Post on 05-Sep-2020
From Russia with Love - Monte Carlo Particle Transport Code SHIELD-HIT12A
Niels Bassler, PhD, <bassler@phys.au.dk>
Department of Physics and AstronomyAarhus University, Denmark
Department of Experimental Clinical OncologyAarhus University Hospital, Denmark
David C. Hansen, Armin Lühr, Nikolai Sobolevsky
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Outline of this talk
• About SHIELD-HIT and derivatives• Current status of the -A fork of SHIELD-HIT• Ongoing projects in Aarhus
• Question ahead of talk to the audience:– What feature would YOU like to see in SHIELD-HIT?
Niels Bassler, SHIELD-HIT12A
Niels Bassler
MC Treatment Planning Publications
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Common MC Codes
• Photons & Electrons only• EGS4, EGSnrc, BEAMnrc (National Research Council of Canada)
• Penelope (Facultat de Fisica (ECM), Universitat de Barcelona) • MCNP (including neutrons)
• Photons & Electrons + Ions• FLUKA (CERN)
• Geant4 (CERN)
• PHITS (Japan Atomic Energy Agency)
• MCNPX (Los Alamos National Laboratory)
• Ions only:• SHIELD(-HIT) (Institute for Nuclear Research RAS)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Common MC Codes in Use
Number of publications
Year
ISI Web of Knowledge
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Common MC Codes in Use
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT
SHIELD-HIT (Heavy Ion Therapy) was designed for precise simulation of interaction of therapeutic beams of protons and ions with biological tissue
Niels Bassler, SHIELD-HIT12A
Niels Bassler
• Compare MC codes with cars:http://willworkforscience.blogspot.com/
→ “Monte Carlo Programs in Particle Therapy Research”
A FLUKA car A Geant4 car A SHIELD-HIT car
Why SHIELD-HIT?
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT12A
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT
SHIELD-HIT (Heavy Ion Therapy) was designed for precise simulation of interaction of therapeutic beams of protons and ions with biological tissue
SHIELD-HITxxA is a fork with several improvements!
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT12A goals
• Black box approach• Simple to use
– easy install– no programming knowledge required– multi platform support
• Less restrictive license (commercial use allowed)• Add missing features from other MC codes (simple LET
scoring, better beam models, simple TRiP integration)• Make it fast and robust
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Vaporware
• “Vaporware is a term in the computer industry that describes a product, typically computer hardware or software, that is announced to the general public but is never actually released nor officially cancelled.”
• Surfaced vaporware:– Duke Nukem Forever– Diabolo III– Warcraft III– ...
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT12A – vaporware?
• As of today: a demo version can be downloaded• Still in beta, expect grave bugs• Demo version: hardcoded seed and statistics
• Download it here:
– http://neptun.phys.au.dk/~bassler/SHIELD_HIT/DEMO
– Or google SHIELD-HIT12A, there is a link from our project page.
Niels Bassler, SHIELD-HIT12A
Niels Bassler
History of SHIELD - Overview
• SHIELD – developed at JINR 1967-72– RSICC No.CCC-667– Transport of nucleons and pions up to 30 GeV– Continued at Institute for Nuclear Research of the Russian
Academy of Sciences (INR)– Rewritten 1989-90, Transport of nucleons, antinucleons,
pions and kaons up to 1 TeV– Fortran 77
• SHIELD-HI– 1997: Transport of ions with arbitrary A and Z
• SHIELD-HIT– 2001: Heavy Ion Therapy (INR, DKFZ, Karolinska)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
The SHIELD-(((HI)T)12A) Family
SHIELD-HIT08
SHIELD-HIT10
SHIELD-HIT10A
SHIELD-HIT12A
SHIELD-HITSHIELD-HI
SHIELD
SHIELD SHIELD-HI
SHIELD-HIT13A
1972
2001
2008
2010
2012
> 2012
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Under the hood
•Dubna cascade model (Toneev et al)•Quark-gluon string model (Amlein et al)•Coalescence model (Toneev et al)•Precompound emission of nucleons and lightest nuclei (Gudima et al)•Equilibrium deexcitation of residual nucleus:
–Fermi break-up of light nuclei (Botvina et al)–Evaporation/Fission competition (Botvina et al and Adeev et al)–Multifragmentation of excited nuclei (Botvina et al)
•Neutron transport LOENT 28 group ABBN (Abagayan et al.)
