Post on 23-Apr-2020
transcript
Treatment planning for Pencil Beam Scanning
Tony Lomax, Alessandra Bolsi, Francesca Albertini,
Damien Weber
Paul Scherrer Institute, Switzerland
1. Field shaping and op0misa0on 2. Plan design
3. Lateral penumbra and PBS
4. Range and posi0oning uncertain0es 5. Mo0on
Overview of presentation
1. Field shaping and op0misa0on 2. Plan design
3. Lateral penumbra and PBS
4. Range and posi0oning uncertain0es 5. Mo0on
Overview of presentation
Single field, uniform dose (SFUD) planning
The combination of individually optimised fields, each of which
deliver a (more or less) homogenous dose across the target volume
SFUD is the spot scanning equivalent of treating with ‘open’ fields.
Field shaping and optimisation
SFUD planning
Spot definition
Incident field Incident field
Spot selection
Selected spots
Initial dose distribution
Dose calculation
Spot weight optimisation
Optimised dose
Dose Calculation
Field shaping and optimisation
An example SFUD plan.
Note, each individual field is homogenous across the target volume
F1 F2
F3 F4
Combined distribution
Field shaping and optimisation
1st series (0-40CGE)
3 field SFUD plan to PTV
2nd series (40-74CGE)
3 field SFUD plan to
‘TechPTV’
Full treatment + =
An example SFUD plan
Field shaping and optimisation
The TechPTV or ‘Virtual 3d block’
In order to carve-out dose to neighbouring critical
structures, need to be able to ‘block’ out dose
Modified target volume used to define ‘Virtual 3d blocks’
Alternatively, dose can be spared to critical structure by
Single Field Optimisation (SFO)
Field shaping and optimisation
Intensity Modulated Proton Therapy (IMPT)
The simultaneous optimisation of all Bragg peaks from all fields (with or
without additional dose constraints to neighbouring critical structures)
IMPT is the spot scanning equivalent of IMRT (and field patching for passive
scattering proton therapy).
Field shaping and optimisation
An example IMPT plan
F1 F2
F3 F4
Combined distribution
Note, each individual field is highly in-homogenous (in dose) across the target volume (c.f. SFUD plans)
Field shaping and optimisation
Example clinical IMPT plans delivered at PSI
Skull-base chordoma
4 fields
3 field IMPT plan to an 8 year old boy
3 fields
Field shaping and optimisation
Two, 5 field IMPT dose distributions
A B
Corresponding spot weight distributions from field 2
3D IMPT DET
There’s more than one way to optimise an IMPT plan…
Field shaping and optimisation
In-fi
eld
inte
nsity
mod
ulat
ion
Technique
DET
Degeneracy: A continuum of modulation*
* Inspired by Stephen Dowdell, MGH, Boston
SFUD IMPT
Field shaping and optimisation
2500 spots per field 150 spots per field 120 spots per field
SFUD/IMPT DET Beyond DET…
Degeneracy: Beyond DET... Field shaping and optimisation
Multiple Criteria Optimisation (MCO)
Pareto surface (David Craft,
MGH)
Modulation
Chen et al 2010, Med. Phys. 37:4938-4945
Field shaping and optimisation
1. Field shaping and op0misa0on 2. Plan design
3. Lateral penumbra and PBS
4. Range and posi0oning uncertain0es 5. Mo0on
Overview of presentation
Geometric avoidance of organs at risk.
The selection of beam incidences which avoid critical structures leads…
…‘automatically’ to reduced doses to the critical structures
Plan design
…and 19%
For same mean dose to target, 15MV photons deliver an integral dose of….
16%…
The corresponding values for two proton fields are..
7%… …and 13%
Field selection and integral dose – protons vs photons
Plan design
Avoidance of coarse density heterogeneities.
• Accuracy of dose calculations
• Effects on dose homogeneity and conformity
• Sensitivity of a plan to spatial delivery uncertainties.
