which raise concerns about unwanted neutron dose to the patients and RT personnel. Regarding the significance of radiation protection in RT, it is important to evaluate photoneutron contamination inside the RT room. Unfortunately, neutron dosimeters used for this purpose have limitations that can under the best conditions cause to > 10% uncertainty. In addition to this uncertainty, the present Monte Carlo (MC) study introduces another uncertainty in measurements (nearly up to 20%) when neutron ambient dose equivalent (H*n (10)) is measured at the patient table or inside the maze and the change in neutron energy is ignored. This type of uncertainty can even reach 35% if H*n (10) is measured by dosimeters covered by a layer of 10B as converter. So, in these cases, neglecting the change in neutron energy can threaten the credibility of measured data and one should attend to this energy change in order to reduce measurement uncertainty to the possible minimum. This study also discusses the change in neutron spectra and H*n (10) at the patient table caused by removing a typical RT room from MC simulations. Under such conditions, neutron mean energy (Ēn) overestimated by 0.2–0.4 MeV at the patient table. Neutron fluence (φn) at the isocenter (IC) was underestimated by 23–54% for different field sizes that caused H*n (10) to be miscalculated up to 24%. This finding informs researchers that for accurate evaluation of H*n (10) at the patient table, simulating the RT room is an effective parameter in MC studies.
1. Introduction Nowadays, high-energy X-rays produced by LINACs are widely used in many RT centers. The usage of these beams, instead of low energy photons, has many advantages, such as good skin sparing, less
⁎
scattering, sharp field edges, small penumbra, and uniform spatial dose distribution (Das and Kase, 1992; Vega-Carrillo and Baltazar-Raigosa, 2011). Despite such advantages, high-energy photons (> 8 MeV) produce undesired neutrons in the LINAC head (Khabaz et al., 2018; Khosravi et al., 2013; NCRP60, 1984; Vega-Carrillo et al., 2015, 2012)
Corresponding authors. E-mail addresses: [email protected] (A.H. Karimi), [email protected] (D. Shahbazi-Gahrouei).
https://doi.org/10.1016/j.apradiso.2018.12.007 Received 23 August 2018; Received in revised form 7 November 2018; Accepted 4 December 2018 Available online 05 December 2018 0969-8043/ © 2018 Elsevier Ltd. All rights reserved.
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Fig. 1. The schematic of simulated Varian Clinac 2100 C/D 18 MV. Table 1 Main components of the LINAC head. Part
Shape
Materials & density (g/cm3)
Target
Cylindrical
Primary collimator Flattening filter
Inside:Conical Outside:Cylindrical Conical
W: 19.30 Cu: 8.96 W: 19.30
Ion chamber
Cylindrical
Mirror
Flat
Secondary collimator
Inside:Conical Outside:Cylindrical Cubical Circular
Jaw Upper circle a b
Ta: 16.65 Fe: 7.86 Cu: 8.96 Kaptona: 1.42 Al:2.70 Mylarb: 1.39 W: 19.30 Pb: 11.34 W: 19.30 Fe: 7.86
2.6% H, 69.1% C, 7.3% N and 21.0% O. 4.2% H, 62.6% C and 33.2% O.
through Giant Dipole Resonance process (γ, n) (Esposito et al., 2008; Mohammadi et al., 2014; Vega-Carrillo et al., 2009). This raises concerns about unwanted neutron dose to the patients (McGinley et al., 1976; Mohammadi et al., 2015; Shahbazi-Gahrouei et al., 2012) and RT personnel (Donadille et al., 2008; Esposito et al., 2008). Regarding the significance of radiation protection in RT, it is important to evaluate photoneutron contamination inside the RT room. Neutron dosimeters used for this purpose are unfortunately saturated by initial or leakage photon flux; also, their response depends strongly on neutron energy (Chibani and Ma, 2003; Zabihinpoor and Hasheminia, 2011). These limitations imply that uncertainty < 10% cannot be achieved in measurements (Chibani and Ma, 2003). In addition to this uncertainty, it has been recently found in a MC study that if the change in neutron energy is neglected when H*n (10) is measured at the patient table, there will be up to 20% uncertainty in measurements (Brkić et al., 2016). Neglecting the neutron energy change can predictably lead to more measurement uncertainty if H*n (10) is measured by dosimeters whose response depends strongly on energy (Brkić et al., 2016). A kind of these dosimeters is nuclear track detectors CR-39 covered by a layer of 10 B as converter. In these detectors, 10B converts neutrons to alpha particles through 10B (n, α) 7Li reaction whose cross-section depends
Fig. 2. The ground plan of simulated bunker geometry. The height of room is 5.5 m. The thicknesses of the floor and ceiling concrete are 1.0 m and 1.7 m, respectively. The position of the detectors was numerated and shown using small circles.
