Proton Therapy Treatment
Planning: Challenges and
Solutions
Proton treatment delivery
Passive scattering in practice
Aperture
Range
modulator
wheel
Jatinder R Palta PhD
Daniel Yeung PhD
Roelf Slopsema MS
Department of Radiation Oncology
University of Florida
Patient
Range compensator
conforms dose to
distal edge target
volume; the aperture
shapes field in lateral
direction.
Brass aperture
Lucite range compensator
Physics of Proton Therapy
Scanning in practice
Tumor divided into
iso-Energy slices
Vacuum
Chamber
Scatterer
Target
Proton treatment delivery
Fast
Compensator
I.
–
–
–
Y
Already
irradiated
slices
Slow
Intensity
Modulated
Beam
basic interactions
Z
energy loss
scattering
nuclear interactions
X
IC’s A
IC’s B
Slice being
Irradiated
with raster
scan pattern
Pair of Quads
Isocenter
Scanning
Magnets
Bragg Peak
II.
clinical beams
–
–
percentage depth dose
lateral penumbra
Courtesy IBA
Proton interactions: Energy loss
Bragg peak
Primarily protons lose energy in coulomb interactions
with the outer-shell electrons of the target atoms.
• excitation and ionization of atoms
• loss per interaction small → ‘continuously slowing down’
• range secondary
e+
p+
<1mm → dose absorbed locally
200MeV
• no significant deflection protons by electrons
Stopping power
Proton interactions: Scattering
Primarily protons scatter due to elastic
coulomb interactions with the target nuclei.
Radial Spread in Water
Approximation:
σ ≈ 0.02 x range
Graph based on NIST data
Radial spread in water
Proton interactions: Nuclear
About 20% of the incident protons have a non-elastic
nuclear interaction with the target nuclei.
• reduction of proton fluence with depth → shape Bragg peak
• secondaries:
– charged (p,d,α,recoils) ~60% of energy → absorbed ‘locally’
– neutral (n,γ) ~40% of energy → absorbed ‘surroundings’
Same general shape applies to
different (relevant) materials.
• production of unstable recoil particles (activation)
– radiation safety – dose verification using PET/CT
Nuclear interactions
Proton Pencil Beam Algorithms
Hong et al, MGH 1996
Szymanowski et al, Institut Curie 2001
Beam Model
™ dosimetry in water
™ inhomogeneity correction
• water equivalent thickness (wet)
• HU
stopping power
Graph courtesy Harald Paganetti.
Proton Pencil Beam Algorithm
Beam Model
Proton Pencil Beam Algorithm
Dose Calculation
™Dose from a pencil beam
CAX Depth Dose (DD)
™ broad beam
™ pristine peak
SOBP
Radial Spread (RS)
™ multi coloumb scatter
• beamline: degrader, nozzle elements (wet only)
• compensator: material dependent scattering power
• patient: wet only
™ gaussian kernel
σ2total = σ2line + σ2comp + σ2patient
™Convolve DD with RS for each pencil
™Sum dose from all pencils
Pristine Peak ‐ Analytical Model Bortfeld Model of Pristine Peak
R=21.86 g/cm2
R=5.86 g/cm2
V 8.0
V 8.1
R=9.0 g/cm2
For R > 15 g/cm2,
Analytical model of proton BP (up to ~ 200 MeV)
• accounts for energy spread
• empirical model of nuclear fragmentation (data fitting)
• numeric depth dose calculation of fitted BP
• assumption – range straggling ‘constant’ with depth
R=19.8 g/cm2
Range straggling
depth
Error in Bortfeld model
Bortfeld Model
Eclipse Model
SOBP Calculated versus Measurement
Beam Profile
Calculated versus Measurement
R=15.1 M=10.4
d=10 cm
Proton beam dose calculation in
water is generally very accurate!
d=5 mm
20%-80% Lateral Penumbra [cm]
Option B5 - R=15.1 g/cm^2, M=10.4 g/cm^2, Air gap = 11.7cm, SSD=220.1, aperture: 15cmx15cm
1.3
Eclipse convolved - @9.9 cm
Eclipse convolved - @0.5 cm
Eclipse convolved - @14.1 cm
Measurement (av) - @9.9 cm
Measurement (av) - @0.5 cm
Measurement (av) - @14.1 cm
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
d=15 cm
UFPTI Data
0.4
0.3
0.00
5.00
10.00
15.00
20.00
Air Gap [cm]
25.00
30.00
UFPTI Data
35.00
Challenges in Proton Therapy Planning
1. Uncertainties:
à
à
à
à
à
2. Evaluation of proton plans
à
How to evaluate a proton plan in the presence of
various uncertainties?
