NEUTRON DOSE CALCULATION AT THE MAZE ENTRANCE OFMEDICAL LINEAR ACCELERATOR ROOMSR. C. Falcao1,�, A. Facure1 and A. X. Silva2

1Comissao Nacional de Energia Nuclear, R. Gal. Severiano 90, sala 405, 22294-900 Rio de Janeiro,RJ, Brazil2[PEN/COPPE—DNC/EE]CT/UFRJ, Ilha do Fundao, P.O. Box 68509, 21945-970 Rio de Janeiro,RJ, Brazil

Received June 28 2006, amended August 14 2006, accepted August 22 2006

Currently, teletherapy machines of cobalt and caesium are being replaced by linear accelerators. The maximum photonenergy in these machines can vary from 4 to 25 MeV, and one of the great advantages of these equipments is that they do nothave a radioactive source incorporated. High-energy (E> 10 MV) medical linear accelerators offer several physical advan-tages over lower energy ones: the skin dose is lower, the beam is more penetrating, and the scattered dose to tissues outside thetarget volume is smaller. Nevertheless, the contamination of undesirable neutrons in the therapeutic beam, generated by thehigh-energy photons, has become an additional problem as long as patient protection and occupational doses are concerned.The treatment room walls are shielded to attenuate the primary and secondary X-ray fluence, and this shielding is generallyadequate to attenuate the neutrons. However, these neutrons are scattered through the treatment room maze and may result ina radiological problem at the door entrance, a high occupancy area in a radiotherapy facility. In this article, we used MCNPMonte Carlo simulation to calculate neutron doses in the maze of radiotherapy rooms and we suggest an alternative method tothe Kersey semi-empirical model of neutron dose calculation at the entrance of mazes. It was found that this new method fitsbetter measured values found in literature, as well our Monte Carlo simulated ones.

INTRODUCTION

The production of photoneutrons by high-energymedical accelerators is a subject of considerableinterest, and has been extensively treated in variousworks(1–3).

At present, a large number of high-energymachines, with photon energies >10 MeV have comeinto routine clinical use in radiotherapy. The accu-racy of the estimation of neutron doses is of interestto regulatory bodies when applying for license, andit has relevance to both patients and occupationalpeople, having implications in the design and shield-ing of the treatment room and maze(4). Some authorshave reported that measurements outside the accessdoor of the treatment room sometimes show a sig-nificant neutron dose equivalent rate(5).

In Brazil, no measurements of neutron doses,inside and around radiotherapy rooms, are requiredwhen installing and commissioning high-energylinacs. Nevertheless, despite the absence of any data,Brazilian standards and regulations concerning thesemachines require that the kerma in tissue due toneutrons inside the treatment field should not be>1% of the X-ray kerma at the isocentre(6). In ourcountry, when designing the shielding of radiother-apy rooms, neutron doses along the maze are usuallycalculated by the Kersey method(7). According to

this method, the neutron dose equivalent at theroom door is given by

H ¼ H0T

T0

·d0ð Þ2

d1ð Þ210�d2=5

, ð1Þ

where H0 is the dose equivalent due to neutrons (Sv),measured at the isocentre (point 1 in Figure 1), d0 isthe distance (m) from the target to the isocentre, d1 isthe distance (m) from the isocentre to a point at thecentral line of the inner maze entrance (point 2 inFigure 1) and d2 is the distance (m) from the centralline of the maze to the door (distance from point 2 topoint 5 in Figure 1). T/T0 is the ratio between thesmallest and biggest cross-sectional area of the maze.The value of H, calculated in Equation 1 is the doseequivalent from neutrons per absorbed dose ofX rays at the isocentre and is given in units ofSv Gy�1. Equation 1 considers only the contributionof direct neutrons to the dose, which means thatscattered and thermal neutrons are not considered.

