CC BY
24
DEVELOPMENT OF A COMPUTER MODEL
OF THE NG-121 NEUTRON GENERATOR
FOR A FAST NEUTRON THERAPY PLANNING SYSTEM
G.N. Malyshkin, E.A. Kashaeva, R.F. Mukhamadiev, V.G. Orlov, S.l. Samarin Russian Federal Nuclear Center- Institute of Technical Physics, Snezhinsk, Chelyabinsk Region, 456770 Russia
The SERA treatment planning system is being adapted to the NG-121 fast neutron facility used at Snezhinsk Neutron Therapy Center, Russia. This requires a computer model of the neutron facility that would match the SERA input format and provide adequately accurate results within a reasonable time. The detailed description of the facility cannot be used in SERA calculations because of its awkwardness, long time for dose calculation, and the requirement that the source must be defined as an energy-angle distribution of neutrons on a plane. First we made the detailed description of the neutron facility in the format of input data for the universal code PRIZMA which is a main tool for Monte-Carlo neutron transport simulations at RFNC-VNIITF. The detailed model of the generator was verified through comparison between calculated and experimental data. Then we considered a number of simplified models in which the neutron source was defined as a source on a plane. The source plane was defined in three positions: at the collimator’s outlet, at the collimator’s inlet, and near the real source. Only the third model provided a good agreement of results. Further analysis into the importance of neutrons in different regions of the third model showed the regions where they were of low importance that allowed us to limit the neutron tracking region by a truncated cone and significantly reduce the dose calculation time. Doses calculated along the neutron beam axis by PRIZMA, SERA and MCNP5 agree well.
Introduction
Since July 1999 a Neutron Therapy Center has been operated at RFNC-VNIITF, Snezhinsk. The Center exploits the NG-121 neutron generator producing 14-MeV neutrons at a rate of 1012 n/s. Medical support to the Center is provided by experts from the Cancer Detection and Treatment Center in Chelyabinsk, where they perform the primary examination of patients, diagnostic studies, collection of topological and metrical information about patients and treatment planning. The photon therapy phase of the treatment is also provided at the Chelyabinsk Center and after that the patients are sent to Snezhinsk for the neutron therapy stage at the Neutron Therapy Center. During 5 years of the Center’s operation, the treatment has been given to more than 350 patients in whom cancers were sited in head and neck regions.
The SERA treatment planning system is being adapted for use at the Snezhinsk Neutron Therapy Center. Foreign collaborators of this work are the Idaho National Engineering and Environmental Laboratory (INEEL) and the Lawrence Livermore National Laboratory (LLNL).
To use the SERA code for treatment planning at the Snezhinsk Neutron Therapy Center, it is necessary to create a model of the NG-121 neutron generator that would meet the following requirements:
• simulations with this model should adequately describe the actual dose deposition in exposed objects;
• the SERA simulation time should be acceptable for clinical practice;
• the model should match the SERA input format.
1. NG-121 neutron generator
The NG-121 neutron generator is a complex unit, whose key elements are a target device, a three-layer biological shield, and a collimating system with replaceable square collimators (see Fig. 1).
The target is an elliptic copper segment 2 mm thick. Its inner surface is covered with a titanium layer ~10 |im thick, saturated with tritium. Deuterons of energy 210 keV interact with this layer and produce 14-MeV neutrons emitted almost isotropically.
The composite biological shielding consists of 45-cm-thick steel, 15-cm-thick borated polyethylene and 5-cm-thick steel. The target is shielded by 10-cm-thick steel and 50-cm-thick borated polyethylene.
The hole in the biological shielding for installation of different collimators is shaped as a truncated cone transforming into a cylinder having inlet 5,5 cm in diameter and outlet 20 cm in diameter. The collimator inserts are similar to the biological shielding in composition except for polyethylene, which is non-borated here. Inlet holes in all collimators measure 4x4 cm2 and the size of outlets is such that the radiation beam at a distance of 10 cm from the collimator’s outlet measures 4x4, 6x6, 8x8 cm2 and so on.
Fig.1. A sectional view of the NG-121 neutron generator
Fig. 2. General 3D view of the NG-121 neutron generator
in sectionand the titanium-tritium layer (brown) of the target assembly
Fig. 3. Locations of three model sources
Rea!
ftOlirCC
\
Ш:
; Three-layer biological shield
/ /
.. ' Collimator
Model source (plate)
m
b)
a)
Fig. 4. Full (a) and simplified (b) models of the NG-121 neutron generator
2. Full model of the NG-121 neutron generator in prizma
Based on the technical documentation, a full simulation model of the NG-121 neutron generator (see Fig. 2) was described in the format of initial data for the universal code PRIZMA which is a main tool for neutron Monte-Carlo simulations at RFNC-VNIITF. The code simulates the separate and coupled transport of neutrons, photons, electrons, positrons and ions, using continuous-energy cross-sections ID, 2D, or 3D geometry, lattice and stochastic geometries are used. Biased sampling techniques are well developed and used for a wide class of problems. The code implements a method of correlated tracking which enables the analysis of results of several variants in a single run.