3
Protons on water
I - dependence
w
The peak spread increases with
energy
2
1
0
PTV?
Error bars of dose distributions?
‚
‚
P122 Iw67
P122 Iw75
P122 Iw80
P183 Iw67
P183 Iw75
P183 Iw80
P230 Iw67
P230 Iw75
P230 Iw80
2
à
4
Intrinsic basic physics uncertainty (I-values)
CT numbers (stopping powers; range),
Dose calculation errors due to complex
inhomogeneities,
Intra-fractional organ motion,
Inter-fractional changes in anatomy and motion
patterns,
Mis-registration of tissue compensators (passively
scattered proton beams,
Uncertainties in immobilization devices and patient
support devices
dE/dz (MeV/g cm ) per incident particle
à
Intrinsic basic physics uncertainty
(I-values)
0
5
10
15
20
25
30
35
2
depth in water (g/cm )
P Andreo, Phys Med Biol, 2009
122 MeV Protons on water: I - dependence
w
4
average I-values of various soft tissues
2
w
P122 I = 75eV
w
P122 I = 80eV
w
3
2
2
dE/dz (MeV/g cm ) per incident particle
0.3 g/cm
P122 I = 67eV
1
Peak spread is .7 g/cm2
for 230 MeV protons
0
9.5
10.0
10.5
11.0
11.5
MUSCLE SKELETAL (ICRP)
LUNG (ICRP)
BLOOD (ICRP)
WATER LIQUID
TESTES (ICRP)
TISSUE SOFT (ICRU-33 4 comp)
HEART ADULT (BLOOD-FILLED)
MUSCLE STRIATED (ICRU)
HEART ADULT (HEALTHY)
GI-TRACT (INTESTINE)
HEART ADULT (FATTY)
EYE LENS (ICRP)
BRAIN (ICRP)
UREA
SKIN (ICRP)
TISSUE SOFT (ICRP)
TISSUE SOFT (ICRU-44 MALE)
TISSUE SOFT (ICRU-44 FEMALE)
BREAST (WHOLE)-50/50
BREAST (WHOLE)-33/67
ADIPOSE TISSUE (ICRP)
2
depth in water (g/cm )
Intrinsic basic physics uncertainty makes the
argument of “sub-millimeter precision” an issue,
which deserves careful consideration
62
74
76
HU-Stopping Power Calibration Curve
Peak spread
assuming 10%
uncertainty
in I-values
2
Realative Stopping Power
1.8
P 164 muscle skeletal ICRP
P 164 tissue soft ICRU-44 F
1.4
1.2
1
0.8
Large phantom, Average position, 140kV,
400mAs, Big bore
0.6
Body phantom, Average position, 140kV,
400mAs, Big bore
Full head phantom, Average position, 140kV,
400mAs, Big bore
Empty head phantom, Average position, 140kV,
400mAs, Big bore
0.4
0
-1000
P 164 muscle skeletal ICRP
P 164 tissue soft ICRU-44 F
17.5
1.6
0.2
P 164 water (I=75)
0.0
17.0
72
2
1.5
0.5
70
P Andreo, Phys Med Biol, 2009
0.7 g/cm
2.0
1.0
68
P Andreo, Phys Med Biol, 2009
2
dE/dz (MeV cm /g) per incident particle
2.5
66
I-value (eV)
164 MeV protons on various tissues
(+/- 10% change in I-values)
3.0
64
18.0
-500
0
500
CT #
18.5
19.0
19.5
2
depth (g/cm )
P Andreo, Phys Med Biol, 2009
1000
1500
2000
S. Flampouri, UFPTI
Range uncertainty due to CT calibration is
generally taken as 3.5%. However, it can be
minimized with better calibration techniques.
HU‐Stopping Power Conversion Uncertainties Results in Range Uncertainties
Impact of CT Hounsfield number
uncertainties on dose distributions
+3.5%
-3.5%
Dong/MDACC
S Flampouri, UFPTI, 2007
CT Artifacts and Hounsfield Numbers
0% uncertainty
Individualized patient determination of tissue composition
along the complete beam path, rather than CT Hounsfield
numbers alone, would probably be required even to reach
“sub-centimeter precision”
Proton Range Uncertainties
10% range error
“It is imperative that body-tissue compositions are not given
the standing of physical constants and their reported
variability is always taken into account” (ICRU-44, 1989).