McGinley and Butker(8) have tested the Kerseymethod in 13 accelerator rooms, through experi-mental measurements using both activation detec-tors and a neutron rem meter. The relation foundbetween the measured and calculated equivalentdoses varied between 0.82 and 2.30, what suggeststhat it is relevant to propose corrections to theKersey method, mainly because it usually under-estimates neutron doses, which is a problem whendesigning radiotherapy room doors.�Corresponding author: rossanafalcao@uol.com.br

Radiation Protection Dosimetry (2007), Vol. 123, No. 3, pp. 283–287 doi:10.1093/rpd/ncl144Advance Access publication 27 September 2006

� The Author 2006. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

at University of W

innipeg on August 23, 2014

http://rpd.oxfordjournals.org/D

ownloaded from

MATERIALS AND METHODS

The Monte Carlo radiation transport code MCNP(9),version 5, and the Evaluated Nuclear Data File B-VI(ENDF/B-VI) continuous energy neutron cross sec-tion library were employed to perform the calcula-tions of photoneutron transport in radiotherapyrooms. The variance reduction technique known aspoint detector (F5 tally), belonging to the class ofpartially deterministic variance reduction methodsimplemented in MCNP5, was used.

We simulated the photoneutrons spectra gener-ated by linear accelerators of 15, 18 and 25 MeVenergy photons and the primary neutron distributionemployed on the simulations was based on the oneproposed by Tosi et al.(5) The accelerator head wasmodelled as a 10 cm radius tungsten sphere, since inprevious papers(10,11) we have tested successfully thevalidity of this initial neutron spectra and acceleratorhead model.

Differently from Ref. (10), where the room wallswere not considered when simulating neutron spec-tra, in the present work we simulated 14 roomgeometries, where the accelerator was housed:9 rooms of 15 MV accelerators, 3 rooms of 18 MVaccelerators and 2 rooms of 25 MV accelerators.Seven of these rooms had the dimensions of existingones, licensed recently by the National NuclearEnergy Commission of Brazil (CNEN). This simula-tion geometry permits us to calculate not only thecontribution of the direct neutrons, but also thermaland scattered neutrons to the dose.

Figure 1 below shows a schematic diagram of therooms simulated, with the points 1–5 where fluenceand dose were estimated. The room dimensions X, Y,Z and K are described in Table 1.

All the room walls were considered to be madeout of ordinary concrete and the ceiling height variedfrom 3.2 m (rooms 9–14) to 4.0 m (rooms 1–5). In oursimulations, the accelerator model was characterisedby its Q-value, given by the manufacturer and definedas the neutron output per each Gray of X rays deliv-ered at the isocentre. Q-values are shown in Table 2for the most commonly used machines in Brazil.

Neutron fluences, mean energies and dose equiva-lent were simulated at various points inside the treat-ment room, as indicated by the points in Figure 1.All detectors were positioned at 1 m from the floorand simulated neutron doses were then comparedwith the values obtained using the Kersey methoddescribed above. In the following section we presentthese results, discuss the possible reasons for thediscrepancies between simulated and theoretical val-ues, and propose some modifications to the Kerseymethod.

MCNP SIMULATION RESULTS

Neutron fluences and mean energy

In Table 3, the simulated values of neutron fluenceat the five detection points described in Figure 1 areprovided. The fluence values are normalised perneutron emitted from the source.

Figure 1. Geometry of the rooms simulated with MCNP5. Neutron fluence and doses were evaluated at points 1–5.

R. C. FALCAO ET AL.

284

at University of W

innipeg on August 23, 2014

http://rpd.oxfordjournals.org/D

ownloaded from

In a previous article(12), we have simulated scat-tered and thermal neutron fluences inside radio-therapy rooms. For the scattered fluence, we foundvalues which are around three times greater than theones predicted theoretically by McGinley(7), whereasthe simulated thermal fluence is about seven timesgreater than the theoretical one. In the present work,for example, the normalised total fluence calculatedin point 2 of room 1 is of 2.61� 10�6 neutron, show-ing that the simulated value is �1.5 times greater.

Also according to our simulations, neutron meanenergies at the maze entrance (point 3 in Figure 1)were found to be between 25 and 100 keV, whereasat the end of the maze, at the room door position(point 5 in Figure 1), these energies fall from 5 to25 keV. These results agree with the ones found byLin et al.(13), who measured, using BF3 proportionalcounters, neutron mean energies in the intervalfrom 50 to 65 keV at the maze entrance and from17 to 20 keV, at room doors of 10, 15 and 18 MVaccelerators.

