Clinical Implementation and Application of
Monte Carlo Methods in Photon & Electron
Dose Calculation – New Issues to Consider in
Clinical Practice
Neelam Tyagi, Ph.D.
Joanna E. Cygler, Ph.D.
2The
Ottawa Hospital Cancer Centre, Ottawa, Canada
University Dept. of Physics, Ottawa, Canada
4University of Ottawa, Dept. of Radiology. Ottawa, Canada
3Carleton
The Ottawa
L’Hopital
Hospital
d’Ottawa
Regional Cancer Centre
Part II: Electron beams
Joanna E. Cygler, Ph.D., FCCPM, FAAPM
The Ottawa Hospital Cancer Centre, Ottawa, Canada
Carleton University, Dept. of Physics, Ottawa, Canada
University of Ottawa, Dept. of Radiology, Ottawa, Canada
1
Outline
• Rationale for MC dose calculations for electron
beams
• Commercially available Monte Carlo based electron
treatment planning systems
• Clinical implementation of MC-based TPS
• Issues to pay attention to when using MC based
system
• Timing comparisons of major vendor MC codes in
the clinical setting.
Rationale for Monte Carlo dose
calculation for electron beams
• Difficulties of commercial pencil beam based
algorithms
– Monitor unit calculations for arbitrary
SSD values – large errors*
– Dose distribution in inhomogeneous media
has large errors for complex geometries
* can
be circumvented by entering separate virtual
machines for each SSD – labour consuming
2
Rationale for Monte Carlo dose
calculation for electron beams
6.2 cm
15
Relative Dose
9 MeV
10
Measured
Pencil beam
Monte Carlo
depth = 6.2 cm
depth = 7 cm
5
0
-10
/tex/E TP /abs/X TS K 09S .O R G
-5
0
5
Horizontal Position /cm
10
98-10-21
Ding, G. X., et al, Int. J. Rad. Onc. Biol Phys. (2005) 63:622-633
Commercial implementations
• MDS Nordion (now Nucletron) 2001
- First commercial Monte Carlo treatment planning for electron
beams
– Kawrakow’s VMC++ Monte Carlo dose calculation algorithm (2000)
– Handles electron beams from all clinical linacs
• Varian Eclipse eMC 2004
– Neuenschwander’s MMC dose calculation algorithm (1992)
– Handles electron beams from Varian linacs only (23EX)
– work in progress to include linacs from other vendors
• CMS XiO eMC for electron beams 2010
– Based on VMC (Kawrakow, Fippel, Friedrich, 1996)
– Handles electron beams from all clinical linacs
3
Nucletron Electron Monte Carlo
Dose Calculation Module
•Originally released as part of Theraplan Plus
•Currently sold as part of Oncentra Master Plan
•Fixed applicators with optional, arbitrary inserts, or
variable size fields defined by the applicator like
DEVA
•Calculates absolute dose per monitor unit (Gy/MU)
•User can change the number of particle histories
used in calculation (in terms of particle #/cm2)
•Data base of 22 materials
510(k) clearance (June 2002)
•Dose-to-water is calculated in Oncentra
•Dose-to-water or dose-to-medium can be calculated
in Theraplan Plus MC DCM
•Nucletron performs beam modeling
Varian Macro Monte Carlo
transport model in Eclipse
• An implementation of Local-to-Global (LTG) Monte Carlo:
– Local: Conventional MC simulations of electron transport performed
in well defined local geometries (“kugels” or spheres).
• Monte Carlo with EGSnrc Code System - PDF for “kugels”
• 5 sphere sizes (0.5-3.0 mm)
• 5 materials (air, lung, water, Lucite and solid bone)
• 30 incident energy values (0.2-25 MeV)
• PDF table look-up for “kugels”
This step is performed off-line.