The advantage of protons
is that they stop.
The disadvantage of protons is
that we don’t always know
where…
Lomax: PTCOG47
Proton Range Uncertainty in the
Presence of heterogeneities
Soft
tissue
Soft
tissue
ΔRhom
Bon
e
Impact of Organ Motion on Proton Dose Distributions
Free breathing treatment
ΔRinho
m
Tsunashima/MDACC
Lomax : AAPM SS 2003
Dong: ASTRO 2008
Plan DVH Evaluation (PRV)
What you see is not what you always get..
Impact of Tumor Shrinkage on Proton
Dose Distribution
Dose recalculated
Original Proton Plan
on the new anatomy
Volume
Proton DVH
Photon DVH
Dose
Bucci et al. ASTRO Abstract, 2007
Free breathing Treatment
Practical Solutions
Gated treated on exhale
L Dong: MDAH
Improving CT number accuracy and reducing metal
artifacts with Orthovoltage CT imaging
H2O
Brain
Al
Cortical
bone
Lung1
(a)
Al
(b)
Lung2
125 kVp CT
Air
125 kVp
320 kVp CT
Brain
Cortical
bone
LDPE
125 kVp CT
LDPE
320 kVp
320 kVp CT
Conventional 125 kVp CT Orthovoltage CT at 320 kVp
(c)
Yang et al. Med Phys 35 (5):1932-1941, 2008
(d)
Yang et al. Med Phys 35 (5):1932-1941, 2008
Match and Patch Fields
¾ used to avoid OARs adjacent to target
Match Fields
™ match fields abutting each other
™ penumbra matching penumbra
Inf - LAT
Sup - RAO
¾ partition target into segments (submatchline
targets)
¾ sub-targets treated with ‘sub-beams’
¾ angle sub-beams to avoid OARs
¾combined with other fields for dose uniformity
Lacrimal Gland Carcinoma
Patch Fields
™ thru beam txt partial target
™ residual txt with patch
™ lateral penumbra (t-beam) ‘matched’ with distal
falloff (p-beam)
™ LPO beam (inferior) patched with SPO
(superior)
(PTV 50.4)
partition into sup + inf targets
brainstem
sup
inf
LPO-Inf (spare optics & BS)
SPO-patch
PTV50.4: 5 fields with match & patch
Target
50%
Patch Field Selection
Fields
PTV
LAO
PTV(Inf)
LPO‐Inf
PTV(Sup)
LAO‐match
Gantry 205°
™ optimal geometric coverage (G = 230°)
™ avoid inhomogeneity along path (G = 205 °)
G=205°
G=230°
Patch line
Gantry 230°
Patch line
Gantry 205°
SPO‐patch
Patch Field – Beam Angle Selection
Gantry 230°
Patch Field Angle Selection
LPO
Distal Blocking
Distal Blocking
added
compensator
™ selective pullback of range to spare OARs
™ pullback achieved with added
compensator
™ potential pitfalls
¾setup error or motion may nullify sparing
¾‘simple’ distal blocking may compromise
target coverage
™ assess robustness of approach
Target
OAR
}
range pullback
Things to remember!
Distal Blocking
•
Proton beams stop - no exit dose
–
underdose
•
Whole BS
distal block
Partial BS
PTV
•
BS
Proton beams are more sensitive to
–
CT Hounsfield number/Stopping Power accuracy
–
Organ motion
–
Anatomy changes
Proton plans are difficult to evaluate
–
Partial BS
•
Although we don’t know exactly where they stop
“What you see is not what is delivered”
Protons demonstrate excellent low dose sparing
RPO Field
Things to remember!
• Inter/Intra-fractional variations have far more
significant consequences in patients treated with
proton therapy
Final thought
What you see is not what you always get….
IMXT Plan
IMPT Plan
(with Confidence-weighted Isodose Distribution)
– Approaches and data to deal with this issue is still
lacking
• Minimize it and develop strategies to deal with the residual
motion
• Empirical approaches used in defining margins
for range uncertainties, smearing, and
smoothing are questionable
– No real data exist to support any of these approaches
• Repeat imaging and reevaluation based on
deformable registration may be necessary
– In some cases repeat planning may be clinically
beneficial
Lomax PSI
Jin UF
180 160 120 80 40 cGy
There is no easy way to
show what patient will get
in proton therapy