Neutron doses

Neutron doses were calculated internally by MCNP,using the conversion factors of ICRP Publica-tion 74(14), at the same points where fluences were

estimated. In Table 4 we illustrate values of simu-lated neutron dose equivalent, per absorbed dose ofX rays at the isocentre, for some of the most com-mon linear accelerators operating in Brazil. Thesevalues were also compared with the ones given bythe manufacturers and the agreement found, exceptfor the SIEMENS machine, was very good. In thepresent simulation we found dose values which are25% higher than the ones we found previously(10),surely because now we considered the room walls inour simulations, taking scattered particles intoaccount.

After calculating neutron doses at the isocentre,we calculated neutron doses at the room doors. InTable 5 we compare the simulated values of neutrondose at the door position (point 5 in Figure 1), withthe ones calculated by the Kersey method, for the14 rooms considered in the present work. Neutrondose equivalent, for each Gray of X rays at theisocentre, are given in units of Sv Gy�1, and thevalues of H0 used in the calculations were the onespresented in Table 4. It can be observed that there isa great discrepancy between the values found byour simulations and the ones calculated using theKersey method, no matter the mega-voltage of themachines: 15, 18 or 25 MV, showing that the Kerseycalculation method seem to underestimate the dosesat the door. In the following section, we discusspossible reasons for this discrepancy and suggestmodifications to the existing equation to calculateneutron doses at radiotherapy room doors.

AN ALTERANTIVE METHOD FORNEUTRON DOSE CALCULATION

One of the possible reasons for the discrepancybetween simulated and calculated values of neutrondoses may be that the Kersey method does not

Table 1. Accelerator model and dimensions of the rooms simulated with MCNP.

Equipment: Model/energy X (m) Y (m) K (m) W (m)

1: Varian 1800 15 MV 10.00 10.00 2.00 5.002: Varian 1800 15 MV 10.00 10.00 1.50 6.503: Varian 1800 15 MV 10.00 10.00 2.00 6.504: Varian 1800 15 MV 10.00 10.00 2.50 6.505: Varian 1800 15 MV 10.00 10.00 2.00 8.506: GE Saturne 18 MV 8.00 4.70 1.50 3.207: GE Saturne 43 25 MV 10.00 10.00 2.00 5.008: Varian 1800 15 MV 8.00 4.70 1.50 3.209: GE Saturne 15 MV 8.50 9.50 1.50 5.5010: SIEMENS MD 15 MV 8.00 10.50 1.50 6.5011: SIEMENSMD2 15 MV 8.50 10.00 1.80 6.5012: Varian 1800 18 MV 10.00 8.50 1.50 6.0013: GE Saturne 18 MV 9.50 8.50 1.50 5.5014: GE Saturne 43 25 MV 8.50 11.50 1.50 7.00

The parameters X, Y, K and W are described in Figure 1

Table 2. Neutron source strength (Q) for some medicalaccelerators(7).

Manufacturer Model StatedMeV

Q (neutronsper Gy)

GE Saturne 43 25 2.4� 1012

Siemens KD 20 0.92� 1012

Varian 1800 18 1.22� 1012

Varian 1800 15 0.76� 1012

NEUTRON DOSE AT MAZE ENTRANCE

285

at University of W

innipeg on August 23, 2014

http://rpd.oxfordjournals.org/D

ownloaded from

consider, as stated in the introduction, the contribu-tion of scattered neutrons to the dose.

Another approximation of the Kersey methodthat may not be valid is that the tenth-value distance

(TVD) for neutron doses along the maze is fixed as5 m (see Equation 1), no matter its shape or wallsconstitution. We then propose that these TVD val-ues shall not be fixed, but vary with the maze design.In order to test this hypotheses, we estimated,through MCNP simulations, the dose TVD for the14 room mazes simulated in the present work. More-over, we also simulated rooms of fixed dimensions,housing 15, 18 and 25 MV accelerators, varying onlythe maze cross-sectional area, from 2 to 10 m2. Weobserved that the neutron dose TVD varies linearlywith the cross-sectional area of the maze, accordingto the following equation:

TVD ¼ 1:7 þ 0:55 CSð Þ; ð2Þ

where the TVD distance is given in metres and CS standsfor the maze cross-sectional area, and is given in m2.