substantially on neutron energy (Khaled et al., 2014; Vukovic et al., 2010). So far, the possible effects of neglecting neutron energy change on measurement uncertainty, when H*n (10) is measured inside the maze, have not yet been studied. Therefore, this work attempted to evaluate the predictions regarding the measurement uncertainties caused by ignoring neutron energy change inside the maze and also at the patient table. 25
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Percentage Depth Dose (PDD) and dose profile, in electron-photon mode for 4 × 4 cm2, 10 × 10 cm2 and 20 × 20 cm2 field sizes (Chegeni et al., 2018). Some of the characteristics of this model are given briefly below: PDDs had been calculated inside a water phantom along the central axis up to 30 cm depth. Here, the depth of maximum dose (dmax) was found 3.4 cm. In addition, the dose profiles had been calculated in offaxis direction for the mentioned field sizes. The calculated data had been benchmarked with measured data using γ-index function, a useful tool for comparing two dose distributions (for more information about γ-index function, please see the Low et al., 1998 paper). In this model, energy and spatial distribution of electrons impinging the target are Gaussian (mean energy = 18.3 MeV, energy FWHM = 1.2 MeV and spatial FWHM = 0.14 cm). In addition, the number of electrons needed to deliver 1 Gy photon dose to dmax, in the standard condition (Field Size (FS): 10 × 10 cm2, SSD = 100 cm), is 1.26 × 1015 electrons per Gy, which is comparable with 1.52 × 1015 electrons per Gy reported in the literature (Martinez-Ovalle et al., 2011). This factor is multiplied by calculated φn and H*n (10) to report these values according to 1 Gy photon dose delivered to dmax (MCNP code normalizes the output to one particle source). The neutron source strength for this model had been found 1.38 × 1012 n Gy−1 (for closed jaws) which is similar to 1.2 × 1012 n Gy−1 and 1.3 × 1012 n Gy−1 reported in other studies (Mao et al., 1997; Mesbahi et al., 2010). These differences seem reasonable as the type of MC code, cross-section libraries for photoneutron production, and simulated geometry of LINAC head differ from model to model. Thus, we were assured that this model is reliable for evaluating photoneutron contamination. The ground plan of the simulated RT room is shown in Fig. 2, where the LINAC has been located (Golestan Hospital, Ahvaz, Iran). The concrete walls had the density of 2.35 g/cm3 which were constructed from: 49.83% oxygen, 31.58% silicon, 8.26% calcium, 4.56% aluminum, 1.92% potassium, 1.71% sodium, 1.22% iron and 0.92% hydrogen (all in terms of their weight percentages). To consider neutron transmission in the simulations, neutron mode was added to electron-photon mode. Also, the fourth parameter of photon physics card was set in the bias state (PHYS:P 40 j j 1 j j) to be focused on photoneutron production. To account all available interactions between different particles into the MC simulation, cross-section libraries of ENDF/B-V2.0, MCPLIB04 and EL03 were used for neutron, photon, and electron, respectively. These interactions included; absorption, dispersion, and capture for neutron, Rayleigh and Compton scattering, pair production, and the photoelectric effect for photon, and finally elastic and inelastic collision, Bremsstrahlung, and annihilation for electron and positron. Additionally, cross-section libraries of LA150u, KAREI01u and CNDC01u were applied to simulate photoneutron production in the LINAC head.
