PatientPatient-based Quality Assurance
for IMRT
Conformal Radiation Therapy
CECE-IMRT4
Dirk Verellen et al.
Department of Radiotherapy, AZ-VUB
Outline
➨ Customize your
Creation of conformal dose distributions & Target oriented positioning !
IMRT & IGRT !
Customize
QA and Treatment
Verification
➨ Tools
➨
➨
●
●
● Dosimeters
● Phantoms
●
➨
➨ Procedures
analysis
➨ Discrepancy analysis
➨ Beware of what has not been verified
How to get comfortable?
What is recommended/required “officially”?
Create an efficient QAQA-procedure
●
➨ Hazard
●
➨
Absolute dose check
Verify each field
Verify composite treatment
“Don’t drown in film measurements”
Efficient processing required
Target oriented positioning or ImageImage-Guided
Radiation Therapy (IGRT)
QA for IMRT: 4 levels
Level 1: Basic linac QA
➨
PrePre-clinical verification of
IMRT treatment
(Patient related)
related)
➨
Verification of fluence
maps, individual IMRT
fields on water phantom
4
➨
Tests for Validation and after every accelerator
check
●
3
2
➨
●
●
●
●
IMRT delivery specific QA
➨
1
➨
Basic QA (linac, MLC)
Level 1: weekly QA pattern
MLC calibration and alignment
Speed Stability
Beam on/off stability
Gravity test
MLC reliability test
Weekly test
●
Garden fence test
Level 2: IMRT delivery specific QA
➨ Acceptance
➨ Commissioning
● e.g.
● e.g.
➨ Small
Test Pattern with Leaf Error
Test Pattern after leaf replacement
and MLC calibration
Solberg et al.
alignment in tomotherapy
MLC specifications and influence on OF
field dosimetry
Level 3: Verification of IMRT delivery
Level 3: Chair
➨ “Chair”
➨ Geometric fluence profile
Leaf transmission only
➨ Irregular
test profile
➨ Making use of EPIDs
Absolute dose
Leaf transmission
&
Leaf separation
Level 3: Test pattern
Level 3: Test pattern
Test Pattern correctly delivered
Test Pattern with 2 mm Leaf Error
and 10% Dose Error
Solberg et al.
Level 4: Different philosophies
➨ Every day ↔
● QA
Every patient ↔ Class solutions
Level 4: Comprehensive Verification
➨ Verification of treatment in toto:
● absolute dose verification: thermoluminescent and alanine detectors
●
Evaluation of dose distribution: film dosimetry
procedure largely dependent on approach
➨ TopTop-down
↔ BottomBottom-up
●Detailed
●analysis
●Time
straightforward
consuming
●Comprehensive
●Discrepancy analysis
Original IMRT Plan simulated for phantom
verification
complicated
Level 4: 2 legs to stand on
QA in IMRT: an example
➨ The
➨ NonNon-patient
ability to create conformal dose
distributions
● How to
make sure it performs adequately each
related
● Comprehensive test
❍
time?
for IMRT delivery capability
Fluence map created by TPS, sequencing from TPS,
transferred and delivered
“Test Pattern”
Pattern”
➨ Target
oriented positioning
● How to
time?