Based on these two observations, we have cal-culated neutron doses at the maze entrance of the14 radiotherapy rooms described above, using thefollowing method:

First we estimated, using the equations proposedby McGinley(7), the total (directþ scattered þthermal) neutron fluence at the inner entranceof the room maze (point 2 in Figure 1).

Table 5. Neutron dose equivalent, at the door position, perabsorbed dose of X rays at the isocentre in units of Sv Gy�1,simulated by MCNP and calculated by the Kersey method,

for the 14 rooms described above.

Room Calculated Simulated Rate calculated/simulated

1 1.22� 10�6 3.67� 10�6 3.012 6.10� 10�7 1.29� 10�6 2.123 7.80� 10�7 1.64� 10�6 2.104 4.11� 10�6 2.40� 10�6 0.585 1.90� 10�7 8.61� 10�7 4.536 4.44� 10�6 8.13� 10�6 1.837 5.13� 10�6 1.47� 10�5 2.878 5.03� 10�6 5.45� 10�6 1.089 8.10� 10�7 2.08� 10�6 2.5710 1.54� 10�7 6.35� 10�7 4.1211 7.80� 10�7 2.50� 10�6 3.2112 1.68� 10�5 2.57� 10�5 1.5313 1.69� 10�5 3.03� 10�5 1.7914 1.37� 10�6 6.01� 10�6 4.39

Table 3. Simulated fluence values for the 14 rooms described above, at the five positions defined in Figure 1.

Room Position 1 Position 2 Position 3 Position 4 Position 5

1 1.17� 10�5 3.98� 10�6 3.41� 10�6 6.63� 10�7 2.60� 10�7

2 1.17� 10�5 3.78� 10�6 3.17� 10�6 3.55� 10�7 1.31� 10�7

3 1.20� 10�5 3.53� 10�6 2.85� 10�6 4.01� 10�7 1.64� 10�7

4 1.23� 10�5 3.22� 10�6 2.44� 10�6 4.30� 10�7 2.86� 10�7

5 1.22� 10�5 2.81� 10�6 1.46� 10�6 1.97� 10�7 8.43� 10�8

6 1.92� 10�5 8.73� 10�6 5.12� 10�6 1.31� 10�7 5.69� 10�7

7 1.17� 10�5 3.98� 10�6 3.42� 10�6 6.64� 10�7 2.61� 10�7

8 1.93� 10�5 8.69� 10�6 5.15� 10�6 1.29� 10�7 5.67� 10�7

9 1.37� 10�5 4.81� 10�6 3.85� 10�6 5.38� 10�7 2.66� 10�7

10 1.43� 10�5 4.57� 10�6 3.58� 10�6 6.51� 10�7 2.07� 10�7

11 1.46� 10�5 4.61� 10�6 3.62� 10�6 6.59� 10�7 2.13� 10�7

12 1.31� 10�5 4.62� 10�6 1.61� 10�6 3.53� 10�7 2.03� 10�7

13 1.30� 10�5 4.65� 10�6 1.64� 10�6 3.59� 10�7 2.10� 10�7

14 1.41� 10�5 4.59� 10�6 3.44� 10�6 6.25� 10�7 1.89� 10�7

Table 4. Neutron dose equivalent at the isocentre (H0) per absorbed dose of X rays in units of mSv Gy�1.

Manufacturer Model Photonenergy

H0 (measured) H0 (simulated) Rate simulated/measured

GE Saturne 43 25 1.38a 1.36 0.99Siemens KD 20 1.10–1.24a 0.53 0.46Varian 1800 18 1.5b 1.35 0.90Varian 1800 15 0.7b 0.82 1.17

aMcGinley(7)

bData given by Varian

R. C. FALCAO ET AL.

286

at University of W

innipeg on August 23, 2014

http://rpd.oxfordjournals.org/D

ownloaded from

Since both in a previous paper(12) and in thepresent one, we have observed that the calculatedfluence is underestimated by a factor of 2, wemultiplied the total fluence at point 2 by thisfactor.