Table 2 Position of detectors located at the patient table (detectors 1–8) and inside the maze (detectors 9–12). Detector number
X (cm)
Y (cm)
Z (cm)
1 2 3 4 5 6 7 8 9 10 11 12
0 0 0 0 0 0 0 0 360 120 − 120 − 360
0 − 10 − 20 − 30 − 40 − 50 − 60 − 70 − 620 − 620 − 620 − 620
0 0 0 0 0 0 0 0 0 0 0 0
Regarding the mentioned limitations for neutron dosimeters, MC studies are more accurate for the evaluation of neutron contamination in RT, especially for neutron dosimetry in the patient's organs (Khabaz et al., 2018; Mohammadi et al., 2014). Some of the MC studies overlooked RT room in the simulations (Brkić et al., 2016; Frankl and Macián-Juan, 2015; Rezaian et al., 2017) while concrete walls of RT room can scatter the neutrons toward the patient table. Recently, the change in neutron spectrum, when neutrons are scattered from walls of vault, has been introduced as an interesting subject for further investigation (Brkić et al., 2016). However, the effect of neglecting the RT room in MC simulations on neutron spectra and H*n (10) has not yet been thoroughly studied. Accordingly, the next aim of this work is to evaluate the change in Ēn , φn and H*n (10) at the patient table when a typical RT room (including floor, ceiling, and walls made of concrete) is removed from MC simulations. 2. Materials and methods This study aims to present key points for MC and experimental studies on photoneutron contamination in RT. For this purpose, the photoneutron production in the head of Varian Clinac 2100 C/D 18 MV (California, Varian Medical Systems, USA) was simulated using MCNPX® (2.6.0) code (Pelowitz, 2008). The main parts of this LINAC has been modeled according to the available manufacture data and literature (Alem-Bezoubiri et al., 2014). Cross-section of simulated LINAC head is shown in Fig. 1. The model consists of target, beryllium window, primary collimator, flattening filter, ion chamber, mirror, secondary collimator, jaws, and upper circle (Table 1). The model had already been verified with measured data,
Table 3 Neutron mean energy (Ēn) at different positions and field sizes with considering the concrete walls in MC simulations (CW) and without (No CW). Position
Ē n (MeV) 0 × 0 cm2
1 2 3 4 5 6 7 8 9 10 11 12
4 × 4 cm2
6 × 6 cm2
10 × 10 cm2
20 × 20 cm2
40 × 40 cm2
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
0.283 0.286 0.287 0.297 0.320 0.343 0.318 0.307 0.172 0.061 0.036 0.028
0.496 0.498 0.510 0.540 0.591 0.640 0.632 0.637 – – – –
0.542 0.368 0.361 0.345 0.341 0.355 0.327 0.313 0.170 0.061 0.036 0.028
0.781 0.576 0.580 0.575 0.591 0.635 0.623 0.629 – – – –
0.597 0.398 0.386 0.366 0.353 0.360 0.332 0.317 0.173 0.061 0.037 0.028
0.820 0.607 0.604 0.594 0.600 0.635 0.624 0.629 – – – –
0.647 0.454 0.419 0.400 0.377 0.371 0.343 0.323 0.171 0.061 0.035 0.028
0.846 0.659 0.639 0.630 0.628 0.646 0.634 0.634 – – – –
0.687 0.643 0.473 0.440 0.416 0.398 0.367 0.339 0.176 0.062 0.037 0.029
0.859 0.839 0.703 0.687 0.681 0.684 0.670 0.661 – – – –
0.738 0.750 0.738 0.433 0.360 0.331 0.308 0.292 0.184 0.065 0.037 0.031
0.938 0.951 0.980 0.768 0.696 0.677 0.670 0.666 – – – –
26
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Table 4 Neutron fluence (φn ) at different positions and field sizes with considering the concrete walls in MC simulations (CW) and without (No CW). φn (107 n cm−2 Gy−1)
Position
0 × 0 cm2
1 2 3 4 5 6 7 8 9 10 11 12
4 × 4 cm2
6 × 6 cm2
10 × 10 cm2
20 × 20 cm2
40 × 40 cm2
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
0.834 0.833 0.821 0.812 0.814 0.816 0.765 0.729 0.211 0.090 0.038 0.024
0.386 0.382 0.370 0.362 0.362 0.362 0.311 0.277 – – – –
1.275 1.011 0.974 0.923 0.883 0.864 0.800 0.755 0.213 0.090 0.038 0.023
0.824 0.559 0.522 0.469 0.426 0.405 0.342 0.298 – – – –
1.452 1.087 1.032 0.970 0.913 0.879 0.813 0.763 0.210 0.090 0.038 0.023
0.999 0.634 0.578 0.514 0.455 0.421 0.355 0.307 – – – –
1.681 1.216 1.099 1.029 0.954 0.900 0.830 0.772 0.211 0.089 0.038 0.023
1.227 0.762 0.645 0.573 0.497 0.441 0.371 0.315 – – – –
1.923 1.640 1.141 1.035 0.955 0.890 0.819 0.757 0.204 0.087 0.036 0.023
1.478 1.196 0.693 0.591 0.508 0.444 0.373 0.314 – – – –
1.572 1.586 1.369 0.747 0.661 0.622 0.589 0.562 0.178 0.076 0.032 0.020
1.183 1.197 0.982 0.359 0.272 0.232 0.199 0.175 – – – –
Table 5 Neutron ambient dose equivalent (Hn* (10)) at different positions and field sizes with considering the concrete walls in MC simulations (CW) and without (No CW). [H*n (10)] (mSv Gy−1)
Position
0 × 0 cm2
1 2 3 4 5 6 7 8 9 10 11 12
4 × 4 cm2
6 × 6 cm2
10 × 10 cm2
40 × 40 cm2
