4. Here we will discuss about
• Physics aspects of Acuros-XB( AXB) algorithm
• How it is comparable accuracy to Montecarlo simulation for
full range of X-ray beams from 4-25 MV especially Lung ,
bone, air and non biological implants.
5. • AXB uses sophisticated technique( Numerical methods) to
solve LBTE explicitly and directly accounts for the effects of
heterogeneities in patient dose calculation.
• MC don’t solve LBTE explicitly, indirectly obtain the solution.
6. • MC simulates a finite number of particles interacting with
medium, errors are random and having statistical noise.
• AXB simulates infinite number of particles ,absence of
statistical noise and errors are systematic and are results from
discretization of the variables in space, angle and energy.
Larger steps in discretization process result in a faster
solution but less accuracy.
7. AXB calculates 3D patient dose deterministically using
4 components
• Primary photon source model
• Scattered photon fluence
• Scattered electron fluence
• Dose calculation :
9. • AXB explicitly models the physical interaction of radiation in
material and solves the radiation transport problem
numerically( Chebyshev-Legendre quadrature).AXB cross
section library includes 25 photons & 49 e groups.
• It uses adaptive angular quadrature that varies by both
particle type and energy.
10. Computational grid
Computational grid in AXB is spatially variable, the local element
size is adapted and provides a rigorously defined solution at
every point in computational domain.
11. Cut off energy :
It employs a spatial transport cutoff for electron
energies <500Kev and for photon energies <1Kev in
the dose grid.
12. Grid Voxel size: up to 3mm
The choice of grid size is based on dimension of the planning geometry. Smaller grid sizes are
normally used for smaller planning target volumes ,for smaller grid size of the dose matrices
becomes larger and total computation times become larger.
The resolution of dose calculation corresponds to the defined grid size, along Z axis AXB
automatically sets grid resolution or closest to user defined grid size
14. • Material assignment:
AXB explicitly model the physical interaction of photon in the
material ,requires chemical composition of each material in
its computational grid to perform dose calculation.
15. From CT calibration curve ,Eclipse assigns chemical composition for each voxel from
Varian system database with density up to 3g/cc (Bone)for biological & 16 non
biological materials of density up to 8g/cc (Steel).
If the densities are more than 3g/cc for biological & 8g/cc for non biological material
requires user assignment of material.
16. Dose reporting:
Dose to medium in medium(Dm,m):
Material for each voxel in the patient image is assigned
according to that voxel CT # ,
AXB uses energy dependent response function is based on the
material properties of that Voxel
Dose to water in medium(Dw,m):Water based response function
Dose to water in water (Dw,w):Heterogeneity is off
17. 6MV,5X5 cm2water bone lung slab phantom 18MV,5X5cm2 on multi layered slab phantom
From the above graphs e- transport is same but deposition of elec energy is different.
This differences highlight the material significance of using actual composition.
18. Calculation time:
• Plan dose calculation: Steps in AXB
1.Transport of source model fluence in to the patient
2.Calculation of scattered photon fluence in the patient
3.Calcultion of scattered e- fluence in the patient
4. Dose calculation
19. Calculation time
Plan dose calculation: Steps in AXB
1.Transport of source model fluence in to the patient
2.Calculation of scattered photon fluence in the
patient
3.Calcultion of scattered e- fluence in the
patient
4. Dose
calculation
20. • Calculation time has very weak dependence on the # fields
,since majority of calculation time is spent on calculating
scattered photon & e- fluence which are performed only once
for all beams in the plan.
• As a result relative calculation speed of AXB increases with
increasing # fields in the plan, when a separate AXB
calculation is performed for each field scattered calculation
phase has to run for every field which significantly increases
calculation time.
21.
22. Validation
AXB Vs MC:
MCNPX (N particle extended) computations run with very large
no of particles to create results that were very smooth and
w/o statistical uncertainties that may have influence the
validation of AXB ,in practice MC results are much less
smooth & statistical uncertainties are clearly visible.
25. Depth dose curves & profiles of high density material of
2cm3 inserted on a phantom
26. Depth dose curves & profiles of high density material of
2cm3 inserted on a phantom
Both codes are in close agreement even in high gradient e- disequilibrium region
surrounded the implant
18x,10X10cm2
27. Depth dose curves on a slab phantom containing air block
2X2 Cm2,6MV, 2x2x10cm3 air block containing water phantom
31. Comparing depth dose & dose profiles of photon beams in both
homogeneous & heterogeneous multilayered phantom the dose
agreement of AXB to MC for almost voxels is within 2% indicating
that AXB was comparable to MC for dose prediction
34. Ojala et al, done Vmat optimization for pelvic case in which hip is replaced by implant
GI is 99.01%
AXB MC
35. Conclusion
• AXB provides comparable accuracy in treatment
planning conditions to MC for full range of X-
ray photon beams.AXB balance of both accuracy
and computation speed.
