Download IMAGE GUIDED RADIATION THERAPY (IGRT)

IMAGE GUIDED RADIATION THERAPY
(IGRT)
Manish Kakar,PhD
YS
47
3
•
Margin reduction
Advanced Methods (State of the Art in IGRT)
ka
rF
•
•
•
Ka
•
Intelligent Systems
4DCT & Adaptive Radiation Therapy
h
•
is
•
What is IGRT ?
Imaging Devices
Errors and Margins
Case Study - Respiratory Motion
M
an
•
0
OUTLINE
WHAT IS IGRT ?
•
YS
is
h
•
ka
rF
•
Treatment decisions (How/whether to treat)
Delineation (ROI)
( )
Positioning
Assessment of the outcome (treatment
Verification)
Ka
•
47
3
0
Image Guided RT (IGRT) is the use of
imaging to guide Radiation Therapy for
M
an
•
h
is
M
an
YS
ka
rF
Ka
IMAGING DEVICES
0
47
3
EPID
Electronic Portal Imaging Device (EPID):
•
•
0
47
3
M
an
is
h
•
YS
•
Name given to Digital flat panel devices.
Metal plate + phosphor screen
Treatment x-rays cause the phosphor to glow
resulting
lti g light is
i detected
d t t db
by an array off
photodiodes
ka
rF
•
Ka
•
Ex.Verification (IMRT);
FLOROSCOPY
•
A fluoroscope
p consists
X-ray source + fluorescent screen
patient is placed in between these
h
Ka
ccd camera (recording and playing on a monitor)
is
•
ka
rF
Modern fluoroscope
M
an
•
YS
•
47
3
0
•
FLOROSCOPY - APPLICATIONS
•
•
47
3
YS
ka
rF
Ka
Disadvantage
h
•
Real-time moving images of the internal structures
of a patient
Pose a potential health risk to the patient
Long exposure times
is
•
0
Usage
M
an
•
47
3
YS
ka
rF
probe
b (Transducer)
(T
d
)
placed directly over the patient
•Reflections from pulses recorded
and displayed
h
Ka
•Hand-held
H d h ld
is
•
Ultrasounds – Cyclic pressure
(created by sound waves) at
frequencies greater than human
hearing (20 KHz)
In medicine – Ultrasound
sonography
M
an
•
0
ULTRASOUND
ULTRASOUND - APPLICATION
47
3
0
Usually used for determining
YS
h
Ex: Prostate thickness and depth of rectum –
measured by Ultrasound for determining dose
to prostate and rectum.
is
•
ka
rF
•
Localization of the target and
D th off the
Depth
th target
t g t
Ka
•
M
an
•
MRI
•
Ka
h
is
•
M
an
•
ka
rF
YS
•
47
3
0
•
Human bodyy is composed
p
of water molecules (p
(protons))
Placed Inside a powerful magnetic field protons align with
the direction of the magnetic field
Additional RF electromagnetic field is briefly turned on
causing the protons to absorb some of its energy
When this field is turned off protons release this energy at
radio frequency which is then detected by the scanner
Position of protons is determined by applying additional
magnetic field (on and off) during scanning allowing image
of the body to build up.
MRI - APPLICATIONS
Used for:
Geometric Distortion :
h
•
Ka
Disavdvantages
is
•
Better target delineation
YS
•
47
3
0
Improved
p
soft tissue visualization
ka
rF
•
•
M
an
•
Variation in Magnetic field - can result in errors in
treatment planning
FAN BEAM CT / CONE BEAM CT
is
h
Ka
ka
rF
YS
47
3
0
CBCT provides faster volumetric imaging
M
an
•
h
is
M
an
YS
ka
rF
Ka
0
47
3
VIDEO - CBCT
OPTICAL TRACKING
RPM System
S
– Real-time
R l i
position management
system by varian
M
an
is
h
Ka
ka
rF
YS
47
3
0
•
•Reflective
marker block
placed on the xyphoid region
of the patient.
