Transition to Heterogeneity
Corrections
Eric E. Klein, M.S., Washington
University, St. Louis, MO
Craig Stevens, M.D., Ph.D., MD Anderson
Cancer Center, Houston, TX
Nikos Papinikolou, Ph.D., University of
Arkansas, Little Rock, AR
AAPM 2004 Annual Meeting
History of Prescriptive Changes
Brought Forth by Physics
• TG-43 changes to Sk based on NIST
Calibration updates.
• Gamma Knife (Elekta) found a 8%
discrepancy in 4 mm output. End result, Direct
Prescription change of 8%.
• Change from LDR to HDR GYN
Brachytherapy. Depends on institution.
• IMRT – Too early to advise if excessive hot
spots (EUD concept) matters
Why have accurate dose
algorithms?
• Effectiveness of radiation therapy
depends on maximum TCP and
minimum NTCP. Both of these quantities
are very sensitive to absorbed dose
• We learn how to prescribe from clinical
trials and controlled studies. Their
outcome depends on the accuracy of
reporting data
Inhomogeneity Corrections
Clinical Examples
• Orton et al (1998)
– Developed benchmark test case
– Reviewed 322 patients in RTOG 88-08
• Results
– Benchmark lung corrections
• Measured: 1.14 (Co-60)-1.05 (24 MV)
• Calculated: 1.17 (Co-60)-1.05 (24 MV)
– Patients: 0.95-1.28, mean=1.05, SD=0.05
• For lateral fields: mean=1.11, SD=0.08
• Conclusion
– Lung corrections lead to significant variations
– Density corrections will help reduce these variations
1
Inhomogeneity Corrections
Clinical Examples
• Mah & Van Dyk (1991)
– reviewed 100 thoracic patients
• Conclusions
–
–
–
–
–
Within lung, corrections are significant (0.95-1.24)
Target dose corrections are significant (0.95-1.21)
Substantial variation over patients (-5% to +21%)
Dose uniformity reduced in corrected distributions
In 80% patients, probability of lung damage
underestimated by >5% (up to 19%) if corrections
not applied
Van Dyk IJORBP 1983
Density
Determination
Assumed
Density
Dose Correction
Factor
CT Based
“real”
1.40
0
CT Measured
(total lung)
CT Measured
(average lung)
Age Related Best
Fit
Age Related Best
Fit (+ 1 SD)
Age Related Best
Fit (-1 SD)
Emphysema
Metastases
0.26
1.45
+4
0.35
1.39
-1
0.31
1.42
+2
% Difference from
“real density
calculation
0.38
1.37
-2
0.24
1.47
+5
0.06
0.60
1.59
1.23
+14
-12
Dose Correction Factors Based on
Different Lung Density Assumptions
Physics of Photon Dose Calculation Problem
TISSUE INHOMOGENEITY CORRECTIONS
FOR MEGAVOLTAGE PHOTON BEAMS
Report of Task Group 65 of the Radiation Therapy Committee
of the American Association of Physicists in Medicine
Nikos Papanikolaou
Jerry J. Battista
Arthur L. Boyer
Constantin Kappas
Eric Klein
T. Rock Mackie
Jeff V. Siebers
Michael Sharpe
Jake Van Dyk
University of Arkansas, Little Rock, Arkansas, USA
London Regional Cancer Centre, London, Ont., Canada
Stanford University, Stanford, California, USA
University of Thessaly, Medical School, Larissa, Hellas
Mallinckrodt Institute of Radiology, St Louis, MO, USA
University of Wisconsin, Madison, Wisconsin, USA
Virginia Commonwealth University, Richmond, Virginia
Princess Margaret Hospital, Toronto, Canada
London Regional Cancer Centre, London, Ont., Canada
• Incident photons
(spectrum)
• Scattered photons
• Scattered electrons
2
Energy transfer to electrons
Photon energy
(MeV)
1.25
2
4
6
Tmean
RCSDA(cm)
muscle lung bone
0.59
1.06
2.4
3.86
0.23
0.44
1.2
1.9
0.92
1.76
4.8
7.6
0.14
0.26
0.72
1.16
Magnitude of Effects
Photon scatter
Depth
(cm)
Field size
(cm)
5
10
20
5x 5
10 x 10
25 x 25
Scatter (% of total dose)
Co-60
6 MV
18 MV
12
24
48
lung=0.25
bone=1.85
g/cc
g/cc
Tmean = h
e
tr
e
Algorithms used for dose
calculation
Measurement based
Algorithms
Model based
Algorithms
Rely on measured data in water,
coupled with empirically derived
correction factors to account for
patient contour, internal anatomy
and beam modifiers (Clarkson,
ETAR)
Use measured data to derive the
model parameters. Once initialized,
the model can very accurately
predict the dose based on the
physical laws of radiation transport
(convolution, MC)
7
14
27
Range of scattered electrons
Range
assumes
8
18
38
Forward (cm)
Lateral (cm)
Co-60
0.5
0.2
Energy
6 MV
18 MV
1.5
3.0
0.4
0.8
How many Dimensions ?