Plan design
Vol = 86%
Vol = 99%
A ‘homogenous’ field direction
Analytical Dose difference (MC-RC %)
0 5 10 -10 -5 10 20 30 40 50
Volu
me
(%)
Difference histogram
MC
Dose difference (MC-RC %) 0 5 10 -10 -5
10 20 30 40 50
Volu
me
(%)
Analytical
Difference histogram Vol = 64%
Vol = 89%
MC
An ‘inhomogenous’ field direction
Plan design
Effects on (single field) dose conformity
Example field through relatively homogenous
anatomy
Example field through very inhomogenous
anatomy
Plan design
Plan design
The importance of multiple fields
• Improve dose homogeneity
• Reduce dose to normal tissues (but increase irradiated volume)
• Improve plan robustness
Plan design The importance of multiple fields
Improved plan homogeneity and reduced dose to normal tissues
Plan design
Albertini et al 2011 PMB 56:4399-4413, Lowe et al 2015 manuscript in preparation
The importance of multiple fields
Improved plan robustness
Dose error bars for frac0onated posi0oning errors
σ85% = 3.3mm
Frac0ons = 23
1. Field shaping and op0misa0on 2. Plan design
3. Lateral penumbra and PBS
4. Range and posi0oning uncertain0es 5. Mo0on
Overview of presentation
~6MV photons
The problem of lateral fall-off for PBS
Safai et al 2008 PMB 21;1729-1750
Passive scaHering (collimated)
PBS (uncollimated)
Lateral penumbra
The problem of superficial spots… • Superficial spots (ranges < ~5 cm) require low energies (< ~80MeV) ~5cm
Typical solution is to use a preabsorber. E.g…
5cm (wer)
Nozzle Patient
• Low energy protons are difficult to transport through the beamline/gantry
• Low energy Bragg peaks are very sharp
Lateral penumbra
The problem of superficial spots… • However, scattering in the absorber broadens the beam…
+
Air gap
Air gap
• …and the beam broadens as a function of the gap between absorber and patient
When using a preabsorber then, the nozzle-patient
distance must be minimised to reduce beam size.
This is not always possible…
Lateral penumbra
Beam widths (σ in air) for PSI Gantry 2 at iso-centre
The problem of the preabsorber…
Without preabsorber 80 MeV, without
preabsorber ~ 4.7mm
80 MeV, with preabsorber ~ 10.5mm
Lateral penumbra
Lateral profiles
-20
0
20
40
60
80
100
120
-10 -8 -6 -4 -2 0 2 4 6 8 10
X coordinate (cm)
Dos
(%)
80-20% Fall offs: No absorber: 8mm
Absorber (9cm gap): 12mm
Absorber (27cm gap): 16mm
No absorber
Absorber (9cm) Absorber (27cm)
The problem of the preabsorber… Effect on lateral fall-off
No absorber
Absorber (9cm gap)
Absorber (27cm gap)
Geometric box calculations
Lateral penumbra
Are superficial spots important?
An orbital rhabdomyosarcoma (pediatric case)
• Critical structure - lacrimal gland
• Maximum range per field: ~5cm
Lateral penumbra
Spot-range histogram
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
Range (mm)
Per
cent
age
of B
ragg
pea
ks
Spot-range histogram
Cumulative BP-range frequency plot
calculated from 3829 fields applied
at PSI
Are superficial spots important?
More than 30% of delivered spots have ranges of <=5cm
Lateral penumbra
1. Field shaping and op0misa0on 2. Plan design
3. Lateral penumbra and PBS
4. Range and posi0oning uncertain0es 5. Mo0on
Overview of presentation
The advantage of protons is that they stop.
The disadvantage of protons is that we don’t always know where…
10% range error
The influence of range uncertainties
Range and positioning uncertainties
Potential magnitude
Beam energy [σ]
Patient positioning [σ]
Inherent CT uncertainties (beam hardening, calibration etc) [Σ]
Distal end RBE enhancements [Σ]
CT artifacts [Σ]
Variations in patient anatomy [Σ,σ]
The ‘Bermuda Triangle’ of range uncertainties
Range and positioning uncertainties
Potential magnitude
Beam energy [σ]
Patient positioning [σ]
Inherent CT uncertainties (beam hardening, calibration etc) [Σ]
Distal end RBE enhancements [Σ]
CT artifacts [Σ]
Variations in patient anatomy [Σ,σ]
The ‘Bermuda Triangle’ of range uncertainties
Range and positioning uncertainties
WaterWater
TissueTissueTissuee
Tissue
NNSP
ρρ
ρ =≈
∑=i
iiA
Tissue
AZNN ω
where
NA is Avagadro’s number Zi, Ai and ωi are the atomic number, atomic weight and the proportion by weight of the ith element of