– Global: Particle transport through patient modeled as a series of
macroscopic steps, each consisting of one local geometry (“kugel”)
C. Zankowski et al “Fast Electron Monte Carlo for Eclipse”
4
Varian Macro Monte Carlo
transport model in Eclipse
• Global geometry calculations
– CT images are pre-processed to
user defined calculation grid
– HU in CT image are converted to
mass density
– The maximum sphere radius and
material at the center of each
voxel is determined
• Homogenous areas → large
spheres
• In/near heterogeneous areas →
small spheres
C. Zankowski et al “Fast Electron Monte Carlo for Eclipse”
Varian Eclipse Monte Carlo
• User can control
– Total number of particles per simulation
– Required statistical uncertainty
– Random number generator seed
– Calculation voxel size
– Isodose smoothing on / off
• Methods: 2-D Median, 3-D Gaussian
• Levels: Low, Medium, Strong
• Dose-to-medium is calculated
5
CMS XiO Monte Carlo system
• XiO eMC module is based on VMC*
– simulates electron (or photon) transport through voxelized
media
• The beam model and electron air scatter functions
were developed by CMS
• The user can specify
–
–
–
–
–
voxel size
dose-to-medium or dose-to-water
random seed
total number of particle histories per simulation
or the goal Mean Relative Statistical Uncertainty (MRSU)
• CMS performs the beam modeling
*Kawrakow, Fippel, Friedrich, Med. Phys. 23 (1996) 445-457
*Fippel, Med. Phys. 26 (1999) 1466–1475
Software commissioning tests
• Criteria for acceptability
– Van Dyk et al, Int. J. Rad. Oncol. Biol. Phys., 26, 261-273,1993;
– Fraass, et al, AAPM TG 53: Quality assurance for clinical radiotherapy
treatment planning,” Med. Phys. 25, 1773–1829 1998
• Homogeneous water phantom
• Inhomogeneous phantoms (1D, 2D, 3D, complex)
– Cygler et al, Phys. Med. Biol., 32, 1073, 1987
– Ding G.X.et al, Med. Phys., 26, 2571-2580, 1999
– Shiu et al, Med.Phys. 19, 623—36, 1992;
– Boyd et al, Med. Phys., 28, 950-8, 2001
• Measurements, especially in heterogeneous phantoms,
should done with a high (1 mm) resolution
6
User input data for MC based TPS
Treatment unit specifications:
• Position and thickness of jaw collimators and MLC
• For each applicator scraper layer:
Thickness
Position
Shape (perimeter and edge)
Composition
• For inserts:
Thickness
Shape
Composition
No head geometry details required for Eclipse, since at this time it only
works for Varian linac configuration
User input data for MC TPS cont
Dosimetric data for beam characterization, as
specified in User Manual
• Beam profiles without applicators:
-in-air profiles for various field sizes
–in-water profiles
–central axis depth dose for various field sizes
–some lateral profiles
• Beam profiles with applicators:
– Central axis depth dose and profiles in water
– Absolute dose at the calibration point
Dosimetric data for verification
– Central axis depth doses and profiles for various
field sizes
7
Clinical implementation of MC
treatment planning software
• Beam data acquisition and fitting
• Software commissioning tests*
• Clinical implementation
– procedures for clinical use
– possible restrictions
– staff training
*should
include tests specific to Monte Carlo
A physicist responsible for TPS implementation should
have a thorough understanding of how the system works.
Software commissioning tests: goals
• Setting user control parameters in the TPS to
achieve optimum results (acceptable statistical
noise, accuracy vs. speed of calculations)
– Number of particle histories
– Required statistical uncertainty
– Voxel size
– Smoothing
• Understand differences between water tank and
real patient anatomy based monitor unit values
8
XiO: 9 MeV - Trachea and spine
Air
Bone
Bone
Air
Bone
Bone
Film
Film
Vandervoort and Cygler, COMP 56th Annual Scientific Meeting, Ottawa, June 2010
Lateral profiles at various depths,
SSD=100cm, Nucletron TPS
20 MeV, 10x10cm2 applicator, SSD=100cm.
Homogeneous water phantom. Cross-plane profiles at
various depths. MC with 10k and 50k/cm2.
110
110
100
100
90
90
80
80
70
meas.@2cm
70
60
calc.@2cm
50
meas.@3.0cm
calc.@3.0cm
40
Dose / cGy
Dose / cGy
9 MeV, 10x10cm2 applicator, SSD=100cm. Homogeneous
water phantom,cross-plane profiles at various depths. MC
with 10k/cm2.
meas.@d=3cm
calc.@d=3cm
calc.@d=3cm,50k
meas.@d=7.8cm
calc.@d=7.8cm
60
50
calc.@d=7.8cm,50k
meas.@d=9cm
calc.@d=9cm
calc.@d=9cm,50k
40
meas.@4.0cm
30
30
calc.@4.0cm
20
20
10
10
0
0
-10
-5
0
Off - axis / cm
5
10
-10
-5
0
5
10
Off - axis / cm
9
Monte Carlo Settings: Noise in the
distributions
Varying MRSU, voxel size=2.5×2.5×2.5 mm3, dose-tomedium, 6 MeV beam, 10x10 cm2 applicator
MRSU=10%
MRSU=5%
Histories=1.2x106
MRSU=2%
Histories=2.8x106 Histories=1.6x107
CMS eMC: Cut-out factors
Cutout Output Factors: 17 MeV
Cutout Output Factors: 9 MeV
1.050
1.050
O u tp u t F a c to r (c G y /M U )
SSD=100 cm
O u tp u t F a c to r (c G y /M U )
1.000
0.950
0.950
0.850
SSD=100 cm
0.900
0.750
0.850
0.800
0.650
SSD=115 cm
0.750
0.550
0.700
Experimental
XiO Calculated
0.450
SSD=115 cm
0.650
Experimental
XiO Calculated
0.600
0.350
1
2
3
4
5
6
Square Cutout Length (cm)
7
8
9
1
2
3
4
5
6
7
8
9
Square Cutout Length (cm)
Vandervoort and Cygler, COMP 56th Annual Scientific Meeting, Ottawa ,June 2010
10
Eclipse eMC no smoothing
Voxel size = 2 mm
Air
Air
Bone
Bone
120
120
depth = 4.7 cm
18 MeV
110
110
100
90
90
Relative Dose
100
Relative Dose
4.7 cm
80
depth = 6.7 cm
70
60
50
depth = 7.7 cm
18 MeV
depth = 4.7 cm
80
70
60
50
40
40
30
20
Measured
eMC
30
Measured
eMC
20
10
10
0
0
-6
-4
-2
0
2
4
6
-6
Off-axis X position /cm
-4
-2
0
2
4
6
Off-axis Y position /cm
Ding, G X., et al (2006). Phys. Med. Biol. 51 (2006) 2781-2799.
