$% &
'
"'
'
!
$)
#
!
"
#(!!
* %
# +,
#
#$
$ '&
-
!
.
1
.
/0 -
.
/
/0
.
.
( & $1
0
'
.
2$
2
-
3
- #
- "
#0
#
2$
/
+
5
&
67
1
8 4
#
%
6
&9*
&
+
5
#
2$
1 :7
;
1
%4
1
1
%
$
& 04
0
%1
1
4
1
.
#
$
- &
< 0/
"
!
8
#
4
2$3$<
&
2$
&
&
4
&
5
1
1
#
67
=
% /
'>
4
#
2$
2$3!
#
6
EFF6
8
? @A +
BACDBCC
C
:G
HAFF
1
IIII:7 !'
J#
&
K
& 1
'
:
4
%&
#
& %
4
$
3
!
-
! .
,
#
!
, '
. 4
'
-
8 .
#
$
+
8 . 4
.
4
-
+
#
#
#
.
3
8
!
9
#
3
.
!
0
!
=
2$
0
%
%
&
-
&
&
.=
$
9
$
0
2$
0
&
0
1 %
2
-
!
0
#
.
'!
8
1
.
0
'
!'
0
'
!
*&
1
#
1
0
1
#
1
'
!
%
1
#
'
!
!
#
.5
5 L!
1
E3
H6B EFFH
M !
L?
EFFE
1
$
!
EI3
:HE
1
# '!
#
Film measurements of a 10-strip test pattern. The linac (a Varian 2100 C/D)
was instructed to deliver 1 MU per strip with the step-and-shoot delivery
mode for a total of 10 MU. The delivery sequence is from left to right.
-
+
$%
O$
'!
# & %1
'
0
N'
Film measurements of a 10-strip test pattern. The linac (a Varian 2100 C/D)
was instructed to deliver 2 MU per strip with the step-and-shoot delivery mode
for a total of 20 MU. The delivery sequence is from left to right.
%
'5 1
$ !
P
EFFEQA6KCE
.5
5 L!
1
$ !
P
EFFHQH6BKH:B
?
$
N'
6
E
O
0
0
3
'
?
1
Two log files are generated by MLC controller
for each field delivery
Each log file records cumulative dose fraction,
beam status, and leaf position data for all
leaves in the bank.
Segment MU Log file data has been validated
by our group at the University Of Florida
'
'
$
&
# +
%
O
N
?
&
%
CIIKBF@P
EFF6Q
'
*
! "
#
$%
&
&
'
%
#
$%
&
#( # ) &
(
'
(
(
*
(
+
&
%
'
(
# &
-
Methodology:
4
'
&
Varian 2100 C/D accelerator
Millenium 120 MLC
91 IMRT patient treatment plans delivered
Plans delivered at dose rates of
3
1
=
100, 300, & 600 MU/min.
For details see: Stell et al.,
H@I6KHAFE P
EFF:Q
1
#
4
$
!
3
! ? # :F
1
$
EF7
HF
$ '$8$
A@
%
HA
,'
,.
,/
%
%
%
%!
%
1 &
!
@KC1
1 &
%-
6FF
%
IMRT treatment plans statistics
1
9
#1
91 clinical IMRT cases
Represents 635 field deliveries for a single
dose rate
34,212 subfield deliveries including all three
dose rates (11,404
(11,404 at each dose rate)
#1
Subfield MU Error (MU)
Subfield MU Error (MU)
.
!
.
Subfield MU Error (MU)
5
1
AFF
#
1
N
9
#
1
&
1 .5
5
1
O
&
+
1
1 &
1
1
HF
#1
.
!
'
#
#=
0
'
#
#=
'
!
300 MU/min.
600 MU/min.
!
Data for recalculated plans…
The recalculation process…
Table I. Relative and absolute plan delivery errors for each of
the plans selected for ADAC recalculation.
relative error (%)
Plan fields subfields
total absolute error
(MU)
300
MU/min.
600
MU/min.
300
MU/min.
600
MU/min.
3.78
8.27
41.31
90.5
1
14
271
2
12
264
3.4
7.01
42.07
86.67
3
12
311
3.63
8.63
60.99
145.16
Obtain actual monitor unit distribution from loglog-file
Generate Varian shaper (.d60) file that combines
monitor unit information with planned leaf positions
(one file for each field)
Combine shaper files and convert into an ADAC
format (.Trial file)
Import .Trial files into ADAC and recalculate dose
distribution
Export RTOG files and compare in Matlab
6
?*
E
Comparison of absolute dose
distributions for Plan 1
original plan
600 MU/min.
E
6
'
#
#
1
1
N+
!
5
O
1
1
<
1
+
67
=
&
7
'
$
'
'
?