the compound
The relationship between chemical composition and
proton stopping power
[ ]KNcoh
cohph
phTissueTissue KZKZKN ++= 86.162.3ρµ
where
and again Zi and ωi are the atomic number and the proportion by weight of the ith
element of the compound
The relationship between chemical composition and
Hounsfield units
[ ] 62.31
62.3∑= iiph ZZ ω
[ ] 86.11
86.1∑= iicoh ZZ ω
Calibrating CT to stopping power (1)
Using these relationships, we can calculate a calibration curve between SP and HU for biological samples
Range and positioning uncertainties
Stoichiometric calibration
1. ‘Parametrize’ CT scanner using tissue substitutes (Kph, Kcoh, KKN)
Parametrization (measured
versus theoretical)
Fitted calibration curve (SP
versus HU)
Accuracy c.f. experiments 1-2% (Schaffner et al, PMB, 1998)
2. From parametrization, calculate HU and SP of biological tissues using their chemical composition
3. Fit multi-linear calibration curve through calculated points (Schneider et al, PMB 1996)
Calibrating CT to stopping power (2)
Range and positioning uncertainties
SFUD
Lomax AJ (2007) in ‘Proton and charged particle Radiotherapy’, Lippincott, Williams and Wilkins
IMPT
Degeneracy and range uncertainty
Range and positioning uncertainties
SFUD
IMPT
Lomax AJ (2007) in ‘Proton and charged particle Radiotherapy’, Lippincott, Williams and Wilkins
Degeneracy and range uncertainty
Range and positioning uncertainties
CTV (IMPT)
0
20
40
60
80
100
120
80 85 90 95 100 105 110 115
Dose (%)
Vol
ume
(%)
NominalCT +5%CT -5%
CTV (SFUD)
0
20
40
60
80
100
120
80 85 90 95 100 105 110 115
Dose (%)
Vol
ume
(%)
NominalCT +5%CT - 5%
+5% CT -5% CT
Lomax AJ (2007) in ‘Proton and charged particle Radiotherapy’, Lippincott, Williams and Wilkins
Degeneracy and range uncertainty
Range and positioning uncertainties
Modeling the effect of delivery uncertainties
SFUD IMPT
Dose distributions
Dose error-bar distributions (+/-3% range)
Albertini et al 2011 PMB 56;4399-4413
Range and positioning uncertainties
Error bars (% of prescription dose)
Volu
me
(%)
5 10 15 20 25 0 0
20
40
60
80
100
Modeling the effect of delivery uncertainties – Error-bar distributions
Error-bar volume histograms (EVH’s)
SFUD (PTV)
IMPT (PTV)
SFUD (CTV)
IMPT (CTV)
Range and positioning uncertainties
Albertini et al 2011 PMB 56;4399-4413
Potential magnitude
Beam energy [σ]
Patient positioning [σ]
Inherent CT uncertainties (beam hardening, calibration etc) [Σ]
Distal end RBE enhancements [Σ]
CT artifacts [Σ]
Variations in patient anatomy [Σ,σ]
The ‘Bermuda Triangle’ of range uncertainties
Range and positioning uncertainties
CT artifacts [Σ]
Range and positioning uncertainties
The problem of metal artefacts
Spinal stabilisation Hip prosthesis
Range and positioning uncertainties
The problem of metal artefacts kV-CT
MV-CT
0
0.5
1
1.5
2
2.5
3
3.5
0 20 40 60 80 100 120 140 160 180 200 220 240
X (voxels)
Sto
pp
ing
po
wer
KV SPMV SP
Prosthesis kV-CT artifacts
Stopping power profiles
PTV
Albertini et al, PTCOG 2006
Range and positioning uncertainties
Correcting metal artefacts
The manual approach
Uncorrected CT Artefact corrected CT
Range and positioning uncertainties
Dietlicher et al 2014 PMB 59:7181-‐7194
Correcting metal artefacts
Experimental validation
Plans calculated to uncorrected and manually corrected versions of the CT Each plan delivered to phantom and dose measured with film in three planes (as indicated)
Range and positioning uncertainties
Gamma agreement Manually corrected CT
97% with γ<1 (3mm/3%)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Dietlicher et al 2014 PMB 59:7181-‐7194
Correcting metal artefacts
Experimental validation
89% with γ<1 (3mm/3%)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Gamma agreement Uncorrected CT
Potential magnitude
Beam energy [σ]
Patient positioning [σ]
Inherent CT uncertainties (beam hardening, calibration etc) [Σ]
Distal end RBE enhancements [Σ]
CT artifacts [Σ]
Variations in patient anatomy [Σ,σ]
The ‘Bermuda Triangle’ of range uncertainties
Range and positioning uncertainties
