Linear accelerator with onboard conebeam CT
Lei Xing , Ph.D.,
Ph D Professor
Department of Radiation Oncology
Stanford University School of Medicine
movies
Use CBCT for Dose Verification
Scatter Artifacts
Planned (IMRT)
• Large Scatter-to-Primary Ratios (SPR) in CBCT cause
severe cupping/shading artifacts.
DVHs (planned vs delivered)
Prostate
PTV
Wide collimator, high scatter
narrow collimator, low scatter
To be delivered (reconstructed dose on CBCT)
Prostate
PTV
Yang Y, Schreibmann E, Li T, Wang C, Xing L:
Evaluation of on-board kV cone beam CT
(CBCT) based dose calculation. Physics in
Medical Biology, 52: 685-705, 2007.
Display window: [min max]; no anti-scatter grid, no scatter correction.
Zhu L, Wang J, Xing L, Scatter correction for cone beam CT in radiation therapy,
Medical Physics 36(6):, 2258-68, 2009.
Work Flow
“Registered” scatter estimate
CBCT Projection
Subtract
Reconstruction
Rigid
registration
Reconstruction
Partial CBCT projection
Reconstruction
Scatter corrected CT
image
Scatter estimation
using interpolation
Scatter estimate
L. Zhu, J. Wang and L. Xing, Med. Phys. 2008
@01-258_via99-011_hepatoma_onscreen
1
Scatter noise in post-processing methods
No scatter correction, no noise Measurement-based scatter
suppression,
correction, no noise
suppression,
Noise in the ROI: 1.01e-6
Noise in the ROI: 1.01e-5
Motion artifacts in fan beam CT and CBCT
Measurement-based
scatter correction, PWLS
noise suppression, (Wang
et al., 2006)
Noise in the ROI: 9.75e-7
L. Zhu, J. Wang and L. Xing, Med. Phys. 2008
CBCT using Trilogy
Scatter
Cone-Beam CT
z
Cone Beam
Tianfang Li et al, Li T, Xing L: IJROBP, 67: 1211-1219, 2007.
Stanford Research
CBCT projections before and after phase sorting
CBCT phantom images
Static phantom - 3D CBCT
motion “switched on” - 3D CBCT
motion “switched on” - 4D CBCT
Li, Koong, Loo, Xing, Med. Phys., 2006
@01-258_via99-011_hepatoma_onscreen
2
0.055
10 mA
80 mA
10 mA PWLS
0.05
Ultra-low dose CBCT
0.045
0.04
0.035
4D CBCT
4D CT
0.03
0.025
0.02
0.015
0.01
210
220
230
240
(a)
250
(b)
10mA
80mA
10mA
Li, Koong, Loo, Xing, Med. Phys.
J. Wang & L Xing, PMB, 2008
T 1
2

R (( p))  
( p)(yˆ w(inpyˆˆ()pipˆp()nyˆT)pˆ)1(yˆR( p)pˆ )  R( p)
n
Noise property of projection images
PWLS (Penalized Weighted Least-Squares method):
 ( p )  ( yˆ  pˆ )T  1 ( yˆ  pˆ )  R( p )
16000
Linear fit of measured data
14000
Variance = 2.7*Mean + 180.7
R ( p )   win ( pi  pn )
Variance
12000
2
n
R = 0.996
10000
8000
6000
4000
win  exp[ (
pi  pn

2000
2
0
) ]
0
1000
2000
3000
4000
5000
6000
Mean
x 10
R ( p )   win ( pi  pn ) 2
n
100
5
iterative Gauss-Seidel updating strategy
200
R( p)   win ( pi  pn ) 2
4
300
n
400
3
500
2
600
1
700
pi( k 1) 
yi   i2 (  win pn( k 1) 
nN i1
1  
2
i
w
nN i
w
in
pn( k ) )
nN i2
in
400 across
600the field
800of view
1000
Incident X-ray200
intensities
with 80 mA
tube current and 10 ms pulse time. Relative intensity is
mainly caused by the bow-tie filter.
@01-258_via99-011_hepatoma_onscreen
3
0.055
10 mA
80 mA
10 mA PWLS
0.05
Ultra-low dose CBCT
0.045
Compressed sensing for CBCT recon with sparse projections
0.04
0.035
0.03
0.025
0.02
0.015
0.01
210
220
230
240
(a)
250
(b)
10mA
80mA
10mA
K. Choi, L. Zhu, T. Suh, S. Boyd, L. Xing, Med. Phys., in press, 2010
J. Wang & L Xing, PMB, 2008
Metal artifacts removal
Ultra-low dose fluoroscopic imaging
(a)
kV imager
kV source
y
x
z
u
v
EPID
J. Wang & L. Xing, X-ray Science & Technology, 2010
MLC log-file generated Fluence Map
MLC log-file
MLC Workstation
Dose Reconstruction: Closing the
Loop of IMRT/RapidArc/Gated
RapidArc treatment
•
•
every 50 ms
leaf position & beam status
TPS
Delivered dose
distribution
Delivered
fluence map
in-house
program
Actual leaf sequences
Lee L, Le Q, Xing L: IJROBP, 70: 634-644, 2008.
@01-258_via99-011_hepatoma_onscreen
4
Case 1: Dose Distribution
pCT
CBCT1
CBCT2
CBCT3
Figure 2.10 (a)
Lee L, Le Q, Xing L: IJROBP, 70: 634-644, 2008.
Case 1: DVHs
Case 2: Dose Distribution
pCT
DVH comparison of the intended and delivered plans
Relative volum
me (%)
100
80
pCT
PT
V
60
CBCT1
CBCT2
Brainstem
40
CBCT1
CBCT2
CBCT3
CBCT3
20
0
0
40
80
120
160
200
240
Dose (cGy)
Case 2: DVHs
Case 2: DVHs
Relative volume (%)
DVH comparison of the intended and delivered plans
100
DVH comparison of the intended and delivered plans
80
pCT
CBCT1
CBCT2
CBCT3
RT Temporal lobe
60
100
40
Relative volume (%)
20
80
PT
V
0
pCT
60
0
50
150
200
250
CBCT2
Brainstem
40
100
Dose (cGy)
CBCT1
Relative volume (%)
CBCT3
100
20
80
pCT
CBCT1
CBCT2
CBCT3
LT Temporal lobe
60
0
0
40
80
120
160
200
240
40
20
Dose (cGy)
0
0
50
100
150
200
250
Dose (cGy)
@01-258_via99-011_hepatoma_onscreen
5

