Acknowledgements
The Role of Imaging in IMRT
n
n
n
n
James F. Dempsey, Ph.D.
Department of Radiation Oncology
University of Florida College of Medicine
Gainesville, Florida USA
n
n
n
n
n
n
n
Outline
n
n
Why IMRT Needs Quantitative Imaging
The “Theory” of Quantitative Imaging
n
n
n
Univ. of Florida
Univ. of Florida
Univ. of Florida
Univ. of Florida
Univ. of Florida
Washington University
M.D. Anderson
Washington University
Washington University
Washington University
Washington University
Clinical Motivation:
Conformality => Better Sparing
Sampling & Reconstruction
Filters for Linear Systems
Survey of the Roles of Quantitative Imaging in
IMRT
n
n
n
n
n
Omar A. Zeidan, Ph.D.
Tony Stell
Chihray Liu, Ph.D.
Jonathan G. Li, Ph.D.
Jatinder R. Palta, Ph.D.
Daniel A. Low, Ph.D.
Clifford K.S. Chao, M.D.
Jeffrey F. Williamson, Ph.D.
Sasa Mutic, M.S.
Robert Malyapa
Malyapa,, M.D., Ph.D.
Perry W. Grigsby, M.D.
Imaging
Imaging
Imaging
Imaging
for
for
for
for
Simulation & Treatment Planning
Target Delineation
Dose Measurement
Delivery Validation
Methods of Delivery Validation via Imaging
Screen Capture from Nomos Peacock Tx Planning System
MIR Washington Univ.
Parotid Sparing in the Presence of
Systematic Setup Error
Effect of Lateral Shift on Left Parotid Dose
Systematic Set-up
errors can have a
significant impact
on critical structure
sparing
1
0.9
Fraction Volume
0.8
15 mm
0.7
0.6
Parotid Sparing in the Presence of
Random SetSet-Up Error
Well No,
With IMRT we
have purposely
placed the parotid
in a high gradient
10 mm
0.5
0.4
5 mm
Correct
0.3
0.2
0.1
0
0
10
20
30
40
50
60
70
80
But Random SetUp Errors will
“wash out” any
overdosing of the
parotid, right?
Dose (Gy)
1
Questions for IMRT
n
n
n
How Do We Ensure Our IMRT Dose Calculations
are Accurate?
How Do We Ensure That Can Achieve Our
Clinical Goals?
How can we best employ IMRT to the greatest
efficacy?
What is Quantitative Imaging?!
n
n
n
n
Could Quantitative Imaging Be the Answer?
n
Where Do We Need Quantitative
Imaging Techniques in IMRT?
n
Imaging for Simulation & Treatment Planning
n
n
n
n
CT based Heterogeneity Corrections
Breath-- Hold Gated CT
Breath
CT or MR Cine’ for Organ Motion
Imaging for Target Delineation
n
n
Imaging for Delivery Validation
n
Imaging for Dose Measurement
n
n
n
n
Fluence Map Validation
n
n
n
Film Dosimetry
Gel Dosimetry
Exit Dosimetry
All of these applications attempt to extract
quantitative information from imaging
We must 1st ask, how do we know that
the imaging devices are capable of
measuring the distributions that we seek?
Secondly, If the device fails to accurately
measure our distribution can we correct or
recover the information?
How to Perform Quantitative
Imaging
n
n
n
n
n
1) Sample Data with High Enough Frequency
2) Characterize the Linearity and Spatial
Independence of the imaging system
3) Determine Line Spread and or Point Spread
functions and Modulation Transfer Functions
4) Evaluate the Ability to Make Quantitative
Measurements
5) Apply Filters to Recover Information if
necessary
e.g. for diagnosis
Radiation Therapy Attempts to Perform
Quantitative Measurements Using
Imaging Techniques
Lets Look at the applications for IMRT
Theory of “Quantitative” Imaging
Pet Registration & Fusion
MR Registration & Fusion
n
An Oxymoron?...
Most “Imaging Science” is concerned with
Qualitative Feature Extraction
Shannon-Nyquist Sampling
ShannonTheorem (1)
n
If a function has all of its frequency components
below some frequency n, then sampling that
function with frequency 2ν
2ν allows for the exact
reconstruction of that function
F(x)
F(ω)
ν
x [dist]
ω [freq.]
2
Shannon-Nyquist Sampling
ShannonTheorem (1)
n
n
Example of the Use of Sampling
Theory (1)
With this theorem and your knowledge of
Medical Physics you can determine the temporal
and spatial resolutions at which distributions
should be sampled to prevent aliasing!