make sure not to miss the target each
❍
e.g. every week
● Regular
detailed QA of linac and MLC (basic
verification)
❍
e.g. every month
QA in IMRT: an example
Tools
➨ Patient
➨ Dosimetry
related
● Comprehensive test
for class solution (pre(pre-
clinical verification)
Commercial “IMRT” phantom
Anthropomorphic phantom
❍ Gel
❍ …
❍
❍
● PrePre-treatment
➨ Phantoms
verification for each patient
Independent MU calculation
❍ Absolute dose check
❍ Verification of fluence patterns
❍
Dosimeters
➨
Integrating
●
●
●
●
●
●
➨
Dosimetric verification
TLD chips
Alanine chips
MOSFET
Radiographic film (X(X-OMAT V, EDREDR-2)
Radiochromic film
Gel
NonNon-integrating
●
●
●
●
Ionization chamber (conventional, micro, pinpin-point)
diodes
diamond
Linear array detectors
➨ Down
scaling of Monitor units
● Losing
small segments
of scatter and leakage dose
● Underestimation
➨ Small
field dosimetry
delivery:
➨ Temporal dose
● integrating
dosimeters (TLD, alanine, film, gel)
(ionization chamber, …)
● nonnon-integrating
Phantoms
Dosimeters
➨
0 dimensional
●
❍
❍
●
●
●
●
➨
➨
●
Conventional
Micro, diamond, …
Diodes
MOSFET
Diamond
TLD, alanine
2 dimensional
●
Ionization chamber
●
➨
Film
EPID
Array of TLD chips
●
Generally 3 types
●
●
●
●
Gel
Stack of film
3D3D-array of TLD chips
Anthropomorphic
❍
❍
3 dimensional
❍
●
❍
❍
❍
Phantoms
Internal construction precise
Multiple dosimeters possible
Alignment straightforward
Geometrically irregular
❍
Stack of TLD chips
Linear array detectors
Internal heterogeneities are anatomically relevant
Multiple dosimeter comparison difficult
Geometric alignment cumbersome
Geometrically regular
❍
●
1 dimensional
●
➨
Create fluence map to obtain a homogeneous dose distribution
Easy for analysis
Phantoms
➨
➨
➨
●
0D:
●
Ionization Chamber
ABSOLUTE
2D:
●
Film Dosimetry
RELATIVE
0D:
●
Ionization Chamber
ABSOLUTE
TLD, alanine
ABSOLUTEABSOLUTE-RELATIVE
➨
1D:
➨
21/2D:
●
●
Stack of TLD
Film Dosimetry
RELATIVE
Phantoms
Phantoms
➨
➨
●
➨
●
RELATIVE
ABSOLUTE?
De Wagter et al.
Film Dosimetry
RELATIVE
Stack of TLD
Phantoms
Phantom Verification
➨
2D
●
A. Van Esch et al.
●
TLD, alanine
ABSOLUTEABSOLUTE-RELATIVE
2D:
●
Williams et al.
3D:
●
0D:
Create fluence map to
generate a
homogeneous dose
distribution
➨
Necessary tools:
●
●
●
Dose Export of a defined area or plane into file or clipboard (ASCI)
(ASCI)
Export of data to beam shaper for 2D phantom verification at
specified depth
Independent registration of measurement and calculation needed
Level 4: Procedures
➨ Fluence
Fluence profiles: Film Dosimetry
profiles
● Measurement
and analysis
measured fluence profiles to recalculate
the dose distribution
● Using
➨ Combining phantoms
➨ Absolute
and dosimeters
dose verification
● Measurement
● Calculation
➨ Verification of
Film
Cadplan
dose distribution
Ahlswede et al.
Fluence profiles: Film Dosimetry
➨ Gamma
●
●
●
●
Fluence profiles: Film Dosimetry
index at the Charité
Charité
Gamma evaluation
Low et al.,
al., 1998
Quantitative evaluation of dose distributions
Measurement within limits, if g < 1
The following limits have been used:
Dose difference:
DD/D < 3%
distance to agreement: DA < 2.0 mm
Ahlswede et al.
Ahlswede et al.
Film Dosimetry
Fluence profiles: EPID
Gamma comparison at University Hospital Leuven:
rel. OD
Film Response
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
clinical implementation of gamma algorithm
on dosimetry with PortalVision:
EDR2
• pre-treatment evaluation:
EPID versus TPS
X-Omat V
0
2
4
6
• treatment evaluation:
EPID versus TPS
EPID versus EPID
8
Dose (Gy)
• error detection
Tournel et al.
