Quality Assurance in Linac-based Stereotactic
Radiosurgery and Stereotactic Body Radiosurgery
Timothy D. Solberg, Ph.D.
Department of Radiation Oncology
University of Texas Southwestern Medical Center
timothy.solberg@utsouthwestern.edu
SRS and SBRT are Ablative Therapies
SRS is a standard of care in the treatment of cranial
neoplasms
SBRT has produced extraordinary control rates in early
stage lung cancer
But ….
In the first 11 liver patients treated with SBRT at
Karolinska, 3 died of treatment induced complications
There are recent reports of severe toxicity in lung and
spine patients treated with SBRT
Blomgren et al, Acta Oncol. 34:861-70, 1995
ED50 =
20.4 Gy
ED50 =
24.9 Gy
Can we hit the target?
Can we put the dose where we want it?
Bijl et al, IJROBP 52(1):205-211, 2002
Quality Assurance in Linac-based Stereotactic
Radiosurgery and Stereotactic Body Radiosurgery
How accurate is radiosurgery?
• Localization Accuracy
•Frame Based
•Image Guided
• Dosimetric Accuracy
Stereotactic Radiosurgery, AAPM Report No. 54, 1995
Other sources:
MRI Distortion
Image Fusion
Relocatable frames
Dosimetric
……
1
Frames & CT
What About Body Frames?
Invasive
Non-invasive
Maciunas et al, Neurosurgery 35:682-695, 1995
First successful spinal application
Rigid skeletal fixation above and below
the involved segments
Linac delivery with circular collimators /
arcs or IMRT
System accuracy ≤ 2.0 mm
8-10 Gy Rx with no portion of cord
receiving > 3 Gy
Stereotactic Body Frame
Reproducibility 5-8 mm for 90% of setups
Diaphragmatic movement limited to 5-10 mm
with pressure.
Lax et al, Acta Oncol. 33:677-83, 1994
Setup accuracy evaluated in 30 patients
using CT and port films; conclude that a 5
mm margin for PTV is sufficient if CT is
performed prior to every treatment. Deviation
of < 10 mm (AP and Lat) in 98% of targets.
Wulf et al, Radiotherapy Oncology 55:225-236, 2000
Hamilton et al, Neurosurgery 36; 311-319, 1995
Other Frame Based SBRT Systems
Negoro et al, IJROBP 50:889-898, 2001
In the initial characterization of the stereotactic
body frame, the authors observed what
reproducibility in patient setup 90% of the time:
1.
less that 2.0 mm
2.
2.0 to 5.0 mm
3.
5.0 to 8.0 mm
4.
8.0 to 11.0 mm
5.
> 12.0 mm
Yenice
Lohr
al,al,
IJROBP
IJROBP
45:521-527,
55:583-593,
1999
2003Herfarth et al, IJROBP 46:329-335, 2000
Shiu etet
57:605-613,
2003
SBRT QA Rule #1:
NO SBRT without modern image guidance
2
In the initial characterization of the stereotactic
body frame, the authors observed what
reproducibility in patient setup 90% of the time:
1.
less that 2.0 mm
2.
2.0 to 5.0 mm
3.
5.0 to 8.0 mm
4.
8.0 to 11.0 mm
5.
> 12.0 mm
Correct answer: 3
Isocentric Accuracy:
The Winston-Lutz
Test
Lax et al, Acta Oncol. 33:677-83, 1994
Mechanical Uncertainties
Is the projection of the
ball centered within
the field?
A daily Lutz test is extremely
important because:
• The mechanical isocenter
can shift over time
• The AMC board in Varian
couches can fail
• The cone or MLC may not be
repositioned perfectly after
service
Before (right) and After (left)
relatively simple couch adjustment
Isocentric Accuracy:
The Winston-Lutz
Test
Is the projection of the
ball centered within
the field?
