COMPARISON OF PREDICTIONS BY MCNP AND EGSNRC OF RADIATION DOSE IMPARTED TO VARIOUS MATERIAL TARGETS BY BEAMS AND SMALL VOLUMETRIC SOURCES.

Transcription

1 Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, 2007 University of Florida (UF), Gainesville, Florida USA on CD-ROM, Amerian Nulear Soiety, LaGrange Park, IL (2007) COMPARISON OF PREDICTIONS BY MCNP AND EGSNRC OF RADIATION DOSE IMPARTED TO VARIOUS MATERIAL TARGETS BY BEAMS AND SMALL VOLUMETRIC SOURCES. Eri Steinfelds Department of Nulear and Radiologial Engineering University of Florida Gainesville, Florida ABSTRACT EGSnr is optimized in order to model the transport of eletrons in a signifiantly more time effiient manner than MCNP. However, it seems that diret eletron ontributions to dose are easier to trae and reord with MCNP than with EGSnr. MCNP makes it lear where the eletrons stop and where seondary photons ontinue for e-beam simulations. Within the proessing algorithms of EGSnr, photon dosimetry and eletron dosimetry are intermixed and very diffiult to separate. The rad transport expert would wonder why EGSnr reports small ontributions from beta partiles whih penetrate well beyond the max CDSA range of the beta, until one onsiders that EGSnr models and reords beta eletron and seondary photon doses simultaneously. For the nulear engineer, it is also instrutive to point out that EGSnr is designed exlusively for the transport of photons and eletrons(+-), not for hadrons (i.e. protons, alphas, et.). This is rather limiting in versatility for EGSnr ompared to MCNPX, whih now does inlude many hadrons in addition to alphas. The predominant modes of radiation therapy are still done with hard photons, semihard photons, and less frequently with energeti eletrons; either from aelerators or radioisotopi soures (e.g. Ir-192 et.). Radiographi imaging relies on positrons, photons, and oasionally neutrons. Thus, EGSnr is versatile enough with its hoies of radiative partiles for para-linial appliations. For geometri onfigurations and designs in the medial realm, virtually all material samples and targets are arranged within retangular voxels, whose planar boundaries are aligned with their respetive x,y, and z axes. While this might seem very limiting to the traditional nulear engineering oriented user of MCNP, this is quite suffiient for the medial physiists and medial physis developers, who almost always work with 3 dimensional images of patients whih are mapped as voxels with a grid-partitioning of often more than 70 by 70 by 50. These mappings of the densities and the images are reorded as files whih are translated from the CT image files of a given patient or speimen. In this paper, it is the simulated doses imparted to eah voxel of a given geometry that are reorded and reviewed. KEYWORDS: MCNP, EGSnr, dosimetry, dosimetri omparison, retangular beam. 1. INTRODUCTION These omputational dose estimates were generated by both MCNP and EGSnr in order to be reviewed for similarities and differenes of performanes between that of the two Monte Carlo odes. EGSnr was developed by the Researh Center of Canada.[1] Dose per voxel is emphasized in this paper beause medial physiists are mostly interested in doses resulting from radiation, and not very muh in the transmissions of partiles. Note that the two omputers used