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
CW
No CW
1.187 1.184 1.170 1.170 1.214 1.263 1.115 1.025 0.171 0.037 0.011 0.006
0.907 0.898 0.882 0.884 0.925 0.974 0.827 0.738 – – – –
2.623 1.694 1.605 1.475 1.385 1.374 1.201 1.085 0.171 0.037 0.011 0.006
2.331 1.403 1.316 1.187 1.098 1.085 0.911 0.797 – – – –
3.192 1.919 1.774 1.608 1.467 1.417 1.237 1.109 0.171 0.037 0.011 0.006
2.902 1.629 1.486 1.320 1.179 1.128 0.948 0.820 – – – –
3.933 2.329 1.994 1.804 1.607 1.484 1.293 1.141 0.170 0.037 0.011 0.006
3.637 2.037 1.703 1.514 1.316 1.192 1.001 0.849 – – – –
4.751 3.850 2.212 1.920 1.691 1.522 1.323 1.153 0.168 0.036 0.011 0.006
4.459 3.558 1.920 1.627 1.398 1.230 1.032 0.865 – – – –
3.989 4.051 3.384 1.299 1.026 0.910 0.816 0.747 0.150 0.032 0.010 0.005
3.720 3.782 3.117 1.034 0.762 0.645 0.552 0.485 – – – –
neglecting the neutron energy change can predictably lead to more measurement uncertainty (Brkić et al., 2016). To investigate the credibility of this prediction, the expected number of alpha particles, recorded by this type of dosimeters, was estimated in various points inside the maze (detectors 9–12). These calculations were performed by multiplying the cross-section of 10B (n, α) 7Li reaction by the neutron spectra, arbitrary in 20 × 20 cm2 FS. In fact, the cross-section of 10B (n, α) 7Li was divided to several energy segments. These parts were chosen to fit curves by the least square method (each segment had its own equation). The midpoint of each logarithmic energy bin was inserted in the obtained equations and was then multiplied by the corresponding neutron fluence. Next, the expected number of alpha particles and H*n (10) were normalized to their maximum values. Subsequently, in each point, the H*n (10) ratio was compared to alpha particles ratio. These comparisons estimate the possible uncertainty in dosimetry (for dosimeters that use 10B) when the change in neutron energy inside the maze is ignored. Finally, by removing concrete walls from simulations, Ēn , φn and H*n (10) were calculated at the patient table (detectors 1–8). As a final point, to achieve valid confidence intervals for MC calculations, at least 5 × 108 electron histories were traced in each simulation. Relative statistical uncertainty for total φn (also H*n (10)) was kept < 1% at the patient table and < 2% inside the maze. Regarding the neutron spectra, scaled by % of total fluence, the average absolute uncertainty for normalized fluence was < 0.1%. In all calculations, to short the run time of programs, energy cut-offs for electron and photon were chosen as 6 MeV (less than threshold energy of photonuclear
Using F5 tally (point detector), neutron spectra were calculated in different field sizes (0 × 0, 4 × 4, 6 × 6, 10 × 10, 20 × 20 and 40 × 40, all in cm2) for several points at the patient table and inside the maze (Table 2 and Fig. 2). Besides, using F5 tally, DF card and ICRP 74 conversion factors (1996), H*n (10) was calculated for all the detectors. Using obtained spectra and according to Eq. (1), Ēn was computed for each detector. 100
Ēn =
20 × 20 cm2
∑a = 1 Ea.ϕ(Ea) 100
∑a = 1 ϕ(Ea)
(1) −9
where energy bin of [10 , 10] MeV is divided into 100 bins logarithmically. Φ (Ea) is the calculated neutron fluence related to the ath energy bin and Ea is the middle point in the ath energy bin. To investigate the amount of uncertainty expected during neutron dosimetry at the IC, only due to neglecting the change in Ēn with FS, the calculated values for φn and H*n (10) at the IC were normalized to their related maximum value (i.e. in 20 ×20 cm2 FS). A similar analysis was performed also along the patient table to determine the amount of possible measurement uncertainty caused only by neglecting the change in Ēn with detector position. In this regard, the calculated values for φn and H*n (10) along the patient table (arbitrary in 20 × 20 cm2 FS) were normalized to their related maximum value (i.e. to IC). Finally, in each case, the φn ratio was compared with the H*n (10) ratio and introduced as expected uncertainty in measurements. Detectors that covered by a layer of 10B as converter are commonly used for neutron dosimetry in different fields, including RT. As mentioned before, if H*n (10) is measured by these types of detectors, 27
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Fig. 4. Neutron fluence (φn ) and ambient dose equivalent (H*n (10)) normalized to their maximum values: (A) for different field sizes at the IC, (B) for different points at the patient table.