Eclipse eMC
Effect of voxel size and smoothing
Air
Air
Bone
Bone
110
2 mmand no smoothing
18 MeV
110
Relative Dose
100
90
80
70
2 mmand with 3D smoothing
60
5 mm and with 3D smoothing
50
120
Relative Dose
120
80
70
60
50
40
30
depth = 4.9 cm
10
0
0
-4
-2
0
2
2 mm and with 3D smoothing
20
10
Off-axis X position /cm
5 mm and with
3D smoothing
90
30
20
4
6
2 mmand no smoothing
18 MeV
100
40
-6
4.7 cm
depth = 4.9 cm
-6
-4
-2
0
2
4
6
Off-axis Y position /cm
Ding, G X., et al (2006). Phys. Med. Biol. 51 (2006) 2781-2799.
11
Dose-to-water vs. dose-to-medium
Hard bone cylinder
2cm
1 cm diameter and 1 cm length
1100
Small volume
of water
Bone cylinder is replaced
by water-like medium but
with bone density
100
90
BEAM/dosxyz
simulation
80
70
Dose
Dm - energy absorbed in
a medium voxel divided
by the mass of the
medium element.
60
50
Bone
cylinder
location
40
30
20
10
Voxel of medium
Dw - energy absorbed in
a small cavity of water
divided by the mass of
that cavity.
0
0
1
2
3
4
5
Central Axis Depth /cm
1.14
9 MeV
SPR
1.13
w
⎛S⎞
Dw = Dm ⎜ ⎟
⎝ ρ ⎠m
1.12
Water/Bone stopping-power ratios
1.11
1.10
0
Ding, G X., et al Phys. Med. Biol. 51 (2006) 2781-2799.
1
2
3
4
5
depth in water /cm
Dose-to-water vs. Dose-to-medium
Dose-to-water vs. dose-to-medium,
MRSU=2%, voxel size=4×4×4 mm3, 6 MeV
beam, 15x15 cm2 applicator, both 602 MU
DTM
DTW
DTWDTW-DTM
12
Good clinical practice
• Murphy’s Law of computer software (including
Monte Carlo)
“All software contains at least one bug”
• Independent checks
MU MC vs. hand calculations
Monte Carlo
Hand Calculations
Real physical dose
calculated on a patient
anatomy
Rectangular water
tank
Inhomogeneity
correction included
No inhomogeneity
correction
Arbitrary beam angle
Perpendicular beam
incidence only
13
9 MeV, full scatter phantom
(water tank)
RDR=1 cGy/MU
Lateral
scatter missing
Real contour / Water tank =
=234MU / 200MU=1.17
14
MU real patient vs.water tank
MC / Water tank= 292 / 256=1.14
MU-real patient vs. water tank:
Impact on DVH
120
PTV-MU-MC
100
PTV-MU-WT
%volume
80
LT eye-MU-MC
LT eye-MU-WT
60
RT eye-MU-MC
40
RT eye-MU-WT
20
0
0.0
10.0
20.0
30.0
40.0
50.0
60.0
dose / Gy
15
Posterior cervical lymph node
irradiation - impact on DVH
45.0
customized
40.0
35.0
30.0
PTV / cm
3
conventional
25.0
20.0
15.0
10.0
Jankowska et al, Radiotherapy & Oncology, 2007
5.0
0.0
0.0
5.0
10.0
15.0
20.0
25.0
30.0
dose / Gy
Internal mammary nodes
MC / Water tank= 210 / 206=1.019
16
How long does it take?