For example, if we know a priori that a dose
distribution takes on a functional form we can
determine the required sampling as follows …
A fitted 1x1 cm2 profile from 0.1 mm pixel RCF
measurements in solid water
Practical Sampling Theory (1)
Linear Systems Theory (2)
n
A system (a transformation T) is considered
“linear” if it have the following properties:
Τ[a × f (x ) + b × g ( x)] = a × F (ϖ ) + b × G (ϖ )
^
In other words, superposition and scaling hold under the
transformation
Here a system is though to be an imaging system of
course!
Linear Systems and Imaging (2)
n
n
n
n
If an imaging system can be approximated
as a linear system …
And any process that degrades the output
of the imager is a linear process …
And the degrading process is spatially or
temporally invariant …
Then…
Spread Functions (3)
n
Convolution with Spread Functions can be used
to characterize the degradation of the imaging
data.
Signal( x) = TrueSignal( x) ⊗ SpreadFucntion( x)
True Signal
Imaged
Signal
Spread
Function
3
Modulation Transfer Function
(MTF) (4)
Spread Functions (3)
n
n
n
n
Spread functions plot the redistribution of data
in a pixel due to a degradating process
Line Spread Functions (LSF) characterize 1D
process
Point Spread Functions (PSF) characterize 2D (or
3D) process
If a PSF is isotropic it can be related to the LSF
by the Abel Transform
The Fourier Transform of the LSF or the Hankel
Transform of the PSF produce the MTF: The MTF
needs to be very close to a value of unity for
quantitative systems !
The Fourier Convolution Theorem
Can Be Used to Recover Data (5)
Quantitative Imaging Summary
n
n
n
n
n
~
~
~
1) Sample Data with High Enough Frequency
2) Characterize the Linearity and Spatial
Independence of the imaging system
3) Determine Line Spread and or Point Spread
functions and Modulation Transfer Functions
4) Evaluate the Ability to Make Quantitative
Measurements
5) Apply Filters to Recover Information if
necessary
TrueSignal (ϖ ) = Signal (ϖ ) / SpreadFucn tion (ϖ )
Virtual simulation for
IMRT Using XX-Ray CT
Brief Survey of Imaging in IMRT
n
Imaging for Simulation &
Treatment Planning
n
n
Due to it’s inherent
spatial integrity XXray CT remains the
“Gold Standard”
for Tx Planning
CT Data can allow
dose calculations
to account for
Tissue
Heterogeneities
We are relying
on accurate
CT#s
4
Heterogeneities in IMRT
CT Artifacts
•Artifactual CT Numbers can be produced in the presence of object that can
•Spatial Integrity of the Data is not compromised
•Assigning Bulk Densities to Structures can overcome Heterogeneity corrections
to Artifactual CT Number regions
•IEEE Trans Med. Imaging 2001 20(10) 1009-1017 Snyder and Williamson et al.
Heterogeneities can cause large
discrepancies in the presences of air
cavities for IMRT.
Shown: CCC w/ and w/o heterogeneities
for nasopharyngeal target
High Quality DRR and Portal Films
are Critical to IMRT
Port Films are required to verify isocenter to within a few mm
Spatial Integrity is of Utmost Importance: Feature Extraction is qualitative
Serial Tomotherapy DRR and Port Film
Adaptive Radiotherapy & IMRT
MegaVoltage CT
Adaptive Radiotherapy & IMRT
X-Ray Cone Beam Reconstruction
Cone Beam Imaging at The Accelerator can Provide Daily Setup
Based on the Imaging of Soft Tissues!
DA Jaffray, DG Drake, M Moreau, AA
Martinez, and JW. Wong, Int. J. Radiation
Oncology Biol. Phys., Vol. 45, No. 3, pp.
773–789, 1999
Brief Survey of Imaging in IMRT
Helical Tomotherapy and Megavoltage CT Imaging at The Accelerator
Imaging for Target Delineation
K J Ruchala, G H Olivera , E A Schloesser and T R Mackie
Phys. Med. Biol. 44 (1999) 2597–2621.
5
Virtual Simulation Software QA
MULTIMODALITY IMAGE REGISTRATION
n
n
Validation of the
Complete Clinical
Process is Important:
Mutic et al.
Multimodality image
registration quality
assurance for
conformal threethreedimensional treatment
planning. Int J Radiat
Oncol Biol Phys, 2001
& FUSION
MRI--CT Image FusionMRI
Fusion -Registration
• Magnetic Resonance Imaging (MRI)
• Excellent soft tissue contrast allows
better differentiation between
normal tissues and many tumors
a)
CT1
b)
CT2
c)
d)
MR
PET
CU--ATSM PET
CU
PET--CT Image Registration &
Target Definition
• It is not limited to imaging in axial
planes
• Disadvantages:
MR
– Susceptible to spatial
distortions
– Image intensity values
do not relate to physical
or electron density
CT
Target Delineation By PET – CT
Registration & Target Definition
CT Scan
FDG-PET Scan
Kidney
Para-aortic L.N.