MLC Workstation
Case 2: Dosimetric comparison
• every 50 ms
• leaf positions &
gantry angle logged
in-house
program
220 cGy at 100%

Treatment Planning System
DCAT
Dose Reconstruction
Regenerated Leaf Sequence File
Dose Distribution Comparison
Planned and Reconstructed Dose Profile Comparison
A
Qian J, Lee L, Liu W, Fu K, Luxton, G, Le Q, and Xing L, PMB
57, 3597–3610, 2010.
R
Reconstructio
n
PTV DVH
Comparison
L
Plan
reconstruction
plan
A-P profile
R-L profile
P
Positioning Errors and Dose Delivered to PTV
Patient Study
Positioning errors intentionally introduced
100% =180 cGy
100%
50%
PTV
PTV
PTV
PTV
10%
reconstruction
plan
pCT
CBCT1
•
Position #1: same as the plan
CBCT2
CBCT1 / CBCT2: monitored the anatomic change, if any
•
CBCTs’ dose distribution very close to pCT’s
Position #2: L R: 0 mm; A P: -2 mm; S I: 2 mm
Position #3: L R: 3 mm; A P: 5mm; S I: 5 mm
@01-258_via99-011_hepatoma_onscreen
6
Use CBCT for Dose Verification
DVH Results
Planned (IMRT)
DVHs (planned vs delivered)
,
Relativ
ve Volume (%)
PTV
54 Gy
Prostate
PTV
dose reconstructed
on CBCT1, CBCT2
Optic
chiasm
planned dose on
pCT
Delivered (reconstructed dose on CBCT)
Bilateral
optic
nerves
Relative Dose
(%)
Prostate
PTV
• slight compromise (< 5%) on the target
coverage
• dose deposited to the critical organs:
in general <10% change;
worst ~20%
Re-optimization
Adaptive Radiation Therapy
Yes

Treatment delivery
No
Reconstruction of
delivered dose
New treatment session
Is replanning
needed?
Forward dose
calculation and
assessment
Volumetric imaging,
Image registration
& auto-contour
propagation
What are needed to bring ART into clinic?
• CBCT.
• Deformable model.
• Automated contour mapping from pCT to
CBCT.
• Retrospective dose reconstruction.
• Deformable registration for cumulative dose
calculation
• Inverse planning for ART
• Dose shaping tool.
IMMOBILIZATION DOES NOT ALWAYS WORK!
CBCT imaging of a rectal cancer patient during a course of RT
1st wk (planning CT)
4 wk
4D Treatment Planning
2 wk
overlay
Static (with
4D CT info
- 3.5D RT)
Gating
Tracking
Adapted from Y. Yang, UPMC
P. Lee, K. Goodman, L. Xing, et al, 2006 ASTRO
@01-258_via99-011_hepatoma_onscreen
7
Radiation Therapy Chain Process
Infrared
Infrared reflective
markers
Real-time information of tumor position
Simultaneous kV/MV imaging guided RT delivery
(R. Wiersma, W. Mao, &L. Xing, Med. Phys., 2008)
xkV
z

d kV  s  d kV  d d kV  s  x
y kV
y

d kV  s  d kV d d kV  s  x
x MV
x

d MV  s  d MV d d MV  s  z
y MV
y

d MV  s  d MV d d MV  s  z
 cos
 0

 sin 
0 sin   x   x 
   
1
0  y    y  
0 cos  z   z  
@01-258_via99-011_hepatoma_onscreen
8
Results – example 1
Example 1
Real-time Image Guidance for Prostate VMAT/IMRT
• The sudden drop represents repositioning.
ACKNOWLEDGEMENT
L. Zhu, T. Li, J. Qian, R. Wiersma, W. Liu, J.
Wang, K. Choi, L. Lee, B. Meng, X. Zhang, Y.
Yang, A. de la Zerda, B. Armbrush,…
Clinical faculty –
A. Koong, Q. Le, B. Loo, G Luxton, C. King, S.
Hancock P.
Hancock,
P Maxim,
Maxim E.
E Mok,
Mok L.
L Wang …
Research supports fromNational Cancer Institute
National Science Foundation
Varian Medical Systems
@01-258_via99-011_hepatoma_onscreen
9