#
&
'
0
10
1
HF
'
'
&9'
#
'
1
J@
-
&
L
&
'
=
&
'
.
9
1
AFF
5
9
/
8
#
1
!
.
8
.
&
2$
#
'
!K
&
P #Q
# &
<
!! !
#
"
P
# HFFQ
Camera System Characterization
How do we know that the imaging device is capable of
measuring the fluence distributions that we seek?
SignalSignal-toto-Noise
Spatial Resolution
Fluence Uniformity and Linearity
Linac Pulse Rate Dependence
We used a 66-MV photon beam delivered from Varian 21000C\
21000C\D
at the nominal dose rate of 600MU/min for all measurements in this
this work
Imaging System Setup
Spatial Resolution
Varian 2100C\
2100C\D +
Millennium 120120-leaf
MLC
Fluoroscopic Material:
Gd2O2S:Tb (0.411 g/cm2) glued to
a stainless steel plate (1.24 g/cm2)
Sony CCD-TRV16 camcorder:
30 FPS
280 × 320 pixels CCD chip
Radial & Transverse profiles of square fields were measured and camera field size determined by the 50% penumbra.
Measure slopes
What can we learn from this simple imaging system?
Fluence Uniformity and Linearity
Image intensity information should be directly proportional to incident
incident beam fluence.
Uniformity
Spatial resolution of ~ 0.9 mm at field of
view focus of 25 × 30 cm2
Pulse Rate Dependence
Two square fields of 10 × 10 cm2 were delivered at two dose rates and at 5 MU/field. The data
were digitized and intensity information/frame were extracted.
Linearity with Field Size
The frameframe-toto-frame intensity variation is attributed to a combination of an aliasing
aliasing effect and variation
in the output intensity of the linac itself.
Total area under both profiles agrees to within 1%
Slope values differences within 2%
9
Linac/Camera Pulse Sequences
At this dose rate, the gun signal
rate matches Klystron signal
rate at 360 pps
The camcorder is not an ideal imaging system. It has several
hardware limitations, but it produced very interesting results …
A schematic of the linac gun signal pulse sequence at two dose rates
rates with respect to the camera
frame capture sequence. The diagram depicts the ideal case where the two sequences are not aliased.
The SMLC Delivery
(Raw Movie Data)
Vendor A SMLC IMRT Delivery
?
Movie Analysis Procedure
CORVUS 4.0
Leaf Sequence File
(shaper)
The SMLC Delivery
(Raw Movie Data)
8# '
!
The Movie: again in detail
Raw Image Data on
8 mm video tape
MLC
Workstation
Video Capture
NTSC => AVI
~ 55 ms cycle reports
DynaLog Files
Frame Grapping Software
AVI => BMP images
~ 0.6 GByte of uncompressed
image data (mostly useless)
Matlab 6.5
10
Feedback Treatment Planning Based on
the Movie
Camera Based Reconstruction
Imaged MLC delivery errors are accounted for in a recalculated treatment plan
Dose volume histograms of the targets and critical structures for
for the reconstructed plans
based on the leaf sequence files from the treatment planning system
system (solid lines) and
those from the MLC log files (dotted line) . The reconstruction was done in using the
Pinnacle treatment planning system with the same patient geometry
geometry and dose grid.
Sudden OutOut-ofof-Tolerance Leaf Movement
"E
Comparison of the dose distributions in an axial plane between those
those from the TPS (Plan Dose)
and those reconstructed based on the information from the MLC log
log files at 600 MU/min (MLC log
dose) and their percentage dose difference.
SMLC Delivery
38#
E#
!
6F
HFFF
3
!! !
!
#!
The FMU for a given segment was extracted from the ratio of
that segment fluence to the total fluence of all segments.
1
'
#'
!
The Discrepancy Pattern
Fluence Map Comparison
?
'!
-
An oscillatory behavior about zero is observed, consistent with a predictorpredictor-corrector control
algorithm attempting to repair the initial overshoot by repeated over compensation. Results
consistent with previous studies*. The segment MUs are redistributed.
redistributed.
* P. Xia et al., Med. Phys. 29, 412412-423 (2002)
* G. A. Ezzell et al., J. Appl.
Appl. Clin.
Clin. Med. Phys. 2, 412412-423 (2002)
11
What Did We Learn so far?
A High Speed Digital Imaging System
Features:
The camera system agrees with log file to within 5%. The
validated log files can be used as a QA tool for IMRT plan
verification.
CMOS camera
Fire Wire I/O rates of 400 MBits/s
Up to 512 × 512 pixels
10 µs Shutter Speed
Up to 1000 FPS
Found three types of delivery errors:
1.
2.
3.
FMU delivery errors (FMU redistribution)
Leaf positioning errors: leaf moving while beam is on
Segment dropping (skipping)(skipping)- only for very small MU segments
Camera hardware limitations prevent us from pursuing further offline
offline
verification of MLC deliveries.
