Download A Finite State Model for Respiratory Motion Analysis in Image

A Finite State Model for Respiratory Motion Analysis
in Image Guided Radiation Therapy
Huanmei Wu (NU) Gregory Sharp (MGH) Betty Salzberg (NU)
David Kaeli (NU) Hiroki Shirato (Japan) Steve B Jiang (MGH)
This work is part of CenSSIS, the Center for Subsurface Sensing and Imaging Systems, under the Engineering Research Centers Program of the National Science Foundation (Award # EEC-9986821).
REOE
3. Technical Details
we have designed and implemented a finite state model that
characterizes a breathing cycle, as containing three breathing states:
exhale (EX), end-of-exhale (EOE) and inhale (IN). These three states
repeat in succession as they do in natural breathing. Any motion that
does not fall into any of the above three state will be treated as
irregular breathing, a fourth state (IRR). The state transition is guided
a finite state automaton as shown in Figure 1.
IN
EX
IN
EX
IN
EX
IN
EX
IN
EX
IN
EX
IN
AVG
- 10
0
0
1
2
3
4
EOE
5
6
7
8
9
EOE
EOE
EOE
EOE

EOE
-5
EX
IN
EX
IRR
IRR
IN
EX
IN
EX
IN
EX
IN
EX
IN
-15
Figure 2 Spatio-temporal tumor movements
-25
0
2
3
5
7
8
10
12
13
15
17
18
20
22
23
25
27
3
4. Results
This model was used to describe the respiratory motion for 23
patients with peak-to-peak motion greater than 7mm. The average
RMS error over all patients was less than 1mm and no patient has an
error worse than 1.5mm, which is shown in the following figure.
EX
EOE
IN
0.5
Amplitude changes
23
22
21
20
19
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Frequency changes
-3
-9
Raw data
-15
predicted value 100ms
-21
1
91
181
271
361
451
541
631
721
811
901
991
1081
1171
1261
Time (1/30 sec)
Figure 5 Prediction results
4. Future Goals
1.5
1
It provides an easy way for motion prediction using subsequence
similarity matching using stream database technology. Figure 5
shows the simple prediction results based on the line segments of
the finite state model. New methods to improve the prediction
accuracy are under development based on sophisticated statistical
analysis of retrieved subsequences
28
Figure 2 Modeling results using the finite state model
Regular
Figure 4 Frequency patterns
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
SI motion (mm)
- 15
23
EX
22
IN
21
EX
20
IN
19
EX
1
18
-5
2
17
0
IN
16
EOE
15
EOE
14
EOE
13
EOE
EOE
12
EOE
EX
11
EOE
10
EOE
3
7
EOE
Shorter
EOE state
Longer
EOE state
6
EOE
Our future research includes several areas.
One area is modifying the current model to
including non-linear states, and more or less
states per breathing period. We would like
to address cardiac motion in the context of
tumor respiratory motion. Another future
work is to extend the 1D implementation to
3D implementation. Finally, we are working
tumor motion prediction based on the new
model we proposed.
Strategic Research Plan
Overview of the Strategic
Research Plan
L3
EnviroS4 Civil S5
Bio-Med
S2 S3
S1
L2 Validating
TestBEDs
L1 Fundamental
Science
R1
Impacting System
Level Projects
R2
R3
Figure 6 Impact on
CenSSIS
References
Base line shifts
Noise
Combinations
Each of these states corresponds to a natural action: EX is the motion
due to lung deflation, EOE is the motion for rest after lung deflation,
and IN is the motion due to lung expansion. An example of
segmentation into these three states is illustrated in Figure 2.
CenSSIS Research and Industrial Collaboration Conference Nov. 2004
4
9
RIN
5
REX
8
IN
4
RIRR
RIRR
 It provides a convenient tool to quantify respiratory motion
characteristics, such as patterns of frequency changes and amplitude
changes. The following figure shows the frequency patterns based on
the finite state model:
RIRR
RIN
3
5
AVG
The goal of our work is to design a quantitative method for describing
breathing motion that captures all of the natural characteristics, and
can be used for both off-line analysis and on-line monitoring. Our
model can be applied to internal or external motion, including internal
tumor position, abdominal surface, diaphragm, spirometry, and other
surrogates
IRR
Figure 1 Finite state model for respiratory motion analysis
SI Motion (mm)
 Motion prediction is required by beam tracking and respiratory
gating due to system latency and imaging rate
REX
REX
 The conformality of delivered dose to thoracic and abdominal
lesions is degraded by respiratory motion
 Most motion compensation methods require an adequate
understanding of the motion characteristics
RIN
RIRR
EX
SI Motion (mm)
Respiratory motion analysis of a moving target is an integral part of
effective image guided radiation treatment.
IN
EX
The three state motion model gives convenient ways to summarize
and analyze motion characteristics, including offline data analysis and
online prediction.
2
REOE
5
2. Challenges and Significance
EOE
1
EOE
SI mot i on ( mm)
Effective image guided radiation treatment of a moving tumor
requires adequate information of respiratory motion characteristics.
For margin expansion, beam tracking, and respiratory gating, the
tumor motion must be quantified for pretreatment planning and
monitored on-line. We propose a finite state model for respiratory
motion analysis that captures our natural understanding of breathing
stages. In this model, a regular breathing cycle is represented by three
line segments, exhale, end-of-exhale, and inhale, while abnormal
breathing is represented by an irregular breathing state. In addition,
we describe an on-line implementation this model in one dimension.
We found this model can accurately characterize a wide variety of
patient breathing patterns. Our model provides a convenient tool to
quantify respiratory motion characteristics, such as patterns of
frequency changes and amplitude changes, and can be applied to
internal or external motion, including internal tumor position,
abdominal surface, diaphragm, spirometry, and other surrogates.
5. Applications
SI Motion (mm)
I. Abstract
Figure 3 Fidelity of our model to raw data
4. Bortfeld T et al 2002 “Effects of intra-fraction motion on IMRT dose delivery:
statistical analysis and simulation”, Phys. Med. Biol. 47(13) 2203-20
5. Jiang S B et al 2003 “An experimental investigation on intra-fractional organ motion
effects in lung IMRT treatments”, Phys. Med. Biol. 48(12) 1773-84
6. Shirato H et al 2000 “Four-dimensional treatment planning and fluoroscopic real-time
tumor tracking radiotherapy”, Int. J. radiat. Oncol. Biol. Phys. 48 1187-95
7. Huanmei Wu et al “Subsequence similarity matching for tumor respiratory motion
analysis”, submitted
8. Huanmei Wu et al “A finite state model for respiratory motion analysis in image
guided radiation therapy”, submitted to Phys. Med. Biol.
Contact: Huanmei Wu, Egan225, Northeastern University, (617) 373 7349, [email protected]