2 for respetive omparisons between MCNP and EGSnr are both Dells with the single Intel Pentium(R) 4M CPU. More most of this paper, a square ollimated beam of some partile is modelled. This beam is transversely shaped as a square whose length of a side is 1.6 m. Within this paper, the target of suh a beam is a ubi water-like material. The beam targets the middle of this soft tissue. The objetive is to ompare the time durations, the ease of operation, plausibility of results, and degree of agreement of results between MCNP and EGSnr. The simulation results between MCNP and EGSnr are similar but not idential. Among preliminary examples of simulation for a given setup of water (or flesh equivalent) filled voxels whih are irradiated by retangular beams of either photons or eletrons, it was generally found at almost all energy levels (.5 thru 18MeV) that EGSnr predits slightly higher doses than MCNP does. These are the doses imparted to samples within the target and to the material regions surrounding the target. It is a major aid to do MCNP modelling in pure photon mode and separately in (photon,eletron) mode in order to effetively aount for the dose ontributions from seondary eletrons. It will beome apparent in the first table that MCNP models primary photons and sattered photons from primary photons (and possibly tertiary photons) almost as fast as EGSnr does, but EGSnr is at least 20 % at lower energies. For energies above 5MeV, the disrepany in dose preditions within the first 3 entimeters of beam penetration leads to preditions by EGSnr whih are more than 10 perent larger than the MCNP preditions. The same number of soure partiles was used in eah simulation, per geometrial target. For simulations of eletron beam, the speed advantage of EGSnr per number of soure partiles is rather extreme, where the EGSnr is easily an order of magnitude faster, even approahing a 50 fold speed advantage. Some therapy uses radio-seed implants and various styles of stereotati radiotherapy. In addition to modelling retangular beam, a brief example of omparison is also given of a point soure emitting the strong γ-ray of Cs-137 toward the end of this text, inluding a figure (4). 2. METHOD, GEOMETRIES PICKED, AND INPUT FOR MC MODELLING For most of the examples a retangular struture of flesh equivalent material or H 2 O ould be desribed as being bombarded by mono-energeti beams of eletrons and photons. One example is given of a spherially symmetrial phantom with 18 m diameter being irradiated by the main gamma ray (i.e..662 MeV) from a Cs-137 point soure. The gamma ray instigated dose predited (by MCNP and EGSnr) to voxels as a funtion of radial distane are given in the last table at the end of this paper. Let us onsider a perfetly ollimated retangular beam pointing along the x-axis hitting a retangular blok whih has been subdivided into 6 by 5 by 3 voxels. Along the x-axis, there are 6 suessive voxels of H 2 O. Figure (1) illustrates the geometry. The red arrows denote the beam of primary eletrons or photons. See figure (1): Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /2

3 Geometry of beam simulations: Length and width of eah voxel is 1m. Beam axis is path of eletron of photon. This is of the beam. Off-beam axis Figure (1). All of the ells onsisting of H2O or soft tissue appear as in Figure (1). Purple ells 202, 222, et. have density of 1.00 gm/m3. There 700,000 soure partiles. In figure (1), note that ells 202, 222, 232, 242, 252, and 262 of Figure (1) lie diretly on the axis of the beam of radiation, whih emerges from inside the blue ell 101. This axis was defined as the X-axis. All of the ells onsisting of H2O or soft tissue appear as in Figure (1). These purple ells have a density of 1.00 gm/m3 and are denoted as voxels, as in the medial onventionalism. In figure (I) and in all of the beam simulations of this paper, the beam is retangularly shaped. The beam traverses along the X-axis. The beam has width-y of 1.6m and width-z of 1.6m. The beam was populated either by 700,000 monoenergeti eletrons or 700,000 monoenergeti photons. The hoies of energy for the initially monoenergeti were.5mev, 1MeV, 6MeV, and finally 18MeV. These simulations were onduted with both MCNP and EGSnr, using 700,000 soure partiles (e.g. nps). This paragraph gives an example of a typial MCNP input file used to by the author in order to generate the alulation of doses of those voxels of figure (1) whih are espeially of interest. The doses due to MCNP were extrated by setting up the MCNP input file to have either F6 ards for MeV's per gram or *F8 ards for MeV's of energy released per hosen type of partile in a given ell. (hoie of partile to read should not hange in a dosimetri file). Appendix A shows the typial type of MCNP file used to model delivery of dose from monoenergeti eletrons to 'purple' geometry of the phantom whih is desribed in Figure (1). Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /3

4 The *F8 ards and the soure ards were speially hosen for the MCNP input file for modelling the eletron beam therapy of the phantom of Figure (1). The following several lines show the *F8 ards of the input and the soure ard governing the generation of a 6 MeV beam of monoenergeti eletrons. mode e p the soure ard defined the emergene of a beam of 6MeV eletrons. sdef erg= 6.0 SUR=1312 POS= RAD=D1 DIR=1 =191 PAR=3 $ Means ookicuttercell; SI1 1.2 $ 1.2 > sqrt(2)*1/2widtbox=sqrt(2)*.8m. $ next: C F18*:p $ energy deposited into ell ; 2nd phot/ from eletrons F28*:p $ energy deposited into ell ; 2ndy photons F38*:p $ energy deposited into ell ; 2ndy photons E $ energy bin splitters.. 3. NUMERICAL DATA FROM THE DOSE SIMULATIONS. 3.1 Durations of Simulations Dosimetri Results of beam Monoenergeti Eletrons hitting retangular target of soft tissue are given in the next tables. Note that soft tissue onsists of 80% H2O and of organis. Soft tissue is equivalent to water in terms of dosimetry, attenuations, and stopping powers. E of partiles from beam Partile of or by the beam: below: E-beam Table I. Durations of Dose Simulations as a Funtion of Energy and Code. Simulation time via MCNP (min) Simulation time via EGSnr (min) Simulation time via MCNP (min) Simulation time via EGSnr (min) Simulation time via MCNP (min) eletron (e) eletron (e) photon (p) photon (p) Seondary* e s generated from p s 0.5 MeV about MeV MeV MeV * Seondary e s are onsequential to stuff lost by photons. 700,000 soure partiles existed for the partiles of the initial beam for eah simulation reorded in Table I. In the last olumn, the results of the seondary eletrons either liberated or generated by the photons are reorded. A summary of observations of the running times of the different partiles as a funtion of energy and hoie ode is offered here. EGSnr simulations of eletrons are muh faster than MCNP simulations of eletrons, often by more than 40 fold. Plain photon simulations via MCNP are almost as fast as those of EGSnr, so long as doses and salar fluxes due to seondary eletrons are not evaluated and omputationally reorded. It will be seen from Table Q that at energies above 1 MeV for photons, dose ontributions due to seondary eletrons annot be ignored. Therefore an MCNP simulation in primary photon mode plus a 2nd MCNP simulation in the (p-e) mode should be arried out - if the photon beam energy energy exeeds 1 MeV. The (p)-mode simulations take only 1 minute approximately, but the (p- Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /4