•Marker movement recorded
and sampled via infrared
camera into a respiratory
signal.
4DCT
47
3
Ka
•
Snapshot in time
May not represent the true position, volume,
shape or trajectory of a moving tumor
YS
•
0
A 3D CT image
ka
rF
•
is
h
4DCT - 4th Dimension (TIME)
•
M
an
•
Captures temporal changes from moving
anatomy
h
is
M
an
YS
ka
rF
Ka
0
47
3
h
is
M
an
YS
ka
rF
Ka
0
47
3
VIDEO – 4DCT
4DCT - APPLICATIONS
4D CT image dataset used for
0
47
3
YS
ka
rF
Ka
h
•
Devisingg breathingg motion model
Normally used with 4DCT & Adaptive Radiation
Therapy (ART) – Advanced planning and delivery
techniques
is
•
M
an
•
0
47
3
YS
ka
rF
M
an
is
h
Ka
ERRORS AND TREATMENT MARGINS
IMAGING ERRORS
Snapshot provided by imaging may not be
correct for all stages of treatment
Sources of error
Ka
ka
rF
YS
•
h
•
Definition of tumor
Setup of the patient
Moving anatomy
is
•
M
an
•
47
3
0
•
VOLUME DEFINITIONS
GTV (Gross Tumor Volume) :
•
•
C t i GTV and/or
Contains
d/ subclinical
b li i l microscopic
i
i malignant
li
t ti
tissues tto b
be iirradiated
di t d
PTV (Planning Target Volume): CTV plus a margin to account for
• Internal organ motion
• Patient setup variations.
variations
• Patient motion during treatment.
OAR (Organs at risk) : Critical normal tissues whose radiation sensitivity may
influence treatment p
planning/dose
g/
p
prescription
p
Ex. Spinal
p
cord
is
M
an
•
h
Ka
ka
rF
•
0
CTV (Clinical Tumor Volume):
47
3
•
Gross visible/demonstrable extent of the malignant growth
YS
•
h
is
M
an
YS
ka
rF
Ka
0
47
3
TUMOR DELINEATION ERRORS
Sources of error:
•
0
is
h
Errors here influences the whole treatment
Can be improved by Automatic Segmentation
M
an
•
Ka
ka
rF
•
Interobserver variations ((manual delineation))
Limited resolution of imaging devices – Partial
Volume effect.
47
3
•
YS
•
SETUP ERRORS
47
3
0
Errors due to patient position and orientation
from fraction to fraction (interfractional)
C G
Ca.
Gaussian
i Di
Distributed
t ib t d
Can be systematic
y
and Random
Ka
h
is
•
M
an
•
ka
rF
YS
•
SYSTEMATIC AND RANDOM ERRORS
Systematic
•
•Originate
- from setup of the
patient for a particular treatment
fraction
•Obs - Vary from fractions to
f
fractions
for
f a single patient
treatment
is
h
Ka
ka
rF
YS
47
3
0
– from patient
planning stage.
stage
•Obs - Constant across all
fractions for a given patient
Random:
•Originate
M
an
•
INTERNAL ORGAN MOTION
M i g anatomy
Moving
t
effects
ff t
0
ka
rF
•
tumors in thorax and abdominal regions
Causes blurring of dose distribution along path
of motion
47
3
•
YS
•
is
h
Ka
Ex. : Breathing Motion (Mainly
intrafractional)
M
an
•
DISTORTION DUE TO INTERNAL
ka
rF
YS
47
3
0
TARGET MOTION
M
an
is
h
Ka
Tumor
Distorted
Undistorted
CURRENT PRACTICES -MARGINS
YS
47
3
0
To overcome the effects of errors and
uncertainities margins are
included/extended
Ka
h
is
CT
V
ka
rF
PTV
M
an
•
Extended Margin
PTV
CT
V
Conventional
CORRECTION STRATEGIES
Setup errors:
47
3
0
Offline:Error accumulated through
g treatment
sessions and corrected at the next fraction
YS
•
h
Ka
Online:Corrections made at start of each fraction
•
is
•
Good for removal of Systematic
y
errors
ka
rF
•
Good for removal of Systematic & Random errors
M
an
•
CORRECTIONS FOR MOVING ANATOMY
0
47
3
ka
rF
YS
•
Large margins
Gating, Breath hold, tracking
Ka
•
h
•
is
•
Non Gaussian Distributed (breathing motion)
Variability from cycle to cycle.