• Bethe and Lindhard-Scharff stopping power or external tables
Niels Bassler, SHIELD-HIT12A
Niels Bassler
• Usability• User's manual (yes, in English!)• Arbitrary scoring grids (Cartesian or cylindrical)• Scoring of particle energy - spectra files in TRiP format (.spc)• Scoring of depth-dose distributions in TRiP format (.ddd)• Plenty of new estimators (dose and track av. LET, β, ...)• Interface to SimpleGEO for geometry visualization• Read external Spread Out Bragg Peak files (.rst from TRiP)• Ripple filter (GSI type)• Binaries in Linux and Windows (32 and 64 bit)
• Computation speed• Parallelization• New random number generator (RANLUX->RANSHI)
• Nuclear models• Fine tuning of inelastic cross sections following experimental data (GSI)• Fine tuning of model parameters in Fermi-Breakup model.
What is new in SHIELD-HIT10A
18Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT10A, Carbon ion depth-dose curve
Experimental data: Herrmann et al. Med. Phys. 2011, Vol. 38, Pages 1859 – 1866Hansen, DC et al., 2012, Int. J. Rad. Biol. 88, p. 195-199
absolute dose
19Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT10A, C-12 fragmentation
Experimental data: Haettner et al, 2006Hansen, DC et al., 2012, Phys. Med. Biol. 57, p. 2393
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT10A - Scoring
Niels Bassler, SHIELD-HIT12A
Niels Bassler
• Usability• New detector: Alanine response• Scoring by zones (parallelizable)• New format of input files
– Simpler and extensible without breaking compability– includes ICRU database with 279 default materials
• New beam model– Divergence– Beam focus (postive and negative)– Arbitrary beam directions– More default beam shapes– RIFI operation mode
• Got rid of atab.dat and tabnuc.dat• Many bug-fixes (solved the fluence in vacuum bug)
• Even more extensive manual
• Computation speed• New Vavilov straggling routine (5-6 times faster than old routine)
WHAT IS NEW IN SHIELD-HIT12A
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Old school USB memory stick
Formatted F77 style input...
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Example: Material definition SHIELD-HIT
BEFORE: for022.dat – (fixed format)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Example: Material definition SHIELD-HIT12A
AFTER: mat.dat – (free format)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
What SHIELD-HIT12A doesn't do
• No graphical user interface (but there is a hack for geometry visualization)
• Not capable of handling voxel geometry (CT-scans)
• No photon/electron transport
• No Russian manual
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Plans for future release(s)
• Usability• Voxel geometry• IAEA phsp import-export (in progress, sort of)• Space related scorers and beam models
• Computation speed• GPU support
• Models• Photon / electron transport• Magnetic fields• Update cross sections of ions heavier than C-12 (in progress -
workshop 26th of January 2013 :-)• Update antiproton cross sections (in progress)• Better stopping powers (heavier targets, heavier ions, higher velocities)
(space applications) (in progress)• Radiobiology scorer
Niels Bassler, SHIELD-HIT12A
Niels Bassler
PyTRiPhelper functions
SHIELD-HIT12A Interfaces
SHIELD-HIT12A
libdEdxelectronic stopping power tables
MCNPneutron sources
TRiPtreatment planning
Particlespectrafiles(.spc)
Depth dosekernels(.ddd)
Rasterscan files(.rst)
SimpleGEOFor 3D visualization
of geometry
FLUKAinput files(.inp)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
VAVILOV STRAGGLING
(applied astronomy)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
New VAVILOV straggling function
• Decided to reimplement VAVILOV straggling due to GPL and copyright issues
• Original version was taken from GEANT 3.21, and could have caused trouble if SHIELD-HIT was released
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT v1 (2001)
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Vavilov straggling
● VAVINV() by Bjarne Thomsen (astronomer in emeritus)
● 5-6 times faster than Geant 3.21 version (Rotondi & Montana)
Results: speedup of code 30-40 % for a typical C-12 calculation with nuclear fragmentation switched on.
