improved ctimproved ct--based voxel based voxel …cmpwg.ans.org/icrs12/presentations/mike...
TRANSCRIPT
-
IMPROVED CTIMPROVED CT--BASED VOXEL BASED VOXEL PHANTOM GENERATION FORPHANTOM GENERATION FORPHANTOM GENERATION FOR PHANTOM GENERATION FOR
MCNP MONTE CARLOMCNP MONTE CARLO
Michael Speiser, Ph.D.p ,Department of Radiation OncologyUT Southwestern Medical Center
Dallas, TX
DEPT OF RADIATION ONCOLOGY
September 1st, 2012CMPWG Workshop
-
Medical Physics Investigations Medical Physics Investigations and Monte Carloand Monte Carlo
Patient + Source = Dose• Source Modelingg• Patient Modeling
• AccurateAccurate• Efficient• Simulation pre- and post-processing• Simulation pre- and post-processing• Radiotherapy analysis
DEPT OF RADIATION ONCOLOGY
-
DEPT OF RADIATION ONCOLOGY
-
DEPT OF RADIATION ONCOLOGY
-
Patient ModelingPatient ModelingPatient ModelingPatient Modeling
“RTMCNP Preprocessor” (DeMarco et al 1997)(DeMarco et al. 1997)
DEPT OF RADIATION ONCOLOGY
-
CT “Information”RESCALINGRESCALINGRESCALING
-707 -475 -225 -116
-787 -665 -416 -200
Hounsfield Units-805 -785 -628 -369
Hounsfield Units
-793 -804 -757 -568
DEPT OF RADIATION ONCOLOGY
-
MCNPMCNPMCNPMCNP
• Use “nested lattice” featureUse nested lattice feature• Bounding cell• Lattice element• Lattice element• Bounding cell origin
DEPT OF RADIATION ONCOLOGY
-
Creating a Lattice StructureCreating a Lattice Structure
-707 -475 -225 -116
di i /
-787 -665 -416 -200-665
y-dimension / “slice thickness”z-dimension /
“voxel height”
805 785 628 369
voxel height
-805 -785 -628 -369x-dimension / “voxel width”
-793 -804 -757 -568
“lattice element”
DEPT OF RADIATION ONCOLOGY
• Cell comprised of repeating identical shapes
-
Creating Creating a Lattice Structurea Lattice Structure• Hounsfield Units to Tissues
1. Air ELEMENT WEIGHT %2. Lung3. Fat HUHU
ELEMENTH (Z=1)C (Z=6)N (Z=7)
WEIGHT %10.310.53.1
HU range954 t 824
ρ (g/cm3)0 048
4. Water5. Muscle
( )O (Z=8)
Na (Z=11)Mg (Z=12)
74.90.2…
-954 to -824-824 to -674-674 to -524
0.0480.12540.2978
6. BoneP (Z=15)S (Z=16)
Cl (Z=17)K (Z 19)
0.20.30.30 2
-524 to -374-374 to -224
0.47210.6455
• Alfidi et al (Radiology 1975)
K (Z=19)Ca (Z=20)
0.2…
DEPT OF RADIATION ONCOLOGY
Alfidi et al. (Radiology, 1975)• ICRP 23 (Reference Man, 1975)• ICRU 44 (1989)
-
Hounsfield Units to MaterialsHounsfield Units to MaterialsLOOKUP TABLELOOKUP TABLE
DEPT OF RADIATION ONCOLOGY (DeMarco et al. 1997)
-
Hounsfield Units to MaterialsHounsfield Units to MaterialsLOOKUP TABLE
HU=0HU=0
HU=0 material definition is H2O ρ=1 0 g/cm3
DEPT OF RADIATION ONCOLOGY
HU 0, material definition is H2O, ρ 1.0 g/cm
-
RTMCNP, DeMarco RTMCNP, DeMarco et alet al..1. Read CT
G t• Geometry• Hounsfield UnitsS C2. Sub-sample CT
• Lattice of 128x128 or 64x643. HU’s into materials (%Z, ρ)4. “Fill” lattice with materials5. Add modified pulse height tally (MeV / g)
DEPT OF RADIATION ONCOLOGY
-
Improving CTImproving CT--Based Model Based Model GenerationGenerationGenerationGeneration
• Flexible geometry modificationsg y• Sub-sampling• Lattice trimming• Isocenter positioning
• Mesh tally definition• Pre- and post-processing
• Material definitions for physical phantoms• Normalization criteria
• Radiotherapy analysis (gamma)
DEPT OF RADIATION ONCOLOGY
-
Efficiency: SubEfficiency: Sub--Sampling the CTSampling the CT512x51264x64128x128256x256
Fewer surfaces, faster simulationsL ti t l ti
DEPT OF RADIATION ONCOLOGY
Less patient resolution
-
DEPT OF RADIATION ONCOLOGY