Corrections are normally made by
M
an
•
0
47
3
M
an
is
h
Ka
ka
rF
YS
CASE STUDY – RESPIRATORY
MOTION
O O COMPENSATION
CO
S O
RESPIRATORY MOTION
•
YS
ka
rF
Ka
•
•
•
Addingg margins
g to CTV
h
´
is
´
M
an
•
47
3
0
•
Respiratory motion effects all tumors in abdomen and
thorax
th
Target Motion dominant than setup errors (intra-fractional)
Precise delivery of dose becomes more challenging
– tumour under dosage
– healthy tissue over dosage
Solution
Drawback: Irradiaties uneccesary tissue
BREATH HOLD TECHNIQUES
Voluntary breath holding by patient:
•
0
47
3
YS
ka
rF
Disadvantages:
•
Ka
•
h
•
Ex. Active Breath Control/Coordinator (ABC)
External scissor valve activates breath control
Pateint controls the switch (pressed – breath hold, released –
breath normally)
is
•
Uncomfortable for patient
M
an
•
h
is
M
an
YS
ka
rF
Ka
0
47
3
BREATH HOLD
GATING
Selectively treat a moving target by
electronicallyy tuningg on and off at specified
intervals.
RPM System (Real time position
Management System):
Ka
h
is
•
•
•
M
an
•
ka
rF
YS
47
3
0
•
Infrared Camera
Passive marker on Xyphoid
yp
region
g
of p
patient
Processing of tracked motion
h
is
M
an
YS
ka
rF
Ka
0
47
3
GATING
YS
47
3
0
EXAMPLE: MARGIN REDUCTION
Ka
h
is
M
an
CT
V
ka
rF
PTV
Conventional
PTV
CT
V
With gating
DRAWBACKS OF GATING
E t
External
l Gating
G ti g
Inaccuracy due to correlation b/w internal motion
and marker.
•
Tracking
•
h
Ka
Implanted fiducial markers - risk of pneumothorax
is
•
ka
rF
Internal Gating :
M
an
•
YS
47
3
•
0
•
Direct tumor tracking and targeting techniques
required without implanting fiducial markers
M
an
is
h
Ka
ka
rF
YS
47
3
0
TRACKING - CYBERKNIFE
TRACKING -ROBOTIC SURGERY
0
47
3
YS
ka
rF
Ka
•
h
•
is
•
Linac mounted on a robot manipulator
6 degrees of freedom
Fudicial markers are placed on the chest wall
Can take several minutes to hrs between
t t
treatment
t
M
an
•
•
Not continous
h
is
M
an
YS
ka
rF
Ka
ADVANCED METHODS
0
47
3
WHAT IS INTELLIGENCE AND AI?
The idea
Th
id off M
Machine
hi IIntelligence
t llig
refers
f
b
back
k tto
1936
0
•
YS
47
3
Combination
C
bi i off fi
five components for
f iintelligence:
lli
Understanding, Perception, Problem Solving,
Reasoning and Learning
is
h
Real Intelligence – What comprises of thought
process of humans.
humans
Artificial Intelligence – Property of machines
gi i g it ability
giving
bilit tto mimic
i i th
the h
human thought
th ght
process.
M
an
•
Ka
ka
rF
•
•
INTELLIGENT SYSTEMS
g
y
y
Intelligent
systems
are systems
which are
capable of learning and making decisions
like in humans.