Fluence profiles: EPID
reference image
A. Van Esch et al.
Fluence profiles: the other way around
measured image
➨
acceptance criteria:
∆ Dmax (e.g. 1 %)
DTA (e.g. 1 mm)
A. Van Esch et al.
Using measured fluence profiles imported back into
the planning system to calculate what has been
delivered!
Combining phantoms and dosimeters
Absolute dose and MU validation
➨ Transferring
➨ MU
the patient’s treatment
parameters to a phantom and recalculate the
resulting dose distribution: “Mapping
“Mapping””
✚
Verification with the actual treatment parameters
validation requires either
● Direct
measurement of dose using TPS MUs and
fluences
❍
- Dose distribution may not be relevant
TimeTime-intensive
■
■
➨ Simulating the patient’s treatment
on a
phantom: “Simulation
“Simulation””
✚
Verification of specific treatment requirements
- Actual treatment is not verified
Absolute dose measurement
■
❍
Temporal
High dose gradients
Small field dosimetry
Currently most thorough method of validation
● Independent
computation of dose
Most efforts still single point
❍ Ideally, recompute entire 3D dose
❍
Alanine dosimetry
Calculated
(Gy)
SD
(Gy)
Measured
(Gy)
SD
(Gy)
meas/calc
case (a)
case (b)
case (c)
20.09
20.04
19.99
0.14
0. 09
0.05
20.01
19.79
19.77
0.20
0.12
0.19
1.00
0.99
0.99
Det. 1
Det. 2
Det. 3
10.73
10.85
4.32
0.14
0.12
0.07
10.74
10.77
4.31
0.13
0.11
0.04
1.00
0.99
1.00
Solitary target:
Target
surrounding OAR:
TLD dosimetry
Ionization chamber*
Calculated
(Gy)
SD
(Gy)
Measured
(Gy)
SD
(Gy)
meas/calc
Det. 1
Det. 2
Det. 3
1.083
1.094
0.324
0.006
0.006
0.014
1.131
1.109
0.324
0.035
0.022
0.007
1.04
1.01
1.00
Det. 1
Det. 2
Det. 3
Det. 4
0.983
0.789
1.024
0.117
0.009
0.014
0.031
0.000
0.980
0.755
0.927
0.122
0.039
0.030
0.037
0.005
1.00
0.96
0.91
1.04
Target
surrounding OAR:
Treatment
simulation
palate tumor:
H&N
IMRT
Calc.
SD
Meas.
SD
(cGy)
(cGy)
(cGy)
(cGy)
5 field
Evenly
distributed
166.6
4.5
165.0
0.78
201.3
3.4
200.1
0.56
5 field
Avoiding air
cavities
*NAC 007 micro ionization chamber, Wellhö
Wellhöfer
Independent computation of dose
Independent computation of dose
unit normal deviate
➨ Single point
approach
➨ A simple method, spread sheet – based:
-10
-8
-6
-4
-2
0
2
4
6
8
10
45
0.21
40
frequency
35
0.18
normal distribution
● Imported from
counts
30
TPS:
0.12
20
0.09
MU per beam
❍ Segment shapes and weights
15
❍
● Not
0.15
25
0.06
10
0.03
5
used from TPS:
0
0
-10 -9
-8
-7
-6
-5
-4
-3
-2
-1
0
1
2
3
4
difference (cGy)
Original TMR data, OF, OAR
❍ Determination of segments that cover measuring point
❍
Linthout et al.
Mean: -0.2cGy
SD: 2.0cGy
5
6
7
8
9
10
Dose distribution: mapping
Original IMRT Plan
Original Plan mapped onto
Phantom
Film Dosimetry : mapping
Tournel et al.
Dose distribution: simulation
Gamma: 4% DD / 4mm DTA
Original IMRT plan simulated for phantom
verification
Film Dosimetry: simulation
Film Dosimetry: simulation
Film Dosimetry: simulation
Film Dosimetry: simulation
Phantom verification: gel measurement
Relaxation rate image
with contours along the
pixels with 90% of the
maximal dose
Transversal CTCT- slice
with PTV and calculated
90%90%-, 50%50%-, 20%20%- and
10%10%-isodose lines
Tournel et al.