5 e) mode simulations to inlude seondary eletrons (e.g. p ollision daughters) endure for more than 20 minutes. The times of these two distint simulations need to be added to get the realisti duration of a omplete dose simulation for beams above 1 MeV. 3.2 Doses to the voxels due to eletron beams In this paragraph and the next few pages, we fous on dosimetry results, on how muh dose is imparted to a given voxel due to a beam of eletrons here and later hard photon beams. Numerous tables of doses generated by eletron beams are given. x- mell (m) DosemnpEle trish [OnbeamX] (gray/disint.) Table II. Doses of Voxels whih are on the Beam Axis. Flutuationm np Eletrish [OnbeamX] DoseEletrish [OnbeamX] (gray/disint.) Flutuation Eletrish[Onbe amx] (deimal) e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e Ebea m (MeV) A key desription to the labels in Table II is given here: /Unit (Gray/disnt) is Grays per disintegration./ X-mell is the x oordinate of the middle of the ell whih reeives dose. DosemnpEletrish[OnbeamX] is the dose imparted to the given ell (or 1m^3 voxel) due to beam eletrons, sattered eletrons, and liberated eletrons via MCNP. DoseEletrish[OnbeamX] is the dose imparted to the given ell of 1m^3 due to eletrons via Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /5

6 EGSnr. FlutuationmnpEletrish[OnbeamX] is the flutuation of the value of the MCNP generated dose. FlutuationEletrish[OnbeamX] is the flutuation of the value of the EGSnr generated dose. The values of flutuation are given as deimals, not perentages. Ebeam is the energy of eah original photon while it is in the original beam trajetory. In the label DosemnpEletrish[OnbeamX], the nom. expression in brakets, [OnbeamX], indiates that voxels exposed are all entered in line with the partile beam axis. The enter of an [OnbeamX] voxel is loated at (x,y,z) = (x-mell, 0m, 0m). In the label DosemnpEletrish[OffbeamX], the nom. expression in brakets, [OffbeamX], indiates that voxels exposed are laterally entered one m away from the partile beam axis. The enter of an [OffbeamX] voxel is loated at (x,y,z) = (x-mell, 1m, 0m). Table III. Doses of Voxels whih are 1m off-beam x- mell( m) DosemnpEletrish [OffbeamX] (gray/disint.) Flutuation mnp Eletrish[Of fbeamx] DoseEletrish [OffbeamX] (gray/disint.) Flutuation Eletrish[Of fbeamx] e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e Ebeam (MeV) For the sake of keeping this paper within reasonable length the numerial results of seondary photons in the MCNP simulations are not inlude. However a simple observation of (e-p) mode simulations of seondary photons from the eletron beams is offered. The dose volume deposited in (ele-photo) mode tends to be almost the same (slightly less than) the dose volume deposited in eletron mode in the ases of.5mev beam and 18 MeV beam, for voxels on the beam axis. This is espeially true in the ells where the journey of e is more than ½ ompleted. 3.3 Doses to the voxels due to beams of monoenergeti photons Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /6