Fig. 3. Neutron spectra: (A) for different field sizes at the isocenter, (B) for different positions at the patient table.
reactions in the LINAC head). The energy cut-off for neutron remained the MCNP default (0 MeV).
table. In addition, it is perceptible that φn inside the maze does not depend on FS and depends only on the detector position. By changing the position from 9 to 12, φn decreases nearly 0.9 times (neutrons were attenuated by concrete walls and fewer neutrons reached the entrance). According to Table 5, H*n (10) for different field sizes at the IC ranged from 1.187 to 4.751 mSv·Gy−1 which was at least 6 times higher than H*n (10) inside the maze. This value becomes more significant (83–333 mSv) when 70 Gy photon dose is delivered to a tumor in a complete treatment. The issue will be more critical for modern RT techniques which use more monitor units for treatment, in turn might increase the risk of secondary cancer (Brkić et al., 2016). So regarding the significance of radiation protection in RT, it is important to evaluate photoneutron dose accurately inside the RT room. It is also evident that with moving off axis from IC to 70 cm distant point, H*n (10) drops to 1.043 mSv·Gy−1 averagely on all field sizes. Table 5 also shows that H*n (10) inside the maze is independent of FS and decreases only by moving away from the end of maze (detector 9), a phenomenon that was previously reported using a simplified MC model (Mohammadi et al., 2014; Zabihzadeh et al., 2009). Dependence of neutron energy spectra to FS (Fig. 3A) and off-axis distance (Fig. 3B) was analyzed at the patient table with the presence of RT room in the simulations. Accordingly, one can observe that the neutron spectra are characterized by two peaks. The first one is located in the thermal energy region, i.e. [10−9, 10−7] MeV, and the second one in the fast energy region, i.e. [10−1, 10] MeV. A flat intermediate
3. Results and discussion At the first step, Ēn was calculated for several field sizes at the patient table (with/without considering vault walls) and inside the maze (Table 3). Table 3 shows that Ēn for different field sizes at the IC ranges from 0.283 MeV to 0.738 MeV in the presence of concrete walls. These results are in fairly good agreement with recent research (Alem-Bezoubiri et al., 2014). After removing concrete walls from MC simulations, Ēn shifted toward the higher energies averagely about 0.2 MeV (0.496 – 0.938 MeV). This effect was more evident for detectors located far away from IC, so as for detector 8 (70 cm of IC), Ēn shifted toward the higher energy up to 0.4 MeV in some cases (FS: 40 × 40 cm2). Next, φn and H*n (10) were analyzed for different field sizes at the patient table and inside the maze. Results are summarized respectively in Tables 4 and 5 which comply well with the relevant data reported in other studies (Alem-Bezoubiri et al., 2014; Kralik et al., 2008). These tables also present the data related to the circumstances where the RT room was removed from MC simulations. In consort with the recent research (Brkić et al., 2016), Table 4 indicates that φn at the IC increases with FS and reaches maximum in 20 × 20 cm2 while it decreases with moving off axis at the patient 28
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
2008). Regarding Fig. 3A, it is evident that neutron energy spectra and consequently Ēn at the IC strongly depends on FS. When FS changes from 0 × 0 cm2 to 40 × 40 cm2, Ēn at the IC will be 2.6 times (Table 3). Moreover, as Fig. 3B illustrates if only the position at the patient table changes, using the same FS, Ēn will shift to lower energies. Herein for detector 8 (70 cm of IC), Table 3 shows nearly 50% reduction (related to IC) in Ēn . Awareness of neutron mean energy is necessary to choose flux-todose conversion factors correctly in neutron dosimetry (ICRP, 1996). If the change in neutron energy with FS is not taken into account when H*n (10) is measured at the IC, up to 19% measurement uncertainty can be expected (Fig. 4A). This finding is well confirmed by Brkić et al.’s work (2016) which introduced 20% measurement uncertainty in this case. This type of measurement uncertainty can be expected when fluxes are converted to doses using constant factors and overlooking the neutron energy change (i.e. with changing FS at IC, φn ratio was assumed as H*n (10) ratio). Besides, when H*n (10) is measured at different points on the patient table (using the same FS) and the shift in neutron energy is not taken into account, the measurement uncertainty can be even > 10% (Fig. 4B) in most detectors, as it was already reported by Brkić et al. (2016). The above findings interested us to perform similar analysis for detectors located inside the maze. The dependence of neutron energy spectra on FS and distance inside the maze were summarized in Figs. 5A and 5B, respectively. According to Fig. 5A, it can be seen that for an arbitrary point inside the maze (detector 11), the change in FS does not have significant effects on neutron energy spectra. Therefore, if the change in neutron