Gamma: 4% DD / 4mm DTA
MRTMRT-slice
with contour along
pixels with 90% of the
maximal dose
Ahlswede et al.
Phantom verification: gelgel-film
GEL
FILM - GEL
FILM
PLANNING - GEL
PLANNING
Hazard analysis
➨ Intuition/experience from
conventional
RT is lost
➨ Find weak links
➨ Define control points
PLANNING - FILM
De Wagter et al.
Hazard analysis: some examples
➨ Leaf
Leaf Sequencing : DMLC ↔ SMLC
calibration
● e.g.
OF can change with 7% for 0.1 cm difference
in small field sizes for an Elekta linear
accelerator.
● Leaf sequence important
➨ Tertiary collimator
● Alignment,
abutting slices
100
90
80
70
60
100
90
50
40
30
20
80
70
60
10
0
0
● Clearance
5
10
15
20
25
50
40
30
20
10
0
0
5
10
15
20
25
Leaf Sequencing : calibration
Leaf Sequencing : after calibration
Leaves+BackUpJaw ↔ Leaves: before MLC calibration
100
90
80
Leaves+BackUpJaw ↔ Leaves: after MLC calibration
70
60
100
90
50
40
30
80
70
60
20
10
0
0
5
10
15
20
50
40
30
20
25
10
0
0
Leaf Sequencing
“CloseClose-in” Technique
15
20
“Sweep“ or
“Sliding Window” Technique
0°°
4
3
2
1
4
3
2
1
Ahlswede et al.
10
Sequential tomotherapy: alignment
Desired
IM - Profile
Trajectory
Sequence
5
270°°
90°°
25
Sequential tomotherapy: alignment
Sequential tomotherapy: indexing
Sequential tomotherapy: alignment
Sequential tomotherapy: alignment
120
115
115
110
105
alanine
TLD
film
planning
105
Dose (%)
Dose (%)
110
alanine
film
planning
100
95
100
90
95
Second rotation
First rotation
90
0
1
2
3
4
5
6
7
8
9
10
11
Underdosage
12
13
14
15
Overdosage
postition (mm)
First rotation
85
0
1
2
Index +1mm
Direction of table movement
3
4
5
6
7
8
9
10
11
12
13
14
15 Underdosage
16 17
Second rotation
18
Overdosage
position (mm)
Direction of table movement
Sequential tomotherapy: alignment
Clearance and choice of origin
120
115
Dose (%)
110
105
alanine
film
planning
100
95
90
First rotation
85
0
1
2
Index -1mm
3
4
5
6
7
8
9
10
11
12
13
14
15
Underdosage
16 17
Second rotation
18
Overdosage
position (mm)
Direction of table movement
Clearance and choice of origin
Clearance and choice of origin
Discrepancy analysis
➨
TPS:
●
●
●
●
●
➨
Discrepancy analysis (cont’d)
➨
Basic beam data (PDD, OF, leaf offsets, penumbra)
Linac model
Dose calculation algorithm
Leaf sequencing algorithm
…
●
●
●
●
●
●
➨
TLD calibration
MLC data transfer
Experimental setset-up (many things can go wrong: MU,
positioning, gantry, … typically afterafter-hours)
…
Discrepancy analysis
MLC calibration
Linac operation
…
Analysis
●
Experiment:
●
Delivery
●
Incorrect registration
DownDown-scaling of MU
❍
❍
●
Losing small segments
Underestimating leakage/transmission dose
…
Discrepancy analysis
R2 = 1/T2
0.8 cm
1.4 cm
Planning minus gel
Planning minus gel
De Wagter et al.
< -5%
> +5%
2 x 15o
2 x 16o
De Wagter et al.