Good results ≤ 0.5 mm
A Lutz test with the MLC is also
important because:
• The Cone-based Lutz test does not
tell you anything about the
mechanical isocenter of the MLC
• The MLC may not be repositioned
perfectly after service
Lutz test with 12 x 12 mm MLC field
3
End-to-end testing: evaluating
multiple sources of uncertainty
Structure
Cylinder
Cube
Cone
Sphere
AP
0.0
20.0
-35.0
25.0
Phantom Specifications
LAT
VERT
0.0
30.0
-17.0
40.0
-20.0
40.0
20.0
32.7
iPlan Stereotactic Coordinates
AP
LAT
VERT
1.0
0.4
30.8
20.8
-17.1
42.4
-34.6
-19.7
40.8
25.5
20.2
33.5
Verification of MLC
shapes and isocenter
System Accuracy
Simulate the entire
procedures: Scan,
target, plan, deliver
Resulting film provides measure of
targeting accuracy …
Offset from
intended target
Phantom with
film holder
Pin denotes
isocenter
4
… as well as falloff for a multiple arc delivery
1
RPC Phantom
Hidden Target Test
scan, plan,
localize assess
0.8
0.6
0.4
0.2
0
-1
-0.5
0
0.5
Lucy Phantom
1
Off Ax is Distance (m m )
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
Imaging Uncertainties
• Use CT for geometric accuracy
• Use MR for target delineation
“MRI contains distortions which
impede direct correlation with CT
data at the level required for SRS”
Stereotactic Radiosurgery – AAPM Report No. 54
1.8 ± 0.5 mm shift of
MR images relative to
CT and delivered dose
Shifts occur in the
frequency encoding
direction
Due to susceptibility
artifacts between the
phantom and fiducial
markers of the
Leksell localization
box
Other References
TS Sumanaweera, JR Adler, S Napel, et al., Characterization of
spatial distortion in magnet resonance imaging and its implications
for stereotactic surgery,” Neurosurgery 35: 696-704, 1994.
Y Watanabe, GM Perera, RB Mooij, “Image distortion in MRI-based
polymer gel dosimetry of Gamma Knife stereotactic radiosurgery
systems,” Med. Phys. 29: 797-802, 2001.
What do we do about MR spatial distortion?
Use Image Fusion
Frequency Encoding = L/R
Frequency Encoding = A/P
Y Watanabe, GM Perera, RB Mooij, “Image distortion in MRI-based
polymer gel dosimetry of Gamma Knife stereotactic radiosurgery
systems,” Med. Phys. 29: 797-802, 2001.
5
Fusion Verification
But image fusion will not work well if
anatomy has deformed
According to Watanabe et al (2001), image distortion in
stereotactic MR imaging:
According to Watanabe et al (2001), image
distortion in stereotactic MR imaging:
1. is completely random
1. is completely random
2. has a systematic component of approximately 2 mm
in the frequency encoding direction
2. has a systematic component of approximately
2 mm in the frequency encoding direction
3. has a systematic component of approximately 2 mm
in the phase encoding direction
3. has a systematic component of approximately
2 mm in the phase encoding direction
4. has a systematic component that depends on the MR
slice thickness
4. has a systematic component that depends on
the MR slice thickness
5. is insignificant and can be ignored for stereotactic
applications
5. is insignificant and can be ignored for
stereotactic applications
Correct answer: 2
Y Watanabe, GM Perera, RB Mooij, “Image distortion in MRI-based polymer gel
dosimetry of Gamma Knife stereotactic radiosurgery systems,” Med. Phys. 29:
797-802, 2001.
Small field depth dose show familiar trends
Dosimetric Uncertainty
120
Stereotactic Diode
6
12
18
24
30
36
42
100
Percent Depth Dose
Standard Diode
Small field measurements
can be challenging; Diodes
and small ion chambers
are well suited to SRS
dosimetry, but their
characteristics / response
must be well understood.
80
60
80
100
60
40
20
0
0
50
100
150
200
250
300
350
400
Depth (mm)
Pinpoint Chamber 0.015 cc
6
Similar machines have similar characteristics
Penumbra: Cones versus MMLC
3.5
120
36A
12A
36 UNMC
12
Penumbra, 80%-20% (mm)
Percent Depth Dose
100
80
60
40
20
0
0
50
100
150
200
250
300
350
Cones
MMLC
3
400
2.5
2
1.5
1
0.5
Depth (cm)
0
0
20
40
60
80
100
Nominal Field Size (mm)
Diode Warnings!!!
3) Reference diode output to
an intermediate field size
1) Diodes exhibit enough energy dependence that
ratios between large and small field measurements
are inaccurate at the level required for radiosurgery
Measure output factors ≤ 2x2 cm2 using diode
≥ 2x2 and ≤ 4x4 cm2 using diode & chamber
≥ 4x4 cm2 using chamber
Output
Factor
=
Reading (FS)diode
Reading (Ref’)diode
X
Reading (Ref’)IC
Reading (Ref)IC
2) Diode response will drift over time
Re-measure reference between each chance in
field size
Radiosurgery beams exhibit a sharp decrease
in output with decreasing field size
3) Reference diode output to
an intermediate field size
Significant
uncertainty
NO!