Dimension of anatomy
Dimension of scatter inclusion
Where will the scatter go ?
(photons, electrons)
z=1
z=0
z=-1
A dose cloud displayed on a 3D rendered volume does
Not necessarily suggest a 3D algorithm
3
Local Energy Deposition - No Electron Transport
Local Energy Deposition - No Electron Transport
Ratio of Tissue-Air Ratios (RTAR)
Power Law (Batho)
• Handles primary accurately
(electronic equilibrium)
• Partial correction for scatter
ICF =
T (d 1, Wd ) 1
T ( d 2, Wd )1
(modified depth in TAR)
• Nothing about size, shape or
location
Local Energy Deposition - No Electron Transport
Power Law (Batho)
2
2
2
1
O’Connor’s Scaling Theorem
• Adapted to CT planning by Webb et al
– layers, multiplicative factors
• sensitive to depth from surfaces
(changes primary & scatter)
• not sensitive to width
• better than RTAR
• under corrects < 1.0, over corrects > 1.0
• under predicts for large fields
• improves with TPR instead of TAR
• problems when di lie in build up region
• Dose at A = Dose at B
• dx
and w x
are equal
4
Local Energy Deposition - No Electron Transport
Equivalent Tissue Air Ratio (ETAR)
ICF =
T (d ' , ~
r)
T (d , r )
• Uses O’Connor’s scaling theorem
d & r are depth & radius of equivalent field
d'& ~
r are scaled versions of d & r
~
r = r~
~=
ijk
i
j
wijk
• Mackie et al
(1985)
• Effects of
electron
transport
– High energy
• Predicted by
convolution
k
wijk
i
j
k
Non Local Energy Deposition - Electron Transport
Convolution - Point Kernel
D =
Inhomogeneity Corrections
Measured and Calculated Data
dxdydz
3D
( x , y , z ) K pt ( x , y , z )
Convolution: Dose Computation
muscle =1 gr/cm3
lung =0.25 gr/cm3
5
Convolution Lung Calculation
Convolution/Superposition
MDAH History
• 5 years ago
– transferred CT info to simulator films by hand.
• CT not in Rx position
– "at least" 1cm from tumor edge to block edge
– dose calculated to midplane in a homogeneous
patient
Homogeneous Scatter
Homogeneous Primary
and Scatter
Now
• GTV contoured on Rx planning CT, with FDG
PET to identify LN
• CTV based on the literature (8mm).
• PTV
– tumor motion measured in ALL patients (ITV).
– Set up uncertainty measured (2SD=7mm)
• then block edge (~7mm)
• GTV to block edge 8+7+7=22mm
• Rx 95% of PTV gets Rx dose.
6
Now
But we've never done it before.
• Able to do this on a service with
– 8 attendings
– 2 physicists
– 6 dosimetrists (that rotate)
– 12 Rx machines
• About 100 therapists
– All while implementing IMRT and other new
technologies
Planning Characteristics
• But the changes to the isocenter are small,
while coverage of the PTV becomes MUCH
better.
Planning Assumptions
• GTV -> CTV 8 mm (Giraud et al., 2000)
• CTV -> PTV 10 mm
• PTV -> Block edge 10 mm
• Plan 1: calculate dose to iso, homogeneous
• Beam geometries and prescription (60-66
Gy) were those used for initial treatment.
• All beams 6MV x-rays
• Plan 3: adjust beam weights so that 95% of
PTV treated to target dose.
• Plan 1H: monitor units from 1, heterogeneous
7
Case 1
Case 1
• T1
• Goal 66Gy
• What was planned in 3D
Case 1
1
1H
Case 1
1H
3
8
Case 1
Why?
• Tumor more anterior
• Lung posterior
Monitor units
Plan 1
Plan 1H
Plan 3
AP
123
123
160
PA
178
178
135
total
301
301
295
Case 2
• T2 tumor
• Goal dose 66 Gy to iso
• Therefore, weighting
should be more AP
9
Case 2
Case 2
• What the dose distribution looks like in 3D
1
1H
Case 2
Case 2
1H
3
10
Case 2
Case 2 18MV
• Dose to GTV
1
2 Gy
3%
1H
Case 2 18MV
1
Patient Characteristics
•
•
•
•
29 patients with 30 tumors
Stage I or Stage II
CTV range: 15-359 cm3
PTV range: 73-760 cm3
1H
11
Planning Characteristics
Planning Assumptions
• GTV -> CTV 8 mm (Giraud et al., 2000)
• CTV -> PTV 10 mm
• PTV -> Block edge 10 mm
• Plan 1: calculate dose to iso, homogeneous
• Beam geometries and prescription (60-66
Gy) were those used for initial treatment.