Discrepancy analysis
Inefficient use of the beam
Dose (OAR) ↓
&
Dose (Target) remains constant
⇓
The number of available ports ↓
⇓
The number of MU/° or MU/segment ↑
⇓
The contribution of leakage & scatter dose ↑
The relationship between measured versus calculated dose in function of increased
constraints to the OAR while maintaining the prescribed target dose at 1.00 Gy
Case 1
Case 2
Case 3
Case 4
Prescribed
Calculated
T
(Gy)
OAR
(Gy)
T
(Gy)
OAR
(Gy)
T
(Gy)
Measured
OAR
(Gy)
T
OAR
(Gy) (Gy)
Meas/Calc
1.00
1.00
1.00
1.00
0.25
0.10
0.06
0.03
1.08
1.07
1.17
1.15
0.32
0.19
0.05
0.03
1.08
1.09
1.09
0.94
0.34
0.19
0.09
0.08
1.00
1.02
0.94
0.82
1.04
0.98
1.83
2.92
MU/º
0.79
0.81
1.30
1.40
Discrepancy analysis: Influence of leakage dose
➨
➨
Ionization chamber measurements showed a transmission of
0.5% through the vanes of the MIMiC.
This enables to calculate an estimated leakage dose based
on the total amount of MU delivered during tomotherapy
Total Leakage Calculated C + L Measured
MU
(cGy)
(cGy)
(cGy)
(cGy)
Case 1
Case 3
458
755
2.29
3.78
32.4
4.67
34.7
8.45
33.6
8.56
Discrepancy analysis: Influence of leakage dose
➨
Phantom measurements (TLD) of the tomotherapy
procedure compared to an identical treatment with all vanes
closed during treatment
Total Leakage Calculated C + L Measured
(cGy)
(cGy)
(cGy)
(cGy)
MU
M/C M/(C+L)
1.04
1.83
0.97
1.01
Case 1
Case 3
458
755
1.60
2.45
32.3
4.70
33.9
7.15
33.6
8.56
M/C M/(C+L)
1.04
1.83
0.99
1.20
Discrepancy analysis
Discrepancy analysis
∆D = 1% DTA = 3mm
≠ Rectal filling
day 1 vs day 6
day 1 vs day 2
day 1 vs day 4
day 1
A. Van Esch et al.
Discrepancy analysis
day 4
day 1 versus day 4
A. Van Esch et al.
Beware of what has not been verified
Wrong energy
➨ Threshold for skin contouring
➨ ExtraExtra-target
ABSOLUTE
RELATIVE
∆D=3.3% and DTA=3mm
∆D=3.3% and DTA=3mm
MLC failure
∆D=5.5% and DTA=3mm
A. Van Esch et al.
dose
➨ Heterogeneity correction
➨ Target localization
➨…
in TPS
ExtraExtra-target dose
ExtraExtra-target dose
Any absorbed dose the patient receives outside the
treatment volume must be considered undesirable.
➨ In addition to the primary beam absorption in
overlying and underlying healthy tissue, the major
sources are:
➨
●
●
●
X-ray leakage
photons scattered out of the treatment volume
neutrons originating in the treatment head and leaking
through the head shielding.