YES!
Reading (6 mm)diode
Reading (100 mm)diode
Reading (6 mm)diode
Reading (24 mm)diode
X
Reading (24 mm)IC
Reading (100 mm)IC
Don’t use high energy
This means that with small collimators,
treatment times can be long
7
Need proof that beam data
acquisition for small
fields is difficult?
1.20
Total Scatter Factor @ 98.5 cm SSD and 1.5 cm depth relative to 10 x 10 cm2 Field
Totlal Scatter Factor
1.00
0.80
0.60
Surveyed Beam Data from 40 identical treatment units:
0.40
Percent Depth Dose
Relative Scatter Factors
0.20
Absolute Dose-to-Monitor Unit CF
Reference Condition
0.00
0.0
20.0
40.0
60.0
80.0
100.0
120.0
Applied statistical methods to compare data
Collimator (mm)
Observations of some treatment units: 6 mm x 6 mm MLC
110
110
100
90
Institution A
90
80
InstitutionAB
Institution
Institution C
80
70
Percent Depth Dose
Percent Depth Dose
100
60
50
40
30
Institution A
Institution B
6.0 %
70
10.3 %
60
50
40
30
20
20
10
10
0
0
50
100
150
200
250
300
350
400
450
0
Depth (mm)
0
50
100
150
200
250
300
350
400
450
Depth (mm)
Observations of some treatment units: 5 mm collimator
Observations of some treatment units: 15 mm collimator
1.2
1.2
1.2
1.2
11
Institution A
Institution A
Institution B
11
Tissue Maximum Ratio
Institution A
A
Institution
Institution B
0.8
0.8
0.6
0.6
0.4
0.4
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0.2
0.2
00
00
00
50
50
100
150
Depth
(mm)
Depth (mm)
200
250
300
00
50
50
100
100
150
150
200
200
250
250
300
300
Depth (mm)
(mm)
Depth
8
Relative Output Factor: 6 mm x 6 mm MLC
Relative Output Factor: 42 mm x 42 mm MLC
Output Factor
Output Factor
~45%
~6%
Institution
Institution
Commissioning your system: Does calculation
agree with measurement?
End-to-end testing
Dosimetric uncertainty
Calculation
Calculation
arc-step
= 10o
arc-step = 2o
Relative Dosimetry
End-to-end testing
Dosimetric
uncertainty
1 isocenter
4 field box
Dynamic Conformal Arcs
2 isocenters
Absolute Dosimetry
2 isocenters
IMRT
9
Difference between calculation and measurement:
the first 160 IMRT patients
Independent MU
Calculations
x = 0.26 1σ = 1.75
Agazaryan et al, JCAMP, 2003
End-to-end testing
Evaluating multiple
sources of uncertainty
Conventional
Localization
RPC SRS Phantom
Does not include MRI imaging
Does not include image fusion
Provides only a spherical target
Hidden Target
Precludes image-guided positioning
image
image-guided
Point dosimetry limited by TLD accuracy
Limited information on dose distribution
Test results require return of phantom
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital Dosimetry Insert
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
10
Absolute Dosimetry
Lucy Prototype
Imaging Insert
MR contouring and
dosimetry based on
MR contours
Circular volumes
for image fusion
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
MR Fusion
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
What about “Frameless Systems?”
Treatment Plan and
Film Dosimetry
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
kV cone beam CT
A “frameless” stereotactic system provides
localization accuracy consistent with the safe
delivery of a therapeutic dose of radiation
given in one or few fractions, without the aid
of an external reference frame, and in a
manner that is non-invasive.
Frameless stereotaxis is inherently image guided
Also required:
Immobilization – need not be linked to localization
Ability to periodically monitor / verify
11
(Stereo)photogrammetry the principle behind
frameless technologies
Stereophotogrammetry in Radiotherapy
Spatial Resolution: 0.05 mm
Photogrammetry is a
measurement technology
in which the threedimensional coordinates
of points on an object
are determined by
measurements made in
two or more
photographic images
taken from different
positions
Temporal Resolution: 0.03 s
Localization Accuracy:
0.2 mm
Optical
Photogrammetry
Stereophotogrammetry in Radiotherapy
In a rigid object, optical stereophotogrammetry can
achieve a localization accuracy of approximately:
1. 0.2 mm
2. 0.5 mm
3. 1.0 mm
4. 1.5 mm
5. 2.0 mm
Bova et al, IJROBP, 1999
In a rigid object, optical stereophotogrammetry can
achieve a localization accuracy of approximately:
1.