7 All of the data in this text involving doses generated by radiative eletrons has been reorded above. In this setion and in the next setion all of the alulation involve doses or energy transfer originating from beams of photons or gamma rays. In the next few tables, we fous on mono energeti beams of photons. The hoies of energy of the beam are: 0.5MeV, 1MeV, 6MeV, and 18 MeV. Q = "IV" Table Q. Doses due to photon beam onto the Voxels whih are on the Beam Axis. x- m ell DosemnpPhot oni [OnbeamX] DosemnpPhoton i [OnbeamX] [filenumber+100 _ini] DosePhotoni [OnbeamX] Flutuation mnpphoto ni[onbea mx] FlutuationP hotoni [OnbeamX] E e E E e E E e E E e E E e E E e E e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e Ebea m (Me V) Here is the key to labels in Table (Q): X-mell is the x oordinate of the middle of the ell whih reeives dose. DosemnpPhotoni[OnbeamX] is the dose imparted to the given ell (or voxel) due to photons via MCNP. DosemnpPhotoni[OnbeamX] [filenumber+100,ini] is the dose imparted to the given ell (or voxel) due to seondary e s and tertiary eletrons resulting from the photons, via modelling and hoie of modes in MCNP. DosePhotoni[OnbeamX] is the dose imparted to the given ell (or vox) due to photons via EGSnr. FlutuationmnpPhotoni[OnbeamX] is the flutuation of the value of the MCNP modelled dose. FlutuationPhotoni[OnbeamX] is the flutuation of the value of the EGSnr generated dose. Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /7

8 These labels in boxes below mirror the boxes of Table (Q), where gray/disnt means Gray/ disintegration. x- DosemnpPhot DosemnpPhoto DosePhoton Flutuationmnp FlutuationP Ebea mell oni ni [OnbeamX] i[onbeamx Photoni[Onbea hotoni m (m) [OnbeamX] [filenumber+10 ] mx] [OnbeamX] (MeV (Gray/disnt) 0,ini] (gray/disnt) ) The next two graphs in figure (2) and figure (3) visually display the dose results of Table Q. 2.50E E E E E E E E E E E Dose by mnp (0.5MV) Dose 2ndy eles (mnp) Dose by EGS (0.5MV) Dose by mnp (1 MV) Dose 2ndy eles (mnp) (1 MV) Dose by EGS (1 MV) Figure (1). Graph of Doses from Table Q, Comparison of Doses by main photons by MCNP and Doses due to seondary mediating ele's. Beam energies are.5mev and 1MeV. 1.80E E E E E E E E E E Dose by mnp (6MV) Dose 2ndy eles (mnp) Dose by EGS (6MV) Dose by mnp (18MV) Dose 2ndy eles (mnp) (18MV) Dose by EGS (18MV) Figure (3). Graph of Doses from Table Q, Comparison of Doses by main photons by MCNP and Doses due to seondary mediating eletrons. The two highest beam energies were seleted. Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /8

9 Now let us onsider the dose imparted to voxels whih are laterally adjaent to the voxels whih lie on the beam axis. The off-beam voxels whih reorded in the next table inlude: ell 201, 221, 231, 241, 251, and 261 of figure (1). Again the dose ontribution from seondary eletrons in reorded in the third olumn of the next olumn, exept for the ase of 1MeV. Table V. Photoni Doses of Voxels whih are 1m away from the enter of Beam Axis. x- mell (m) DosemnpPhotoni [OffbeamX]] (Gray/disnt) DosemnpPhotoni [OffbeamX] [filenumber+100_i ni] DosePhotoni[Offbea mx] (Gray/disnt) FlutuationPh otoni [OffbeamX] Flutuationm npphotoni[ OffbeamX] e-13 almost same e e-13 almost same e e-13 almost same e e-13 almost same e e-13 almost same e e-13 almost same e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e e Ebeam (MeV) As with the other dose tables the initial energy of primary photon (i.e. Ebeam) equals.5, 1, 6, or 18 MeV. Next we onsider a point soure of photoni Cs-137 surrounded by soft tissue or water. Consider good agreement between photon simulations of MCNP and EGS for spherial phantom with 18 m diameter. A point soure or mirosopi ubi soure ontaining highly onentrated Cs-137 is plaes in the enter of the spherial phantom. Note that EGS geometry mimis an approximate sphere with numerous retangular ells of H 2 O. Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /9