We need a faster imaging system ( hundreds of FPS) with more controls:
controls:
external strobing,
strobing, frame rate, shutter speed, …etc
6 GB onboard memory
~$25K
Redlake
MotionPro 500
!
6AF #
1 1
'
.
Linac/Camera Pulse Sequences
*
#
Linac/Camera Pulse Sequences
!
*
1
Very Fast MLC Motions: 72 FPS from
360 FPS Data
*
#
!
#
!
$
'
*#!
*#!
*#!
&
'!
&
12
-
$1
#
"
'!
1
Log file
<
>
@
&
J6
%
1 &
'
#(!!
Sup
Post
Ant
2 cm
Inf
jaws
Overlay
Overlay
Dose differences (log-dva)
110 MU
dose distribution
'!
'!
#
Leaf sequence file
110 MU
Dose differences (log-dva)
Timing / communication errors
55 MU
2X
Is there an analogous problem for DMLC?
!"# $$%& '(% )* ++,& !+
+
- !+
14 MU
8X
28 MU
4X
Yes, if
• leaf sequencers actively utilize
the beam hold-off to control leaf speed
• if leaves can not maintain their rated speed
2$L
Motor replacement history
MLC installed
1st MLC Tx
-
1st dMLC Tx
90
Room
445
245
442
444
241
441
80
70
60
#
50
40
30
20
Mark 1
Mark 2
Mark 2
Millennium
Millennium
Millennium
#
#
10
0
Ja
2
n-9
n-9
Ja
3
n-9
Ja
4
n-9
Ja
5
n-9
Ja
6
n
Ja
-97
n
Ja
-98
Ja
n-9
9
Ja
n-0
0
Ja
n-0
1
Ja
n-0
2
Ja
n-0
3
Ja
n-0
4
Ja
5
n-0
.
1
'
2$0
#
#
2$.
0
!
.
'
!K8
#'
!
'
!
Date
13
'
!
!
8
5
#
3. E
CAX Cross Profiles: 5.0 cm depth, shift = 1.0 cm.
Transmission dose is subtracted.
•Radiographic meas. of accuracy of leaf travel/ offset
And Integral Fluence Maps
•Absolute vs. Relative
LS
LE
'#
EFFHR
EBP
HHQ
3
EEFIKEEHI
8
#
'!2$ # '!
.
3. E '
.
How is the superposition changing with distance between LE profiles?
0.5 mm
1.5 mm
1.0 mm
2.0 mm
S
HFK
S
M
9
&
M
1
S$
&
'
!2$
.
S
&
'
!
K %
S
M
&
<
S$
M
%
%
Suggested Confidence Limit and Action
Level Values for IMRT Planning
Region
Confidence Limit*
(P=0.05)
Action Level
δ1 (high dose, small dose
gradient)
3%
5%
δ1 (high dose, large dose
gradient)
10% or 2 mm DTA
15% or 3 mm DTA
δ1 (low dose, small dose
gradient)
4%
7%
δ90-50% (dose fall off)
2 mm DTA
3 mm DTA
&
'&
P
EFFHQC@EKC@A
* Mean deviation used in the calculation of confidence limit for all regions is expressed
As a percentage of the prescribed dose according to the formula,
δi = 100% X (Dcalc – Dmeas./D prescribed)
14
Segmental Multileaf Collimator
(SMLC) Delivery System
Tolerance
Limit
Action
Level
1 mm
0.2 mm
0.2 mm
2 mm
0.5 mm
0.5 mm
0.75 mm
radius
1.00 mm
radius
2%
2%
3%
3%
MLC*
Leaf position accuracy
Leaf position reproducibility
Gap width reproducibility
Gantry, MLC, and Table
Isocenter
Beam Stability
Low MU Output (<2MU)
Low MU Symmetry (<2MU)
Dynamic Multileaf Collimator
(DMLC) Delivery System
Leaf position accuracy
Leaf position reproducibility
Gap width reproducibility
Leaf speed
Low MU Output (<2MU)
Low MU Symmetry (<2MU)
'!1
=
1
K
1
+ * FH!$HFFA6A$&
!
1
1
'
!2$
$% &
+
3%
2%
1.00 mm
radius
5%
3%
1
#
#
1
1
#
1 mm
0.5 mm
0.5 mm
±0.2 mm/s
#
=
+
0.5 mm
0.2 mm
0.2 mm
±0.1 mm/s
0.75 mm
radius
Gantry, MLC, and Table
Isocenter
Beam Stability
#
<
Action
Level
MLC*
* Measured at all four cardinal gantry angles
.
1
Tolerance
Limit
#'
!
'!
=
%
,
$
1
#
?
%
# +
15