Example:
ka
rF
²
Ka
Hybrid Intelligent Systems (HIS)
ANFIS (Adaptive Neuro Fuzzy Inference System)
h
•
is
•
M
an
•
YS
47
3
0
•
ADVANTAGES OF HIS
Ad
Advantages:
t
0
Usage:
Ka
Adapt to the moving anatomy in an intelligent and automatic manner.
h
•
is
•
47
3
•
YS
•
Adapation to the input
Automatically rule extraction from numerical inputs
Can be used for modelling and prediction of Non-linear dynamics
ka
rF
•
M
an
•
ADVANCED STRATEGIES
Conventional solution for respiratory motion expand PTV
47
3
0
•
YS
is
h
Ka
Optional - Continous motion tracking and
correction
•
M
an
•
Delivering high dose to adjacent structures.
ka
rF
•
Requires advanced correction methods like ART
(Adaptive Radiation Therapy) + 4DCT.
4DCT & ADAPTIVE RT
0
Typical process:
Structure
Segmentation
(Contouring)
YS
47
3
Deformable registration
Ka
ka
rF
(Modelling & Prediction)
is
h
IS
M
an
´
DMLC/Couch
Ad t ti &
Adaptation
delivery
Adaptive
p
Replanning
AUTOMATIC SEGMENTATION
Delineation
D
li
ti off anatomical
t i l structures
t t
automatically based upon a distance
measure or similarity
YS
47
3
0
•
•
•
ka
rF
Ka
h
is
•
Euclidean Measure – K-Means Clusteringg
FCM - Fuzzy C Means Clustering
Texture Analysis
M
an
•
Drawback – heavy dependence on the
model.
h
is
M
an
YS
ka
rF
Ka
0
47
3
GTV
USAGE
Advantages:
•
•
0
heavy dependence on the model.
h
•
Ka
Drawback
is
•
47
3
•
YS
•
Removes interobsever variability
Can help in margin reduction
Helps in 4DCT to automatically propagate contour with moving anatomy
Tissue and tumor modeling for navigation and intervention.
Fuzzyy approach
pp
helps
p in removal of partial
p
volume effect
ka
rF
•
M
an
•
M
an
is
h
Ka
ka
rF
YS
47
3
0
VIDEO – AUTOMATIC SEGMENTATION
REGISTRATION
h
is
M
an
•
0
Non-Rigid Deformation:
• Elastic, Flow (Navier Stokes)
Utility: Patient Realignment and Moving anatomy modeling
Ka
•
b
bony
anatomy
t
b
based
d
47
3
•
YS
•
Registration algorithms determine the transformation that
map a point from a source image to target image
Rigid : Affine registration (6 paramters)
ka
rF
•
OPTICAL FLOW MAPS
Target
g
Rigid
g
Non Rigid
g
Ka
ka
rF
YS
47
3
0
Source
h
Rigid
is
Target
M
an
Source
Non Rigid
M
an
is
h
Ka
ka
rF
YS
47
3
0
VIDEO – TUMOR DEFORMATION MAPS
PREDICTION
h
Ka
•
System latency
Detection of irregular
g
tumor motion ((ex.
unexpected patient cough or movement)
YS
•
47
3
0
Prediction filters are needed :
ka
rF
•
•
is
Prediction from the model
Ex. ANFIS
M
an
•
LESION SEQUENCES
Q
– SEG/PRED.
/
ka
rF
YS
47
3
0
Lesion sequence from segmentation
M
an
is
h
Ka
Lesion from ANFIS Prediction
LESION SEQUENCES
is
h
Ka
ka
rF
YS
47
3
0
Segmented
g
- FCM
M
an
VIDEO –
Predicted - ANFIS
ADAPTIVE REPLANNING
Ka
Dose changing
D
h
i with
i h moving
i anatomy to b
be
replanned
Time dependent deformable field (ex
.Deformable registration) can be used.
is
h
•
ka
rF
YS
47
3
0
Treatment to be best matched with the
moving anatomy.