The energy-angular distribution of secondary neutrons was obtained in a preliminary simulation with the PRIZMA code modeling the effect of 210
Based on the performed simulations, a conclusion was drawn that the source plane should be located as close to the real source as possible because only in this case the angular distribution of directly transmitted neutrons can be defined correctly. Further analysis of neutron importance showed that neutrons scattered in the biological shielding far from the collimator contributed insignificantly to the dose in the phantom. Therefore, calculation efficiency can be improved almost at the same accuracy level by ignoring those neutrons.
For model 3, a simplified geometry (model 4) was considered which contained only a part of the biological shielding inside a truncated cone with the base radii of 3 and 20 cm (see Fig. 4); neutrons escaped from the cone were never back in the system.
Fig. 5 shows neutron dose rate calculated in the full statement (in red) in terms of 1012 neutrons per second and in the statement with the model 4 (in blue).
a) Phantom center
b) Phantom haif-shade
Fig. 5. Neutron dose rate along z-axis for the model 4
keV deuteron beam on the elliptic layer of the target. The detailed model of the generator was verified through comparison between calculated results and experimental data obtained in experiments on the NG-121 generator.
3. Simplified source models for sera
However, the full-scale model of the neutron generator developed for PRIZMA cannot be used in SERA calculations because of its awkwardness, long time for dose calculation, and requirements for source definition in SERA as an energy-angular distribution of neutrons on a plane (called the source plane) located between the real source and exposed object perpendicular to the radiation beam axis.
Several simplified models of the neutron generator were considered. The source plane was defined in three positions (see Fig. 3): at the collimator’s outlet, at the collimator’s inlet, and near the real source. The model sources were checked for validity by comparing doses in a phantom, calculated by PRIZMA in simplified and full statements.
As for the efficiency of all the above models, it can be evaluated from the Table I.
Table 1
Comparison of model efficiencies
Source model Relative calculation time
Model 1 1
Model 2 50
Model 3 100
Model 4 5
Full-scale model 8000
Based on the above consideration, it was decided to use model 4 as a basic one in the SERA treatment planning system.
Finally, comparative dose calculations for model 4 were done by PRIZMA, SERA and MCNP5 [8]. Doses were calculated along the segment located on the beam axis. The results were averaged over one row of cubic cells lxl x lcm3 whose centers lied on the above the segment. Calculated doses are shown in Fig. 6.
It is seen that doses calculated by the three codes agree well, minor differences may result from the use of different cross-section libraries.
Fig. 6. Neutron dose rate along the beam axis
4. Conclusions
As our research shows, the model of the neutron generator incorporating the model source located near the real source and the simplified geometry of the biological shielding ensures the acceptable accuracy of dose calculation within a reasonable time and can be used to define the NG-121 neutron generator in the SERA code.
Reference
1. Litvin VI., Mokichev G. V, et al. «Dosimetric parameters of radiation beam produced by NG-12 facility». Abstract of presentation at the 3rd Conference of Health Physicists Association of RF, Obninsk, 1997.
2. Vazhenin A. V., Vasilchenko M. V., Shmygin V.A., Munasipov Z.Z., Magda E.P., Mokichev G. V.
«Results of clinical trial of fast neutron therapy». Abstract of presentation at the Conference «To W0'h anniversary of N. V. Timofeyev-Ressovsky», Snezhinsk, 2000.
3. SERA Workshop Lab Manual. Idaho National Engineering and Environmental Laboratory, INEEL/EXT-99-00766, 1999.
4. Kandiev Ya.Z., Kuropatenko E.S, Lifanova I. V., Orlov A.I., Plokhoi V. V, Shmakov V.M. Monte-Carlo simulations of the particle-matter interactions with PRIZMA code. Abstracts of presentations made at the III All-Union Conference on Shielding against ionizing radiation from nuclear engineering installations. Tbilisi, 1981.
5. Arnautova M.A., Kandiev Ya.Z., Lukhminsky B.E., Malyshkin G.N. Monte-Carlo Simulation in Nuclear Geophysics. Incomparison of the PRIZMA Monte Carlo Program and Benchmark Experiments. Nucl. Geophys. Vol. 7, Ns 3, 1993, P. 407-418.
6. Vasilyev A.P., Kuropatenko E.S., Lyutov V.D., Orlov A.I., Shmakov V.M.. Nuclear Data Library -BAS. The history of development and validation for criticality safety calculations. 1CNC’95. Proceedings of the international conference of nuclear criticality safety, Albuquerque, New Mexico, USA, September 17-21, 1995, P. 2.56-2.60.
7. Kandiev Ya.Z., Malyshkin G.N. Modeling by Value Implemented in PRIZMA Code. V Joint Rus-sian-American Computational Mathematics Conference. Sandia Report. SAN98-1591, 1998, P. 149-158.
8. X-5 Monte Carlo Team, «MCNP - A General Monte Carlo N-Particle Transport Code, Version 5, Volume I: Overview and Theory», LA-UR-03-1987, Los Alamos National Laboratory (2003).