10 Dose in Grays per d i s i n t 1.00E E E E-14 Gray/dis (EGS) Grays/dissint (mnp) 1.00E-15 Radius from point soure of γ-rays in m. Figure (4). Dose per disintegration ('disint') versus r from Cs-137 soure in H CONCLUSIONS From preliminary review it is lear that EGSNr is muh faster than MCNP at modelling primary eletrons. MCNP is more versatile than EGSnr. Only dose and flutuations are reorded by EGSnr. Based on the understanding of differenes, it is expeted that not any time within the span of several years will EGSnr be modified to inlude protons for proton therapy. EGSnr internally is most reliable for retangular geometries. EGSnr an read spetra and geometry and ompositions from external files. - ontributing to onveniene. MCNP when speialized an read external geometry. The (p-e) mode simulations of photons should be done within MCNP in addition to the (p) mode simulation. A study of the dose tables herein makes this lear. Conlusive statements: A thorough study of the tables given for the dosimetry of photon targeted Figure (1) makes it very apparent that the dose values in (p-e) mode of MCNP are almost onsistently loser to the dose values in (p)mode of EGSnr than the values of the (p) mode simulations of MCNP. Moreover, the dose distributions as a funtion of beam depth from EGSnr and MCNP in (p-e) mode math the dosimetri beam profiles given in the graphs of the text The Physis of Radiation Therapy [2]. On the other hand, the dose distribution by MCNP simulation in purely (p) mode fails to aount for signifiant build up and the graphs published in Kahn s book. The agreement between the dose table of values from MCNP s (pe) mode and the values of EGSnr are reasonably lose. The patterns of the graphs of MCNP s (p-e) mode the orresponding graphs of EGSnr are very lose, suh that the ratio of the two graphs remains onstant, although not exatly equal to one. Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /10

11 REFERENCES 1. Code EGSnr; Ionizing Radiation Standards; national researh ounil of Canada; (2006-7). 2. Faiz Kahn, The Physis of Radiation Therapy (Third Ed.), pg. 163, (2003). 3. Frank Attix, Introdution to Radiologial Physis and Radiation Dosimetry, pg. 229, Wiley- Intersiene (USA), (1987). Page 229 has a photon dose graph. APPENDIX A new tool in 2007 on yl ookie uttermakng sqr. Erg eletr= 6MeV $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh imp:e 1 $ targ flesh Shift in z oordinate by -1 entimeter: imp:e 1 $ 1702; targ flesh Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /11

12 imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh imp:e 1 $ 1702; targ flesh this is a parallel planar repetition of ell imp:e 1 $ targ flesh this is a parallel planar repetition of ell imp:e 1 $ targ flesh this is a parallel planar repetition of ell imp:e 1 $ targ flesh this is a parallel planar repetition of ell imp:e 1 $ targ flesh this is a parallel planar repetition of ell ells 1224 through 1264 have been omitted from this reprodution in Appdx. A. Shift in z oordinate by +1 entimeter: imp:e 1 $ 1702; targ flesh this is a parallel planar repetition of ell imp:e 1 $ 1702; targ flesh this is a parallel planar repetition of ell imp:e 1 $ 1702;targ flesh this is a parallel planar repetition of ell ells 2203 through 2264 have been omitted from this reprodution in Appdx. A imp:e 1 $//orretd imp:e 1 $//orretd imp:e 1 $//orretd imp:e imp:e imp:e imp:e imp:e imp:e 1 $//orretd imp:e imp:e imp:e 1 These were equatorial ells imp:e 1 $ ell whene beam emerges imp:e imp:e 1 These were equatorial ells; imp:e imp:e imp:e 1 $ 30* boundry inrses removed from target ell; imp:e 1 $ australz sphere imp:e 1 $ borealz sphere imp:e 1 $ australz sphere imp:e 1 $ borealz sphere Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /12

13 imp:e 0 surfae ards 4 /y s $ lastn is radius 302 px px px px px px px py py py py py py py py py pz pz pz pz px -5.0 $ plane of beam emergane 900 s mode e p the soure ard defined the emergene of a beam of 6MeV eletrons. sdef erg= 6.0 SUR=1312 POS= RAD=D1 DIR=1 =191 PAR=3 $ Means ookicuttercell; SI1 1.2 $ 1.2 > sqrt(2)*1/2widtbox=sqrt(2)*.8m. C next: F18*:p $ 2nd phot ; eletrish/ from eletrons F28*:p $ 2ndy photons ; eletrish F38*:p $ 2ndy photons ; eletrish E $ energy bin splitters m $ mass omposition type 1 m1 ould be replaed by biologial soft tissue omposition. m $ mass omposition type 3 m nps $ # of soure partiles released for simulation Computational Medial Physis Working Group Workshop II, Sep 30 Ot 3, /13