Model-Based Dose Calculation Algorithms in Brachytherapy  

Luc  Beaulieu  Professor,  Department  of  Physics,  Université  Laval    

Medical  Physicist  and  Head  of  Research,  Department  of  Radia>on  Oncology,  CHU  de  Quebec  

Contents •  Introduction: TG43 and beyond…

•  Advances in brachytherapy dose calculations

•  Recommendations from AAPM/ESTRO/ABS/ABG Task Group 186

•  Conclusion

Brachytherapy is state-of-the-art •  Exquisite dose distribution and intensity modulation

•  Dose deposition "kernel" better than proton

•  Real-time image guidance and dose guidance

•  Addition of robotic brachytherapy

•  Possibility of shielding, directional source, multiple isotope/energie tx

…but dose calculation is not

≠

Interstitial Contura

Mammo SAVI

One size does not fit all!

Vision 20/20 Paper Medical Physics The evolution of brachytherapy treatment planning Mark Rivard,1 Jack L. M. Venselaar,2 and Luc Beaulieu3 1Department of Radiation Oncology, Tufts University School of Medicine, Boston, Massachusetts, USA 2Department of Medical Physics, Instituut Verbeeten, P.O. Box 90120, 5000 LA Tilburg, The Netherlands 3Département de Radio-Oncologie et Centre de Recherche en Cancérologie de l’Université Laval, Quebec

Brachytherapy is a mature treatment modality that has benefited from technological advances. Treatment planning has advanced from simple lookup tables to complex, computer-based dose calculation algorithms. The current approach is based on the AAPM TG-43 formalism with recent advances in acquiring single-source dose distributions. However, this formalism has clinically relevant limitations for calculating patient dose. Dose-calculation algorithms are being developed based on Monte Carlo methods, collapsed cone, and the linear Boltzmann transport equation. In addition to improved dose-calculation tools, planning systems and brachytherapy treatment planning will account for material heterogeneities, scatter conditions, radiobiology, and image guidance. The AAPM, ESTRO, and other professional societies are coordinating clinical integration of these advancements. This Vision 20/20 article provides insight on these endeavors. Med. Phys. 36, 2136-2153 (2009)

Sensitivity of Anatomic Sites to Dosimetric Limitations of Current Planning Systems

anatomic site

photon energy

absorbed dose attenuation shielding scattering beta/kerma

dose

prostate high low XXX XXX XXX

breast high XXX low XXX XXX XXX

GYN high XXX low XXX XXX

skin high XXX XXX low XXX XXX XXX

lung high XXX XXX low XXX XXX XXX

penis high XXX low XXX XXX

eye high XXX XXX XXX low XXX XXX XXX XXX

Rivard,  Venselaar,  Beaulieu,  Vision  20/20,  Med  Phys  36,  2136-­‐2153  (2009)  

Importance of the Physics: Water vs Tissues

< 100 keV large differences

Impact of tissue composition: 192Ir

Importance of the Physics: Attenuation by Metals

10−3 10−2 10−1 100 101100

101

102

103

104

Photon Energy (MeV)

Atte

nuat

ion

coef

ficie

nt (r

atio

to w

ater

)

WCuAgAuTi

From NIST website

http://physmed.fsg.ulaval.ca/ 12

Poon et al., IJROBP (2008)

How important in the clinic?

Site / Application Importance

Shielded Applicators Huge

Eye plaque -10 to -30% (TG129)

Breast Brachy -5% to -40%

Prostate Brachy -2 to -15% on D90

GYN Depends on applicators

Accurate dose calculation should be a priority

•  The dose-outcomes, tolerence doses, prescription doses will probably need to be revisited

•  e.g. 192Ir breast skin tolerance dose •  -16% Raffi et al, Med Phys 2010

Rule of tumb

Energy Range Effect

192Ir Scatter condition

Shielding (applicator related) 103Pd/125I/eBx Absorbed dose (µen/ρ)

Attenuation (µ/ρ)

Shielding (applicator, source)

Contents •  Introduction: TG43 and beyond…

•  Advances in brachytherapy dose calculations

•  Recommendations from AAPM/ESTRO/ABS/ABG Task Group 186

•  Conclusion

TG43 PSS CCC MC

Brachytherapy Dose Calculation Methods

GBBS Physics  Content  

Analytical  /  Factor-­‐based   Model-­‐Based  Dose  Calculation  :  MBDCA  

Rivard,  Beaulieu  and  Mourtada,  Vision  20/20,  Med  Phys  2010  

TG43 PSS CCC MC

BT Dose Calc.