Variations in patient anatomy [Σ,σ]
Planning CT Planning CT
Repeat CT
Repeat CT, 1st fraction
E.g. Variations in bowel filling
Francesca Albertini and Alessandra Bolsi (PSI)
Range and positioning uncertainties
Recalculated dose Francesca Albertini and Alessandra Bolsi (PSI)
Range and positioning uncertainties
Robust planning in the bowel
Planning (modified) CT
Original CT Planned dose
IMPT for robust planning
Lomax et al 2001, Med. Phys. 28:317-324
Range and positioning uncertainties
IMPT for robust planning
Range and positioning uncertainties
Liu et al 2012, Med Phys 39:1079-1091
Unkelbach et al 2009, Med Phys 36:149-163
1. Field shaping and op0misa0on 2. Plan design
3. Lateral penumbra and PBS
4. Range and posi0oning uncertain0es 5. Mo0on
Overview of presentation
1.8σ 1.3σ
Assume σ = 0.5cm
For this example, dose errors of ~20% can result from motion
(positioning) errors of 2.5mm
Phillips et al., PMB, 37:223-234,1992
Interplay
Motion
Max dose 111.2%
Max dose 112.8%
Static Max dose 120.2%
With motion (~6mm S-I)
Single field
Max dose 104.5%
Three fields
Interplay and multiple fields Motion
Knopf et al, Phys. Med. Biol. 56 (2011) 7257-7271
Max dose 111.2%
Static 4x rescanning
Single field
Max dose 104.5%
Three fields
Interplay and rescanning Motion
Max dose 108.6%
Max dose 104.9%
Knopf et al, Phys. Med. Biol. 56 (2011) 7257-7271
Zhang et al 2015, PTCOG 54, San Diego.
Motion Range adapted PTV’s
3 field plan to gITV
3 field plan to raITV
4D-‐CT
Motion
Zhang et al 2015, PTCOG 54, San Diego.
Range adapted PTV’s and rescanning
gITV
raITV
no rescanning
HI: 32%
Rescanning
HI: 20%
HI: 25% HI: 12%
• Treatment planning for PBS is a flexible and automated process
• Lateral penumbra is a limi0ng factor for PBS. Collimated PBS therapies will help to improve this in the future.
• The planning process is highly degenerate. This can be good and bad…
• Uncertainty in range is unavoidable, but is manageable if understood.
• Organ mo0on can severely degrade the delivered dose distribu0on, but re-‐scanning and ‘smart’ target design can help mi0gate the
problem
• The easiest form of robust planning is to use mul0ple beams. They can mi0gate (and hide) a mul0tude of evils…
Take home messages
The lateral dose fall-‐off of PBS proton therapy is…
20%
20%
20%
20%
20% 1. Below 5cm range is worse than photons but beHer than passive scaHering
2. Above 15cm range is beHer than for passive scaHering but worse than photon therapy
3. Is always beHer than photons or passive scaHered protons
4. Is always worse than passive scaHered protons 5. Is only dependent on the beam width in air
10
The air gap in PBS therapies …
20%
20%
20%
20%
20% 1. Is important due to large amount of scaHer in air that substan0ally broadens the beam
2. Can be neglected
3. Should be minimized only for the treatment of deep seated tumours
4. Should be minimized, in par0cular for the treatment of superficial tumors
5. Is only determined by the shape of the pa0ent.
10
20%
20%
20%
20%
20% 1. Is mainly due to uncontrollable varia0ons in proton beam energy
2. Is independent on the quality of CT data used for planning
3. Is typically in the sub-‐millimeter range
4. Can easily be reduced to sub-‐millimeter levels with careful calibra0on of the CT scanner
5. Will typically be of a few percent of proton range in the best case
10
Range uncertainty for proton therapy…
20%
20%
20%
20%
20% 1. Is not a problem
2. Is less problema0c than for passive scaHering
3. Is less of an issue than for IMRT
4. Is subject to interplay effects that effect target dose homogeneity
5. Is only a problem if mo0on is more than 1cm
10
The treatment of moving targets with
PBS…
20%
20%
20%
20%
20% 1. Use as few beam orienta0ons as possible
2. Use the distal fall-‐off to spare cri0cal organs such as the spinal cord and brain stem
3. Maximise the path length to the target to minimize pencil beam size
4. Use mul0ple beam orienta0ons to improve plan robustness
5. Use mul0ple fields to minimize integral dose
10
When designing plans and selecSng beam orientaSons for PBS proton therapy, it is good
pracSce to…