We used the conversion factor given by ICRUPublication 74(14) to convert fluence to dose atthe inner entrance of the maze.

Knowing the dose at the inner entrance ofthe maze, we used neutron TVD dose valuesestimated by this work (see Equation 2), to cal-culate neutron dose equivalent values at theroom door.

CONCLUSION AND FUTURE WORK

We observed that the method of neutron dose cal-culation proposed above is in good agreement withMCNP simulated values. One of the advantages ofthis method is that, due to its simplicity, it can beused by clinical physicists to design door shielding,without having to perform Monte Carlo simulations,which, in the majority of cases in Brazil, is notavailable in hospitals. In the present method neutrondoses would not be underestimated, avoiding aradiological problem.

At present, we have been doing neutron dosesmeasurements at radiotherapy facilities in Brazil,using bubble detectors, in order to validate theseMonte Carlo predictions.

REFERENCES

1. National Council on Radiation Protection andMeasurements. Neutron contamination from medicalaccelerators. NCRP Report No. 79 (Bethesda, MD:NCRP) (1984).

2. Agosteo, S., Foglio Para, A. and Maggioni, B.Neutron fluxes in radiotherapy rooms. Med. Phys.20(2), 407–414 (1993).

3. Carinou, E. and Kamenopoulou, V. Evaluation ofneutron dose in the maze of medical accelerators. Med.Phys. 26(12), 2520–2525 (1999).

4. Waller, E. J., Jamieson, T. J., Cole, D., Cousins, T.and Jammal, R. B. Experimental and computationaldetermination of neutron dose equivalent aroundradiotherapy accelerators. Radiat. Prot. Dosim.107(4), 225–232 (2003).

5. Tosi, G., Torresin, A., Agosteo, S., Foglio Para, A.,Sangiust, V., Zeni, L. and Silari, M. Neutron measure-ments around medical electron accelerators by activeand passive detection techniques. Med. Phys. 18(1),54–60 (1991).

6. Norma de Radioprotecao CNEN. Requisitos de Radio-protecao e Seguranca para Servicos de Radioterapia.NE-3.06 (1990).

7. McGinley, P. H. Shielding Techinques for RadiationOncology Facilities. (Madison, WI: Medical PhysicsPublishing) (1998).

8. McGinley, P. H. and Butker, E. Evaluation of neutrondose equivalent levels at the maze entrance of medicalaccelerator treatment rooms. Med. Phys. 18(2),279–281 (1991).

9. X-5 Monte Carlo Team. MCNP—a general MonteCarlo n-particle transport code, Version 5, Volume I:overview and theory. LA-UR-03-1987 (Los AlamosNational Laboratory, Los Alamos, NM) (2003).

10. Facure, A., Falcao, R. C., Silva, A. X. and Crispim, V.R. Neutron dose rate evaluation for medical linear accel-erators. Radiat. Prot. Dosim. 111(1), 101–103 (2004).

11. Facure, A., Falcao, R. C., Silva, A. X., Crispim, V. R.and Vitorelli, J. C. A study of neutron spectra frommedical linear accelerators. Appl. Radiat. Isot. 62,69–72 (2005).

12. Facure, A., Silva, A. X. and Falcao, R. C. MonteCarlo simulation of scattered and thermal photoneutronfluences inside a radiotherapy room. Radiat. Prot.Dosim. Advance Access Published in June 30, 2006;doi: 10.1093/rpd/ncl080.

13. Lin, J. P., Chu, T. C., Lin, S. Y. and Liu, M. T. Themeasurement of photoneutrons in the vicinity of asiemens primus linear accelerator. Appl. Radiat. Isot.55(3), 315–321 (2001).

14. International Commission on Radiological Protection.Conversion coefficients for use in radiological protectionagainst external radiation. ICRP Publication 74(NY: Pergamon Press) (1995).

NEUTRON DOSE AT MAZE ENTRANCE

287

at University of W

innipeg on August 23, 2014

http://rpd.oxfordjournals.org/D

ownloaded from