•
M
an
•
DYNAMIC ALIGNMENT OF BEAM/COUCH
0
47
3
YS
ka
rF
Ka
h
is
•
Couch motion
DLMC (Dynamic Multileaf Collimator) M
an
•
M
an
is
h
Ka
ka
rF
YS
47
3
0
ADAPTIVE DELIVERY - DMLC
M
an
is
h
Ka
ka
rF
YS
47
3
0
A TYPICAL ART SYSTEM
ADVANTAGES (4DCT + ART)
47
3
0
Customize the margin tailored to each patient
patient’ss
unique tumor characteristics
P
Personalized
li d C
Cancer Th
Therapy
Holds p
potential for ”Margin
g removal”
Ka
h
is
•
M
an
•
ka
rF
YS
•
IGRT COMPANIES & TPS
0
47
3
YS
ka
rF
Ka
h
•
is
•
Varian – Eclipse TPS
Elekta – CMS Software
Siemens – Artis Zeego- intervensjonssenter
M
an
•
REFERENCES
•
•
YS
ka
rF
Ka
h
•
is
•
M
an
•
47
3
0
•
Innovations in image-guided radiotherapy. D. Verellen et. al.
Nature
Reviews
Cancer
7,
949-960
N t
R i
C
7 949
960 (2007)
Anatomical Imaging for Radiotherapy, Philip M Evans. Phys.
Med. Biol. 53 (2008)
(
)
Errors and Margins in Radiotherapy, Marcel Van Herk
Seminars in Radiation Oncology. 2004.
Ma age e t off respiratory
Management
e i at
motion
ti iin radiation
adiati oncology
c l g
report of AAPM task group 76. Keall PJ, et. al., 2006
Respiratory
p
y motion p
prediction byy using
g the adaptive
p
neuro
fuzzy inference system (ANFIS). Kakar et. al. Phys Med Biol.
2005 Oct 7;50(19):4721-8
Effect of different lung densities on the accuracy of various
radiotherapy dose calculation methods: Implications for
tumor coverage. LR Aarup et al.,Radiotherapy and Oncology.
In press 2009
REFERENCES
Automatic segmentation and recognition of lungs and lesion from CT
scans of thorax. Kakar M, Olsen DR. Comp. Med. Imag. Graph;33(1);
2009.
•
4-D image based treatment planning: Target volume segmentation and
presence of respiratory
p
y motion. Rietzel E. et. al. Int. J
dose calculation in p
Rad. Oncol. Biol. Phys. (61) 1535-1550, 2005.
Hybrid intelligent modeling and prediction of texture segmented lesion
from 4DCT scans of thorax. Kakar M, Olsen D.R., IEEE Conference on
Fuzzy Systems 2009.
2009
ka
rF
Ka
•
YS
47
3
0
•
Inverse Planning for Adaptive Radiation Therapy Using Dynamic
Algorithm. De la Zareda A, et. al, Int J. Radiat Oncol Biol Phys (66): S123124 2006
124,2006
•
Determination of maximum leaf velocity and acceleration of a dynamic
p
for 4D radiotherapy.
py J Wijesooriya
j
y et. al.
multileaf collimator: Implications
Med. Phys (32), 932-941. 2005.
M
an
is
h
•
SUMMARY
0
47
3
Case Study -Respiratory
Respiratory motion Compensation
•
Methods for respiratory motion compensation
Advanced Strategies
g ((4DCT & ART))
h
•
ka
rF
•
TTumor D
Delination,
li ti
S
Setup
t E
Errors, M
Moving
i anatomy
t
Correction Strategies
YS
•
Ka
•
•
is
•
What is IGRT ?
I
Imaging
i D
Devices
i
Errors and Margins
M
an
•