• Plan 3: adjust beam weights so that 95% of
PTV treated to target dose.
• Plan 2: monitor units from 1, heterogeneous
Isocenter Dose
Dose (Gy)
Number of patients
Isocenter Dose
Plan 1H
Plan 3
Mean difference
3%
% Difference
Plan 3 - Plan 1H
12
CTV Maximum
Plan 3
Plan 1H
% Difference
Plan 3 - Plan 1H
Plan 3
PTV Minimum
Dose (Gy)
Number of patients
CTV Maximum Dose
% Difference
Plan 3 - Plan 1H
PTV Maximum
Dose (Gy)
Plan 1H
CTV Minimum Dose
Number of patients
Dose (Gy)
Dose (Gy)
CTV Minimum
Plan 1H
Plan 3
Plan 1H
Plan 3
13
PTV Maximum Dose
Number of patients
Number of patients
PTV Minimum Dose
% Difference
Plan 3 - Plan 1H
% Difference
Plan 3 - Plan 1H
% PTV Coverage
95% of PTV to Goal Dose
Number of patients
Number of patients
% PTV Coverage
Plan 1H
% Coverage to Goal Dose
p=0.05
% Coverage to Goal Dose
14
Beam arrangement for IMRT plan
Summary
• Monte Carlo is very similar to convolutionsuperposition with heterogeneity
• Hetero plans are close to Monte Carlo
– On average PTV coverage is better.
– Case-by-case can be quite different
• And it's not hard.
•
But block margin, weighting, and energy will be chosen more
accurately
Transverse plane isodose comparison
Pencil Beam calculations
0.25 cm
0.5cm
1cm
2cm
EGS4-BEAM calculations on 2100C
Inhomogeneous
Homogeneous
Inhomogeneous with
Homogeneous calc MUs
Calculation accounts
for varying tissue
densities
Calculation assumes
all tissue has water
density
Calculation accounts for
varying tissue densities but
forces the MU from
Homogeneous calc
Note the difference in coverage in absolute dose between plans
15
TG-65 Recommendations
Dose comparison and Ratio
Homogeneous
Calculation
Inhomogeneous
Calculation
Inhomogeneous
w/ Homogeneous
MU’s
Max.
Dose
cGy
Max.
Dose
cGy
Max.
Dose
cGy
Mean
Dose
cGy
Mean
Dose
cGy
Mean
Dose
cGy
Ratio of
Inhomogeneous
w/ Homog. MU’s
and Homogeneous
Max.
Dose
Ratio
in %
Mean
Dose
Ratio
in %
GTV
5399.71 5348.06 5428.59 5383.83 5091.50 5044.67 -5.71
-5.67
Adenopathy
5420.27 5341.31 5451.35 5399.08 5109.21 5048.74 -5.74
-5.48
Lt.
Lung
5392.60 1056.51 5451
-4.21
Soft
Tissue
5420.27 292.801 5456.66 299.35
1076.57 5110.07 1011.98 -5.24
5114.79 276.946 -5.64
-5.42
Under-dose of 5% or more is introduced by the monitor units as calculated by the homogenous plan
• The physicist needs to understand the
algorithm(s) within the TPS and MU
calculation programs.
• The physicist is strongly advised to test the
planning system to ascertain if the system can
predict common trends.
• The physicist is advised to measure
benchmark data for their own beam and
compare with the calculated (planning
system or hand calculations) data. If possible,
the physicist may also use Monte Carlo
calculations to support measured data.
TG-65 Recommendations
• The physicist should understand the dose
calculation resolution grid, due to
volumetric averaging.
• The physicist should maintain an open
dialogue with clinicians and be clear on
limitations of the TPS. For each clinical
site (eg. left breast, right lung, larynx etc),
there should be 5-10 treatment plans
generated, with & without inhomogeneity
corrections. The dose prescription should
be the same for both cases.
TG-65 recommends energies of 12 MV or less
for lung radiotherapy.
TG-65 Recommendations
•
•
•
The physicist should keep abreast of new
algorithms. The vendors should provide
clear documentation of the inhomogeneity
correction methods implemented.
When physicists teach residents, tissue
inhomogeneity effects on doses should be
discussed.
The physicist should finally confirm that
the method to calculate treatment time or
monitor units, whether it is derived by the
treatment planning software, or with an
alternative method, is accurate to deliver
the planned absolute dose to the point of
interest.
16
Implementation Recommendations
• In addition, planning volume margins
may be affected according the
algorithms’ ability to calculate
penumbra in the presence of
inhomogeneous media, particularly
lung.
• Classic beam arrangements may
need to be scrutinized due to the
impact of increased exit dosing
observed with corrections applied.
17