GBBS

Current  STD:  Full  scatter  water  medium  

No  particle  transport.  No  heterogeneity,  shields.  Primary  can  be  used  in  more  complex  dose  engine  

Implicit  particle  transport:  Heteregoneities.    Accurate  to  1st  scatter.  GPU  friendly  

Only  commercial  MDBCA.  Solves  numerically  transport  equtations.  Full  heteregoneities.        

Explicit  particle  transport  simulation.  Gold  STD  for  source  characterization  and  other  applications  

TG43 PSS CCC MC

BT Dose Calc.

GBBS

First-scatter kernel

Outline of the brachy-CC MBDCA IV. Summation I. Raytrace source II. CC convolution III. CC convolution

S1sc S2sc

Dprim D1sc Drsc + + = Dtot  

Scatter transport line Residual-scatter kernel

-4

-12

log(D/R) 192Ir msel-v2

Details in Carlsson and Ahnesjö (2000) Med Phys p 2320-2332

∝1sc prim

CPES D ∝2sc 1sc

CPES D

Scatter transport line

First scerma Second scerma

Primary source rays Material info

Material info Material info

I.  TG43  

Superposition of sngle-source water-dose Imaging in TG43: localise dose -anatomy

Dm,m Collapsed Cone Dw-TG 43

II.  MBDCA  

Information on tissue, etc composition from images or elsewhere

water

Brachy-CC MBDCA

From Åsa Carlsson-Tedgren

Monte Carlo simulations

•  Mimics  the  discrete  par>cle,  sta>s>cal  nature  of  ioniza>on  radia>on  

•  "Golden  standard"  for  dose  calcula>ons  •  TG43  parameters  •  Primary  ScaKer  Separa>on  

•  Model  complex  geometries    

•  Derive  informa>on  not  accessible  in  measurements    

DWO Rogers, Review paper, PMB 51 (2006); TG43-U1 by Rivard et al., Med Phys 2004;

Monte Carlo Dose Calculations: Brachy

Williamson (1987) Med Phys p 567-576, Hedtjärn et al (2002) Phys Med Biol p 351-376

•  General Purpose •  EGSnrc •  MCNP (5,X) •  Penelope •  Geant4

•  Brachytherapy specific •  MCPI – Seeds (Chibani and Williamson (2005) Med Phys

3688-3698) •  BrachyDose - Seeds (Taylor et al (2007) Med Phys

445-457) •  PTRAN CT (Williamson et al (1987) Med Phys p 567-576) •  ALGEBRA (Afsharpour et al., (2012), PMB)

2D: Daskalov et al (2002), Med Phys 29, p.113-124 3D: Gifford et al (2006), Phys Med Biol vol 53, p 2253-2265

–  Position: mesh position discretization (finite elements)

–  Energy: E Energy bins (cross section) –  Direction: Angular discretization

Ω̂ ⋅∇Ψ(r ,E,Ω̂) +σ t (

r ,E)Ψ(r ,E,Ω̂) =Qscat (r ,E,Ω̂) +Qex (r ,E,Ω̂)

«  multi-­‐group  discrete  ordinates  grid-­‐based  …»  

r = (x, y, z)

Ω̂ = (θ ,φ)

Grid-­‐Based  Boltzmann  Solver  (GBBS)  

•  Varian BV-Acuros® implementation: only commercial MBDCA solution at this time •  CPE assumption : Primary dose analytical (ray-tracing

with scaling) • Dprim = Kcoll • First scatter from primary : Scerma = Dprim•((µ-µen)/uen) • Share this step with CCC

•  3D scatter integration through GBBS

•  Source modeling done in Atilla® (Transpire Inc)

Grid-­‐Based  Boltzmann  Solver  (GBBS)  

Example

Figure  from  :  L.  Petrokokkinos  et.  al.  Med.  Phys.  38,  1981-­‐1992  (2011).  .  More  references  on  the  algorithm,  see  e.g.:    K.  A.  Gifford  et.  al.  Med.  Phys.  35,  2279-­‐2285  (2008)  

•  Speed: 40 sec to 12 min depending on complexity

Factor-based vs Model-based

Superposition of data from source characterization

Dw-TG43

Dm,m Dw,m

Source characterization

Tissue/applicator information

Source characterization

INPUT OUTPUT CALCULATION

TG43

MBDC

INPUT OUTPUT CALCULATION

From Åsa Carlsson-Tedgren

Model-Based Dose Calculation Algorithms

Contents •  Introduction: TG43 and beyond…

•  Advances in brachytherapy dose calculations

•  Recommendations from AAPM/ESTRO/ABS/ABG Task Group 186

•  Conclusion

TG-186 •  For early adopters

•  Report approved by •  AAPM (BTSC, TPC) •  ESTRO (BRAPHYQS, EIR) •  ABS (Physics committee) •  ABG

•  Published in Medical Physics

1.  Definition of the scoring medium

2.  Cross section assignments

(segmentation)

3.  Specific commissioning process

Three main areas identified as critical

Dx,y  x: dose specification

medium  

y: radiation transport medium  

1. Definition of the scoring medium

�  x,y: Local medium (m) or water (w)

!