Gonick & Huffman
WBED for a prostate case
➨ Assuming
Comparison w literature
identical scatter conditions:
Hp(10) = 1.55 x 10-2 mSv/MU
Hp(70 Gy) = Hp(10) x #MU x #fractions
Verellen et al
Followill et al
Mutic et al
20475
127050
6.2
8400
67900
8.1
94500
-
1.18 x 10-2 mSv/MU
1.55 x 10-2 mSv/MU
0.8 x 10-2 mSv/MU
0.8 x 10-2 mSv/MU
0.4 x 10-2 mSv/MU
Hp,conv(70 Gy)
Hp,tomo(70 Gy)
ratio
242 mSv
1969 mSv
8.1
67 mSv
543 mSv
8.1
406 mSv
-
Prob. coeffconv
Prob. Coefftomo
ratio
1.2 x 10-2
9.9 x 10-2
8.3
0.4 x 10-2
2.8 x 10-2
7.0
2.0 x 10-2
-
MUconv
MUtomo
ratio
Hp(10)conv
Hp(10)tomo
● serial
tomother. (654 MU, 5 arcs):
1 (490 MU, 6 fields):
● IMRT 2 (128 MU, 6 fields):
● Dynamic arc (292 MU, 1 arc):
● IMRT
1774 mSv
1595 mSv
417 mSv
158 mSv
Heterogeneity correction
Heterogeneity correction
CC Algorithm
Clarkson Algorithm
PB Algorithm
Cranial
measurement
Caudal
measurement
air cavity :
target volume :
Linthout et al.
Target volume
Linthout et al.
Heterogeneity correction
Choice of phantom
CC Algorithm
Clarkson Algorithm
PB Algorithm
Linthout et al.
Target volume
Gamma: 4% DD / 4mm DTA
IGRT
The radiotherapy chain
Physical patient
➨ Conformal Dose distribution
● High dose volume is shaped to the volume occupied by
the target.
● Don’t miss the target!
∴PTV and PRV should reflect setset-up accuracy!!!
➨ Temporal Intensity Modulation
● Optimization based on snapsnap-shot.
● Target displacement/movement influences dose
distribution.
∴RealReal-time knowledge of anatomy required!!!
Target Delineation
➨
➨
➨
➨
➨
➨
➨
➨
➨
CT room coords
Lasers
Skin markers
Images
Bone
Tumor
Delineation
Margin
Planned beam
Van Herk et al.
➨
➨
➨
➨
➨
➨
➨
➨
Treatment room coords
Lasers
Skin markers
Bone
Tumor
Beam
Linac
Treatment room
17 possibilities for geometrical
errors
Virtual patient
Temporal intensity modulation
Field 2
2
1
2
1
OAR
1
2
PTV
2
Field 1
Ahlswede et al.
Field 3
Temporal intensity modulation
Temporal intensity modulation
Field 2
Field 2
2
1
2
2
Field displacement
1
2
overdosage
1
4
3
4
Field 1 1
1
Field 1
Field 3
2
6
5
6
2
Field
displacement
2
3
4
5
2
3
4
3
4
5
6
4
5
3
6
4
underdosage
Ahlswede et al.
Ahlswede et al.
Temporal intensity modulation
Patient immobilization
1
2
Field 3
Field displacement
Target localization
Target localization
Target localization≠
localization≠immobilization
Target localization
➨
Tomotherapy
To treat sequential
transaxial slices the
patient is translated
longitudinally between
consecutive gantry
rotations using the
“Crane”.
Crane”.
Macky et al.
al.
RealReal-time Verification
Electronic Portal Imaging
“perfect alignment”
+
room lasers
+
+
+
+
+
+
+
skin markers
bone references
Ultra Sound Guidance
Anatomy tracking
As a positioning tool: only 2D information
Cone beam CT
Jaffray et al.
al.
Infrared Guidance
Stereoscopic XX-ray Imaging
Stereoscopic XX-ray Imaging
Stereoscopic XX-ray Imaging
Implanted marker matching
Automated DRR fusion
Boyer et al.
al.
IMRT: Patient related QA
Conclusions
➨ Analyze
the chain of events in your IMRT
treatment procedure
● Hazard
analysis: define control points
QA/QC procedure
● Get comfortable with each step
● Customize
Analysis?
➨ Complementary dosimetry
➨ QA
procedure should be efficient
of personnel!
➨ Training
Acknowledgements
Special thanks to:
Ann Van Esch
(University Hospital Leuven)
Leuven)
Carlos De Wagter
(Ghent University Hospital)
Julia Ahlswede
(Charité
Charité, Berlin)
Tim Solberg
(UCLA)