energy with FS is ignored during neutron dosimetry inside the maze, the measurement uncertainty will be less pronounced (up to 7%, Fig. 6). On the other hand, Fig. 5B shows that the change in detector position from 9 to 12, using the same FS, lead to a significant change in neutron energy (also see Table 3). So if this energy shift is not considered when H*n (10) is measured at different positions inside the maze, there will be nearly 21% uncertainty in measurements (Fig. 7, difference between H*n (10) and φn ). This type of uncertainty can even approach 35% (Fig. 7, difference between H*n (10) and the expected number of alpha particles) if measurement is performed by dosimeters use 10B as converter. As Brkić et al. (2016) had predicted, the large energy dependence of 10B (n, α) 7Li cross-section can lead to more measurement uncertainty when the change in neutron energy is neglected. Finally, with/without considering the RT room in MC simulations, neutron energy spectra were compared together at the IC in 0 × 0 cm2 FS, illustrated in Fig. 8. Accordingly, one can observe major changes in the neutron spectra with removing the RT room from MC simulations. The figure also demonstrates when the concrete walls were not accounted for MC simulations, there were only fast neutrons in the spectrum. This means that all the neutrons produced in the LINAC head lie in the fast region, an occurrence that was also encountered in an earlier study (Mohammadi et al., 2015). In consort with Pena et al. (2005) study, Fig. 8 illustrates that with considering the RT room in MC simulations, the fluence of fast neutrons increases at the IC. It seems rational since some of the fast neutrons which were produced in the LINAC head can be scattered elastically from the concrete walls toward the patient table and consequently can increase the fluence of fast neutrons at the IC. Removing concrete walls from MC simulations had another effect on the shape of neutron spectrum. Neutrons could also be scattered inelastically for several times from concrete walls toward the patient table and subsequently appeared as thermal and epithermal neutrons in the spectrum (Fig. 8). It is noteworthy that with increasing distance from IC, the amount of thermal and epithermal neutrons remain practically the same while the fast neutrons are reduced (Table 4), a phenomenon which was also reported previously by Vega-Carrillo and Baltazar-
Fig. 5. Neutron spectra inside the maze: (A) for different field sizes, (B) for different positions.
Fig. 6. Neutron fluence (φn ) and ambient dose equivalent (H*n (10)) at detector 11 for different field sizes normalized to their maximum values (i.e. to 0 × 0 cm2).
energy region, i.e. [10−7, 10−1] MeV, related to epithermal neutrons is located in the middle. The shapes of the calculated spectra correspond well with those measured before (Esposito et al., 2008; Kralik et al.,
29
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Fig. 7. Expected number of alpha particles, neutron fluence (φn ) and neutron ambient dose equivalent (Hn* (10)) inside the maze normalized to their maximum values (i.e. to detector 9).
Fig. 9. Change in neutron ambient dose equivalent (H*n (10)) with removing concrete walls from MC simulations: (A) at the IC, (B) at 70 cm of IC.
Fig. 8. Neuron spectra at the IC with and without considering the concrete walls in MC simulations.
photoneutron dose in RT. At first, when H*n (10) is measured at the patient table or inside the maze and the change in neutron energy (with FS or detector location) is ignored, there will be nearly up to 20% measurement uncertainty. This type of uncertainty can approach 35% if H*n (10) is measured by dosimeters whose response depends strongly on energy (e.g. CR-39 covered by layer of 10B as converter). So, in these cases, neglecting the change in neutron energy can threaten the credibility of measured data and one should attend to this energy change in order to reduce measurement uncertainty to the possible minimum. For this purpose, MC calculations can be used as a correction for measured data where neglecting the neutron energy change lead to large differences between φn ratio and H*n (10) ratio (Figs. 4, 6 and 7). Secondly, it was found when the RT room is not considered in MC simulations, Ēn is overestimated by 0.2–0.4 MeV at the patient table. φn at the IC will be underestimated by 23–54% for different field sizes that cause H*n (10) to be miscalculated up to 24%. This finding informs researchers that for accurate evaluation of H*n (10) at the patient table, simulating RT room is an effective parameter in MC studies. Although the calculated values for H*n (10) in the air (at the patient table and inside the maze) can be used as a good parameter for radiation protection aspects, it is far better to consider a more realistic situation. With considering this point and using an anthropomorphic phantom in MC simulations, our future study will concentrate on unwanted H*n (10) in the patient's tissues/organs (risk of secondary cancer) for different RT techniques.