-
Efficiency: Voxel TrimmingEfficiency: Voxel Trimming
~60% reduction in time required(Speiser et al. 2007)
DEPT OF RADIATION ONCOLOGY
( p )
-
DEPT OF RADIATION ONCOLOGY
-
Original lattice:S b l d l tti
512 x 512 x 160256 256 160Sub-sampled lattice:
Trimmed lattice:256 x 256 x 160208 x 132 x 100
Trimmed lattice < 6 6% of the original scanDEPT OF RADIATION ONCOLOGY
Trimmed lattice < 6.6% of the original scan
-
Isocenter / Source PositioningIsocenter / Source Positioning
DEPT OF RADIATION ONCOLOGY
-
Energy Deposition Mesh TallyEnergy Deposition Mesh TallyEnergy Deposition Mesh TallyEnergy Deposition Mesh Tally
• Scores energy deposition from all particlesScores energy deposition from all particles• Independent of problem geometry
M V / 3 / ti l• MeV / cm3 / source_particle• Rectangular format (RMESH3)
DEPT OF RADIATION ONCOLOGY
-
Efficient Tally DefinitionEfficient Tally Definitiond
TALLY OUTPUT• Uncertainty matrix
“Filled” Lattice
slice thicknessUncertainty matrix• Energy matrix
MeV / cm3 / spvoxel height
1 ID the sub volume of interest
pPREPROCESSOR OUTPUT• Density matrix
voxel width
2. Define MESH tally elements as t i ll id ti l t l tti l t(M V / 3) ( / 3) (d )
1. ID the sub-volume of interestDensity matrix g / cm3
geometrically identical to lattice elements(MeV / cm3)i,j,k ÷ (g / cm3)i,j,k = (dose)i,j,k3. Further define MESH to occupy the
L MESH ffi i llDEPT OF RADIATION ONCOLOGY
same spaceLess MESH = more efficient tally
-
DEPT OF RADIATION ONCOLOGY
-
Simulation Results ProcessingSimulation Results Processing• Post-processing software linked with
voxelized phantom creationvoxelized phantom creation• Preprocessor generates:
• MCNP formatted input file• MCNP-formatted input file• Density matrix
CT d l i• CT underlay images• MCNP simulation creates:
Dose matrix
• Energy deposition matrix• Uncertainty matrix
DEPT OF RADIATION ONCOLOGY
-
Post ProcessingPost Processing• Compare simulation results with
somethingsomething– Film Measurement– Treatment Planning System Calculation– Treatment Planning System Calculation– Water Phantom profileGamma Analysis (Low et al )• Gamma Analysis (Low, et al.)–Usually 2D, can be 3D
DEPT OF RADIATION ONCOLOGY
-
10x10 cm10x10 cm22
DoseDose
DEPT OF RADIATION ONCOLOGYUncertainty
-
NormalizationNormalization
This voxel has relatively• low energy deposited?• low density• high energy deposited• equivalent density• low energy deposited• equivalent density
y
y• med-high dose• high uncertainty
q y• high dose• low uncertainty
q y• low dose• high uncertainty
DEPT OF RADIATION ONCOLOGY
-
Normalization Normalization MethodMethod
(energy)x y zMESH tally (density)x y z( gy)x,y,zy ( y)x,y,z
(dose)x,y,z(uncertainty)x,y,z
Max dose with sufficient uncertaintyDEPT OF RADIATION ONCOLOGY
Max dose with sufficient uncertainty
-
Source Model DevelopmentSource Model Development• Develop a clinically accurate Monte Carlo
source model of the Novalis®source model of the Novalis®
• Compare Monte Carlo simulation results against• Measurements• TPS calculations
DEPT OF RADIATION ONCOLOGY
-
Gamma AnalysisGamma Analysis• Low et al., Med Phys, 1998• Quantitative comparison of two dose distributions
Measured dose point (M)
Calculated dose point (C)Calculated Dose Distribution (C)Overlay Data Sets
Measured Dose Distribution (M)
z z’
IF the minimum Γ(rm,rc) > 1, THEN GREEN overlaySubset of both
data setsx x’
DEPT OF RADIATION ONCOLOGY
-
Comparative AnalysisComparative Analysis