0.2 mm
2.
0.5 mm
3.
1.0 mm
4.
1.5 mm
5.
2.0 mm
X-ray Stereophotogrammetry
Correct answer: 1
Schlegel et al, Radiotherapy and Oncology 29:197-204, 1993
Bova et al, IJROBP 38:875-882, 1997
12
X-ray Stereophotogrammetry
Calibration of IGRT Systems
Establish spatial
dimensions
Specify the
isocenter
Frameless (Image Guided) Radiosurgery
Calibrate the x-ray system
How do we know the system is targeting properly?
End-to-end evaluation that mimics a patient procedure
X-ray
Identify target & plan
Results of Phantom Data
DRR
(mm)
Lat.
Long.
Vert.
3D vector
Average
-0.06
-0.01
0.05
1.11
Standard
Deviation
0.56
0.32
0.82
0.42
Set up in treatment room
Irradiate
• Sample size = 50 trials (justified to
95% confidence level, +/- 0.12mm)
Evaluate
1.2
Lateral
Sup/Inf
Ant/Post
Single Fraction
1
0.8
0.6
0.4
0.2
0
-8.0
Comparison in 35 SRS patients
and 565 SRT fractions
-6.0
Single
Fraction
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
(mm)
AP
Lat
Axial 3D vector
Average
0.05
-0.06
0.26
1.01
Standard
Deviation
0.55
0.67
0.72
0.54
13
1.2
Lateral
Sup/Inf
Ant/Post
Multiple Fractions
1
1.2
Superior / Inferior
Multiple Fraction
Single Fraction
1
0.8
0.8
0.6
0.6
0.4
0.4
0.2
0.2
0
-8.0
-6.0
Multiple
Fraction
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
(mm)
AP
Lat
Axial 3D vector
Average
0.17
0.17
0.47
2.36
Standard
Deviation
1.03
1.24
2.11
1.32
0
-8.0
-6.0
-4.0
-2.0
0.0
2.0
4.0
6.0
8.0
• Frameless localization appears equivalent to frame-based rigid fixation
• Frameless localization improves accuracy of relocatable frames
Localization using implanted fiducials
Localization using
implanted fiducials
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
Extracranial Applications - Spine
Courtesy Sam Hancock, PhD, Southeast Missouri Hospital
Extracranial Applications - Spine
14
Extracranial Applications - Liver
What do you see on oblique radiographs of
the chest or abdomen?
Extracranial Applications - Lung
The Better Approach
Implanted Markers
Not much
Lots of redundancy
End-to-end evaluation:
Extracranial
• Patient is marked
• Immobilization device
is marked
• Location of IR markers
is recorded
• Table heights and
shifts are provided
3D error 1.2 ± 0.4 mm
15
End-to-end evaluation: CyberKnife
End-to-end evaluation: Calypso
3D error 1.1 ± 0.3 mm
3D σ = 0.34 mm (N=20)
Chang et al, Neurosurgery 2003
Management of Respiratory motion is essential
Can’t rely on the
“Ignorance is Bliss”
concept anymore
Active Breathing Coordinator
Eccles et al, IJROBP 64:751-759, 2006
Monitor the motion and
limit it if necessary
Courtesy Laura Dawson, PMH
16
Radiosurgery Guidelines
ASTRO/AANS Consensus Statement on stereotactic radiosurgery
quality improvement, 1993
RTOG Radiosurgery QA Guidelines, 1993
AAPM Task Group Report 54, 1995
European Quality Assurance Program on Stereotactic
Radiosurgery, 1995
DIN 6875-1 (Germany) Quality Assurance in Stereotactic
Radiosurgery/Radiotherapy, 2004
AAPM Task Group 68 on Intracranial stereotactic positioning
systems, 2005
ACR Practice Guidelines for the Performance of Stereotactic
Radiosurgery, 2006
ACR Practice Guidelines for the Performance of Stereotactic Body
Radiation Therapy, 2006
AAPM Task Group 101, Stereotactic Body Radiotherapy, 2008
17