Raigosa (2011). So, if concrete walls are overlooked in MC simulations, the contribution of the scattered neutrons in spectra is ignored and consequently φn at the IC is underestimated by 23–54% for different field sizes (Table 4). In another MC study, Alem-Bezoubiri et al. (2014) showed for 10 × 10 cm2 FS, the contributions of the scattered neutrons in total φn at the IC and 40 cm of IC were 27% and 58% respectively which are similar to the corresponded values of 27% and 48% in this project (Table 4, reduction in φn with removing the RT room). This underestimation will be even higher when moving off axis at the patient table, so as for detector located at 70 cm from IC, it approaches 69% in some field sizes (40 × 40 cm2). It seems reasonable as by taking distance from IC, the contribution of thermal and epithermal neutrons in the spectrum will increase (Fig. 3B). As this reduction occurs mainly in thermal and epithermal energy region (Fig. 8), where flux-to-dose conversion factors have low values (compared to fast region), H*n (10) is miscalculated on average 0.286 mSv Gy−1 at the patient table (Table 5). This miscalculation cause to maximum 24% underestimation in the evaluation of H*n (10) at the IC (Fig. 9A) that can reach even 35% for detector located at 70 cm of IC (Fig. 9B). 4. Conclusion This study discussed critical issues (via MC simulation) and suggested precautions to be considered for the accurate evaluation of 30
Applied Radiation and Isotopes 145 (2019) 24–31
A.H. Karimi et al.
Conflict of interest
Phys. 72, 524–529. Martinez-Ovalle, S.A., Barquero, R., Gomez-Ros, J.M., Lallena, A.M., 2011. Neutron dose equivalent and neutron spectra in tissue for clinical linacs operating at 15, 18 and 20 MV. Radiat. Prot. Dosim. 147, 498–511. McGinley, P., Wood, M., Mills, M., Rodriguez, R., 1976. Dose levels due to neutrons in the vicinity of high-energy medical accelerators. Med. Phys. 3, 397–402. Mesbahi, A., Ghiasi, H., Mahdavi, S.R., 2010. Photoneutron and capture gamma dose equivalent for different room and maze layouts in radiation therapy. Radiat. Prot. Dosim. 140, 242–249. Mohammadi, N., Miri-Hakimabad, H., Rafat-Motavlli, L., Akbari, F., Abdollahi, S., 2015. Neutron spectrometry and determination of neutron contamination around the 15 MV Siemens Primus LINAC. J. Radioanal. Nucl. Chem. 304, 1001–1008. Mohammadi, N., Miri-Hakimabad, S.H., Rafat-Motavalli, L., 2014. A Monte Carlo study for photoneutron dose estimations around the high-energy linacs. J. Biomed. Phys. Eng. 4, 127–140. NCRP60, 1984. Neutron Contamination from Medical Electron Accelerators: Recommendations of the National Council on Radiation Protection and Measurements, Ncrp Reprt 60, Bethesda, MD, USA. Pelowitz, D.B., 2008. MCNPX User's Manual, Version 2.6.0. Los Alamos National Laboratory, Los Alamos. Pena, J., Franco, L., Gomez, F., Iglesias, A., Pardo, J., Pombar, M., 2005. Monte Carlo study of siemens PRIMUS photoneutron production. Phys. Med. Biol. 50, 5921–5933. Rezaian, A., Nedaie, H.A., Banaee, N., 2017. Measurement of neutron dose in the compensator IMRT treatment. Appl. Radiat. Isot. 128, 136–141. Shahbazi-Gahrouei, D., Khosravi, M., Jabbari, K., Gheisari, R., 2012. Measurement of photoneutron dose in the linear accelerator at the radiation therapy section of SeyedAl-Shohada hospital, Isfahan, Iran. J. Isfahan Med. Sch. 29. Vega-Carrillo, H.R., Baltazar-Raigosa, A., 2011. Photoneutron spectra around an 18 MV linac. J. Radioanal. Nucl. Chem. 287, 323–327. Vega-Carrillo, H.R., de Leon-Martinez, H.A., Rivera-Perez, E., Luis Benites-Rengifo, J., Gallego, E., Lorente, A., 2015. Induced radioisotopes in a linac treatment hall. Appl. Radiat. Isot. 102, 103–108. Vega-Carrillo, H.R., Hernández-Almaraz, B., Hernández-Dávila, V., Ortíz-Hernández, A., 2009. Neutron spectrum and doses in a 18 MV LINAC. J. Radioanal. Nucl. Chem. 283, 261–265. Vega-Carrillo, H.R., Martínez-Ovalle, S.A., Lallena, A.M., Mercado, G.A., Benites-Rengifo, J.L., 2012. Neutron and photon spectra in LINACs. Appl. Radiat. Isot. 71, 75–80. Vukovic, B., Faj, D., Poje, M., Varga, M., Radolic, V., Miklavcic, I., Ivkovic, A., Planinic, J., 2010. A neutron track etch detector for electron linear accelerators in radiotherapy. Radiol. Oncol. 44, 62–66. Zabihinpoor, S., Hasheminia, M., 2011. Calculation of neutron contamination from medical linear accelerator in treatment room. Adv. Stud. Theor. Phys. 5, 421–428. Zabihzadeh, M., Ay, M.R., Allahverdi, M., Mesbahi, A., Mahdavi, S.R., Shahriari, M., 2009. Monte Carlo estimation of photoneutrons contamination from high-energy Xray medical accelerators in treatment room and maze: a simplified model. Radiat. Prot. Dosim. 135, 21–32.