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Ripple Filter
Niels Bassler, SHIELD-HIT12A
Niels Bassler
SHIELD-HIT10A – Ripple Filter
Niels Bassler, SHIELD-HIT12A
Niels Bassler
RiFi-induced inhomogeneities
• RIFI -> various degree of lateral straggling of the ions
• inhomogeneous dose distribution close to RIFI
• project by MSc student Toke Printz in collaboration with Uli Weber at Marburg
Niels Bassler, SHIELD-HIT12A
Niels Bassler
• Detailed model of Marburg beamline implemented, including beam focus
• maximum dose inhomogeneity seen where lateral scatter is comparable to the RiFI-period of 0.15 cm.
• At Bragg peak, no effect seen (due to scatter and straggling)
• → can one make thicker RIFIs to save E-steps? Check DVHs for plans, using SH12A ↔ TRiP interface
RIFI dose ripple for inc. target distances
Niels Bassler, SHIELD-HIT12A
Niels Bassler
“Nuclear reaction models very important for particle therapy”
When? Where? How much?
37Niels Bassler, SHIELD-HIT12A
Niels Bassler
Nuclear Models in Particle Therapy
Nuclear models were changed in SH08 → SH10. What does this mean?
• Influence on1) Dose distribution2) Dosimetry
a) Stopping Power Ratios - b) Fluence Correction Factors3) Radiobiology4) PET isotope production for imaging5) Secondary neutron generation (secondary cancer)
● RELATIVE tests:A) Reference: nuclear models as they are now in SHIELD-HIT10AB) Turn off nuclear reactions entirelyC) Increase and decrease inelastic cross section by 20 %D) Fermi-Breakup model: change free volumes Vf/V0 and Vfc/V0
Lühr et al. 2012, Phys. Med. Biol. 57, 5169
38Niels Bassler, SHIELD-HIT12A
Niels Bassler
3) Effect on Radiobiology (LEM I, via TRiP)
39Niels Bassler, SHIELD-HIT12A
Niels Bassler
CONCLUSION
Impact of nuclear models on particle therapy:
1) Dose distribution20 % change in nucl. Cross sections -> 10 % change in SOBP
2) Dosimetry:a) Stopping power ratios: NOb) Fluence correction factors: A little (< 5%)
3) Radiobiology: No. (mostly attributed to dose shift from 1 -> non constant end point)
Here we only looked at inelastic cross sections, and Fermi-Breakup, due to newly available experimental data.
40Niels Bassler, SHIELD-HIT12A
Niels Bassler
Nuclear Models in Particle Therapy
Nuclear models were changed in SH08 → SH10. What does this mean?
• Influence on1) Dose distribution2) Dosimetry
a) Stopping Power Ratios - b) Fluence Correction Factors3) Radiobiology4) PET isotope production for imaging <--- NEW5) Secondary neutron generation (secondary cancer)
● RELATIVE tests:A) Reference: nuclear models as they are now in SHIELD-HIT10AB) Turn off nuclear reactions entirelyC) Increase and decrease inelastic cross section by 20 %D) Fermi-Breakup model: change free volumes Vf/V0 and Vfc/V0
Lühr et al. 2012, Phys. Med. Biol. 57, 5169
41Niels Bassler, SHIELD-HIT12A
Niels Bassler
4) PET Isotope distribution
Lühr et al., submitted. 2012
Niels Bassler, SHIELD-HIT12A
Niels Bassler
B) NO nuclear reactions --> NO PET isotopes C) +/-20% change inelastic cross sections: 20-40% change positron yieldsD) Fermi-Breakup model adjustment: mixed responsefewfrag: increase less than 10%, except 10C (50%) manyfrag: all yields decrease → improves agreement
Niels Bassler, SHIELD-HIT12A
Niels Bassler
Conclusions
• SHIELD-HIT12A– Fast and simple to use– Many features for research in particle therapy– Buggy demo version free to download (beta)
• Outlook– Voxel geometry– Photon/electron transport– [your idea here]
Niels Bassler, SHIELD-HIT12A
Niels Bassler
HIT Wednesday seminar, as seen from Aarhusduring Armin's talk in November last year: Jakob, Toke, Niels, Jørgen, Ricky