FROM: G Landry, Med Phys 2011

Which dose to report? •  Dm,m is inherently computed by Model-based algorithms •  Dm,m must be reported along with TG43 •  Dw,m can also be reported but method must be specified:

•  e.g. large cavity theory, small cavity theory •  Could be energy and target size dependent (voxel, cells, …)

!

#1: #2:

2- Cross section assignments � Accurate tissue segmentation, sources and

applicators needed: identification (ρe ,Zeff) �  e.g. in breast: adipose and glandular tissue have

significantly different (ρe ,Zeff); dose will be different

�  If this step is not accurate è incorrect dose �  Influences dosimetry and dose outcome studies �  Influences dose to organs at risk

Recommendation - segmentation � Extract  electron  density  from  CT  calibra>on  (see  TG53,  TG66  …)    �  Use  the  density  from  CT  for  each  voxel  

�  Use  recommended  >ssue  composi>ons  �  Organ-­‐based  (contoured)  assignments  

�  Prostate  from  Woodard  et  al,  BJR  59  (1986)  1209-­‐18  �  All  others  from  ICRU-­‐46  composi>on  

 

�  From  CT  calibra>on:  breast,  adipose,  muscle  and  bone  

Recommendation - segmentation � If artifacts (e.g. from metals)

�  Override the density using the recommended default organ/tissue density (TG-186 table)

�  Assign tissue composition based on organ contours

Recommendation - segmentation �  If no CT (US and MRI)

� Use contoured organs with recommended tissue compositions �  For 192Ir, water is a good approximation for soft tissues only. �  Air, lung, bone, … should be assigned correctly

�  Use accurate source and applicators geometry and

composition

3- Specific commissioning process

� MBDCA specific tasks �  Currently, only careful comparison to Monte Carlo with or

w/o experimental measurements can fully test the advanced features of these codes

�  This is not sustainable for the clinical physicists

Recommendation - Commissioning

•  Two parts process

•  Level 1: MBDCA should fall back to TG43 in well controlled conditions •  Full scatter: R-r ≥ 5 cm or 20 cm •  All water

192Ir Test Geometry for MBDCA Water

20 cm at least

LEVEL 2

� MBDCA specific tasks �  Monte Carlo remains the gold standard for comparison

�  Might not be appropriate for all clinics

Need  Standardized  MBDCA  Benchmarks  � Excellent reference HDR 192Ir benchmarks in MedPhys

� Acuros BrachyVision

Petrokokkinos et al., MedPhys 38, 1981-1992 (2011)

3- Specific commissioning process

� AAPM/ESTRO/ABG working group to tackle this issue Luc Beaulieu (chair), Frank-André Siebert (vice-chair) Facundo Ballaster, Åsa Carlsson-Tedgren, Annette Haworth, Goeff Ibbott, Firas Mourtada, Panagiotis Papagiannis, Mark J Rivard, Ron Sloboda and Frank Verhaegen. Strategy: Registry of validated cases with reference doses calculations. We will try to involved the vendor as much as possible to make the cases compatible with all TPS.

TG-186 Recommendations •  The full TG-186 report has a detailed rational

supporting the various recommendations

•  Following the recommendations should ensure uniformity of implementation across centers

•  NOTE: there is one MBDCA commercial system and it is for 192Ir only at this time.

Clinical relevance of MBDCA � More accurate dose calculation

�  Impact on prescription, dose to OARs, … �  Current dose-outcome relationships could be wrong!

� Enable better treatment approaches �  Directional sources, mixed sources, shielded

applicators, …

Conclusion •  Advanced dose calculation is a necessary step for

better brachytherapy treatments

•  Change in dose calculation standard is not new (e.g. lung EBRT)

•  Transition period •  Revisiting dose-outcomes, dose prescription

•  The future of brachytherapy is exciting

Merci!

beaulieu@phy.ulaval.ca http://physmed.fsg.ulaval.ca