The authors have no conflicts of interest. References Alem-Bezoubiri, A., Bezoubiri, F., Badreddine, A., Mazrou, H., Lounis-Mokrani, Z., 2014. Monte Carlo estimation of photoneutrons spectra and dose equivalent around an 18 MV medical linear accelerator. Radiat. Phys. Chem. 97, 381–392. Brkić, H., Ivković, A., Kasabašić, M., Poje Sovilj, M., Jurković, S., Štimac, D., Rubin, O., Faj, D., 2016. The influence of field size and off-axis distance on photoneutron spectra of the 18 MV Siemens Oncor linear accelerator beam. Radiat. Meas. 93, 28–34. Chegeni, N., Karimi, A.H., Jabbari, I., Arvandi, S., 2018. Photoneutron dose estimation in grid therapy using an anthropomorphic phantom: a Monte Carlo study. J. Med. Signal. Sens. 8, 175–183. Chibani, O., Ma, C.M.C., 2003. Photonuclear dose calculations for high‐energy photon beams from Siemens and Varian linacs. Med. Phys. 30, 1990–2000. Das, I.J., Kase, K.R., 1992. Higher energy: is it necessary, is it worth the cost for radiation oncology? Med. Phys. 19, 917–925. Donadille, L., Trompier, F., Robbes, I., Derreumaux, S., Mantione, J., Asselineau, B., Amgarou, K., Martin, A., Bottollier-Depois, J.F., Queinnec, F., Aubert, B., Clairand, I., 2008. Radiation protection of workers associated with secondary neutrons produced by medical linear accelerators. Radiat. Meas. 43, 939–943. Esposito, A., Bedogni, R., Lembo, L., Morelli, M., 2008. Determination of the neutron spectra around an 18MV medical LINAC with a passive Bonner sphere spectrometer based on gold foils and TLD pairs. Radiat. Meas. 43, 1038–1043. Frankl, M., Macián-Juan, R., 2015. Monte Carlo simulation of secondary radiation exposure from high-energy photon therapy using an anthropomorphic phantom. Radiat. Prot. Dosim. 168, 537–545. ICRP, 1996. Conversion coefficients for use in radiological protection against external radiation. Ann. ICRP 26, 3–4. Khabaz, R., Boodaghi, R., Benam, M.R., Zanganeh, V., 2018. Estimation of photoneutron dosimetric characteristics in tissues/organs using an improved simple model of linac head. Appl. Radiat. Isot. 133, 88–94. Khaled, N., Ghanim, E., Shinashin, K., El-Sersy, A., 2014. Effect of X-ray energies on induced photo-neutron doses. Radiat. Eff. Defects Solids 169, 239–248. Khosravi, M., Shahbazi-Gahrouei, D., Jabbari, K., Nasri-Nasrabadi, M., BaradaranGhahfarokhi, M., Siavashpour, Z., Gheisari, R., Amiri, B., 2013. Photoneutron contamination from an 18 MV saturne medical linear accelerator in the treatment room. Radiat. Prot. Dosim. 156, 356–363. Kralik, M., Turek, K., Vondracek, V., 2008. Spectra of photoneutrons produced by highenergy X-ray radiotherapy linacs. Radiat. Prot. Dosim. 132, 13–17. Low, D.A., Harms, W.B., Mutic, S., Purdy, J.A., 1998. A technique for the quantitative evaluation of dose distributions. Med. Phys. 25, 656–661. Mao, X., Kase, K., Liu, J., Nelson, W., Kleck, J., Johnsen, S., 1997. Neutron sources in the Varian Clinac 2100C/2300C medical accelerator calculated by the EGS4 code. Health
31