R. Malyapa et al. MO-D-517A-07
Chao et al. Int J Radiat Oncol Biol Phys 2001 Mar
15;49(4):1171-82
Imaging for Lung IMRT?
n
n
Attractive IMRT Site Due to Need for
Conformal Dose Distributions and Target
Encompassing Critical Structures
Significant Concern Over Loss of
Superposition Due to Breathing Motion
and Significant Heterogeneity Corrections
Spirometer Gated Multislice CT
Multi -slice CT gated on tidal
lung volume as measured by
digital spirometry during free
breathing
Reconstruction is performed
using sinograms that are
assembled by the coincident
spirometer reading
Produces time dependent CT
data to map out breathing
motions
D. Low et al. WED-C-517D
6
Brief Survey of Imaging in IMRT
Radiographic Film Dosimetry for
Patient Specific QA
Imaging for Dose Measurement
Radiochromic Film Dosimetry for
Commissioning IMRT Treatment
Planning Systems
Polymerizing Gel Dosimetry in
IMRT
600
900
1200
1400
300
n
n
n
Imaging for IMRT Delivery
Validation
MLC Delivery Device
Characterization via Film
How Do We Validate the Complex
Orchestration of MLC Leaf Motion and
Dose Delivery That Occurs in SMLC - and
DMLC--Based IMRT?
DMLC
Imaging of the integral dose or fluence
with film or EPID
Time Resolved Imaging of the Fluence or
Dose
•Radiographic meas. of accuracy of leaf travel/ offset
And Integral Fluence Maps
LoSasso T et al. Med.
Phys. 2001;
28(11):2209-2219.
7
Accelerator Log Files ~20 FPS
(~50 ms sample rate)
EPID and IMRT
Delivery Validation
EPID Records Integral or Time Resolved Exit Dose to Validate
Fluence Map Delivery: Most Time Resolved Systems <10 FPS
LoSasso T et al. Med.
Phys. 2001;
28(11):2209-2219.
Partridge M et al., Med. Phys. 2000;
27(7):1601-1609.
Xia P et al. Med.
Phys. 2002;
29(3):412-423.
Ezzell and Chungbin J
Appl Clin Med Phys
2001; 2(3):138-148.
Litzenberg et al. J
Appl Clin Med Phys
2002; 3(2):63-72.
Scintillation-CCD Camera Validation
Scintillationof MLC Delivery
Scintillation-CCD Camera
ScintillationCharacterization
n
n
n
n
Scintillation-CCD Camera
ScintillationCharacterization
n
n
n
Fast Gd2O2S:Tb Scintillation Plate coupled to a
CCD or CMOS Camera for high frame speed
capture.
Only 1.24 g/cm2 stainless steel, 0.411 g/cm2
Gd2O2S:Tb, and 0.008 g/cm2 aluminized mylar
in the beam path: allows verifcation upstream of
the patient
CCD Camera up to 30 FPS
CMOS Camera up to 1000 FPS (200 FPS is good
enough)
Scintillation Camera Validation of
IMRT Delivery
Sampling in Time & Space
Demonstration of Linearity as well as
Spatial and Temporal Invariance
Spatial PSFs,Temporal LSFs
LSFs,, and
associated MTFs
Gd 2 O2S:TB Scintillation
plate
Camera:
30 fps CCD Camera
1000 fps CMOS Camera
Observation of Leaf Motion During SMLC Delivery
8
n
30 FPS
Comparison of Log Files, Camera,
and the Intended Delivery
100 FPS
Much Ado About Nothing?
Feedback Treatment Planning
What Should I Do when
Quantitative Imaging is used In My
Clinic
Imaged MLC Delivery Errors are accounted
for in a recalculated Treatment Plan
n
n
n
Understand the Issues
Understand the Theory
Insist that vendors of a device that is intended
for quantitative imaging measurement
demonstrate
n
n
n
Linearity
LSFs,, PSFs
LSFs
PSFs,, and MTFs where appropriate
Use your knowledge of dosimetry to evaluate
the practical limits of the device
9
One should ask:
“Does my Imaging
system accurately
measure
...with high enough
resolution?
… in high gradients?
Where are the LSF,
PSF, or MTF?
You should not assume
it will be correct just
because I paid a lot of
money for it!
10