• Gamma analysis• 2D or 3D capability
• CT image import and registration for g p gisodose “underlay”
• MCNP simulation compared with:MCNP simulation compared with:• Film measurements• TPS calculations• TPS calculations
DEPT OF RADIATION ONCOLOGY
-
Voxelized Models of Real PhantomsVoxelized Models of Real Phantoms
Source benchmarkingg• Monte Carlo• Measurement
DEPT OF RADIATION ONCOLOGY
-
Lookup Lookup TablesTablesHU RANGE
MATERIAL &DENSITY (g/cm3)HU RANGE MATERIAL g/cm3
≥ 796 2.1
200
CIRS AVG BONE1.600
616
436
256
76
BONE1.891.68
1.48
1.26200
CIRS PLASTIC WATER1 039
76
29
19
-29
MUSCLE
WATER
1.14
1.06
1.0
0 98
-224
1.039-94
-159
-224
-374
FAT
0.98
0.925
0.85
0.6455
CIRS LUNG0.210
374
-524
-674
-824
LUNG
0.4721
0.2978
0.1254
0.048
DEPT OF RADIATION ONCOLOGY
-954
-1024 AIR 0.0012-954
-1024 AIR
0.048
0.0012
-
Validate Validate Source and Phantom Source and Phantom Models Against MeasurementModels Against MeasurementModels Against MeasurementModels Against Measurement
“S lid t ” l b• “Solid water” slabs• Kodak EDR-2 Film• AP fields
DEPT OF RADIATION ONCOLOGY
-
Source Model vs. MeasurementsSource Model vs. Measurements
24x24 mm2 field24x24 mm field
DEPT OF RADIATION ONCOLOGY
Red line: filmBlack line: MC simulation
-
Source Model vs. MeasurementsSource Model vs. Measurements
IMRT segment
R d li fil
DEPT OF RADIATION ONCOLOGY
Red line: filmBlack line: MC simulation
-
Source Model vs. MeasurementsSource Model vs. Measurements
Composite IMRT field
DEPT OF RADIATION ONCOLOGY
Red line: filmBlack line: MC simulation
-
Source Model vs. MeasurementsSource Model vs. Measurements
• “Solid water” slabs• “Lung” slabs• Lung slabs
DEPT OF RADIATION ONCOLOGY
-
Source Model vs. MeasurementsSource Model vs. Measurements
Composite IMRT fieldComposite IMRT field
DEPT OF RADIATION ONCOLOGY
Red line: filmBlack line: MC simulation
-
DEPT OF RADIATION ONCOLOGY
-
Monte Carlo Patient Model
Monte Carlo Virtual Source
ModelAnalysisAnalysisTPS Simulation Film
yy
Calculation ResultsMeasurement
Γ analysis(Low et al.)
DEPT OF RADIATION ONCOLOGY
-
Monte Carlo Patient Model
Monte Carlo Virtual Source
ModelAnalysisAnalysisTPS Simulation Film
yy
Calculation ResultsMeasurement
DEPT OF RADIATION ONCOLOGY
-
Lung “Target”Lung “Target”
lung
wat
er
wat
er
lung
wat
erDEPT OF RADIATION ONCOLOGY
-
Patient SimulationsPatient Simulations
DEPT OF RADIATION ONCOLOGY
-
DEPT OF RADIATION ONCOLOGY
-
DEPT OF RADIATION ONCOLOGY
-
ConclusionsConclusions• Improvements in CT-based voxelized
phantom generationp g• Flexible geometry modifications
• Sub-sampling• Sub-sampling• Lattice trimming• Isocenter positioning• Isocenter positioning
• Mesh tally definition• Pre- and post-processing• Radiotherapy analysis (gamma)
DEPT OF RADIATION ONCOLOGY
py y (g )
-
ConclusionsConclusions“All models are wrong – some are useful.”
George Box-George Box• CT-based phantoms can be used for
lid ti di th d l ivalidating radiotherapy source models in preparation for and in addition to i ti ti
It’s easy to make a model
investigations.
It s easy to make a model……but the model needs to be accurate.
DEPT OF RADIATION ONCOLOGY
-
Ongoing / FutureOngoing / Future• Modeling techniques
used to supplement smallused to supplement small animal SBRT research
• Validate brachytherapytreatment planning p gsystem for heterogeneous dose gcalculations
DEPT OF RADIATION ONCOLOGY
-
Thank YouThank You
• John DeMarco, Ph.D.• Adam Kesner, Ph.D.
DEPT OF RADIATION ONCOLOGY