• MC gives entire distribution, not just a few points
• time for N beams is the same as for 1 beam
• timing is a complex question since it depends on
– statistical uncertainty and how defined
– voxel size
– field size
– beam energy and whether photons or electron
– accuracy sought
– speed of CPU and optimization of compiler
- complexity of patient specific beam modifiers
Monte-Carlo Settings: Effect on
computation time
Timing Results XiO:
30
-2.1
y = 6.4x
25
9 MeV 2.5 mm voxel
For 9 and 17 MeV beams,
time / min
17 MeV 2.5 mm voxel
20
y = 3.4x-2.0
15
17 MeV 5 mm voxel
10x10 cm2 applicator and the
9MeV 5 mm voxel
trachea and spine phantom,
timing tests were performed
10
for a clinical XiO Linux
-2.0
y = 0.7x
5
workstation, which employs 8
y = 0.4x-2.0
0
0
0.5
1
1.5
MRSU %
2
2.5
processors, 3 GHz each, with
8.29 GB of RAM.
17
Timing – Pinnacle3
dual processor 1.6 GHz Sun workstation, 16 GB RAM.
Overall uncertainty
2%
1%
0.5%
Patient
#
histories
CPU time
(min)
#
histories
CPU time
(min)
# histories
CPU time
(h)
1(cheek)
3.4x106
4.8
1.4x107
20
1.6x108
3.9
2 (ear)
1.7x106
2.1
6.5x106
8.1
7.1x107
1.5
3 (breast)
3.3x106
7.1
1.4x107
29.9
1.5x108
5.4
4 (face)
1.1x107
32.1
4.7x108
134.5
5.2x108
24.5
Fragoso et al.: Med. Phys. 35, 1028-1038, 2008
Timing – Nucletron TPS
Oncentra 4.0
Anatomy - 201 CT slices
Voxels 3 mm3
10x10 cm2 applicator
50k histories/cm2
4 MeV Timer Results:
Init = 0.321443 seconds
Calc = 42.188 seconds
Fini = 0.00158201 seconds
Sum = 42.5111 seconds
20 MeV Timer Results:
Init = 0.311014 seconds
Calc = 110.492 seconds
Fini = 0.00122603 seconds
Sum = 110.805 seconds
Faster than pencil beam!
18
Timing – Varian Eclipse
Eclipse MMC, Varian single CPU Pentium IV
XEON, 2.4 GHz
10x10 cm2, applicator, water phantom,
cubic voxels of 5.0 mm sides
6, 12, 18 MeV electrons,
3, 4, 4 minutes, respectively
Chetty et al.: AAPM Task Group Report No. 105: Monte Carlobased treatment planning, Med. Phys. 34, 4818-4853, 2007
Summary - electron beams
• Commercial MC based TP systems are available
– fairly easy to implement and use
– MC specific testing required
• Fast and accurate 3-D dose calculations
• Single virtual machine for all SSDs
• Large impact on clinical practice
– Accuracy of dose calculation improved
– More attention to technical issues needed
– Dose-to-medium is calculated, although some systems calculate
dose-to-water as well
– MU based on real patient anatomy (including contour
irregularities and tissue heterogeneities)
• Requirement for well educated physics staff
19
New clinical dilemma
• Accurate dose calculation systems
– at last we know what dose we are giving to
tumors and OAR
• How are we to incorporate this knowledge
into clinical practice?
– Will we change the dose prescription?
– We have done it for photon beams and
inhomogeneity corrections in lungs…
Discussion: dose prescription issues
• MC-calculated doses can in some instances be
significantly different (10-20%) from conventional
algorithms, such as radiological pathlength, and
convolution-based methods or pencil beam in case of
electron beams
• In light of these differences:
- should dose prescriptions change with MC-based
calculations ?
- How?
20
Dose prescription - AAPM TG-105
perspective
•
MC method is just a more accurate dose algorithm
•
Dose prescription issues are not specific to MC-based
dose calculation
•
As with other changes to the therapy treatment
process, users should correlate doses and prescriptions
with respect to previous clinical experience
Conclusions
• Clinical implementation of MC-based systems must be
performed thoughtfully and users, especially physicists,
must understand the differences between MC-based and
conventional dose algorithms
• Successful implementation of clinical MC algorithms
requires strong support from the clinical team and an
understanding of the paradigm shift with MC algorithms
• A properly commissioned MC-based dose algorithm will
improve dose calculation accuracy for electron and
photon beams
• More accurate dose calculations may improve dosebiological effect correlations
21
Acknowledgements
George X. Ding
George Daskalov
Gordon Chan
Robert Zohr
Ekaterina Tchistiakova
Indrin Chetty
Margarida Fragoso
Charlie Ma
Eric Vandervoort
David W.O. Rogers
In the past I have received research support from
Nucletron and Varian.
TOHCC has a research agreement with Elekta.
I hold a research grant from Elekta
Thank You
22