1. IGRT:
MV
CBCT
Calibra0on
and
ATP
on
Siemens
Oncor
Linac
By
Vibha
Chaswal,
Ph.D.

2. Ac0vi0es*
• Flat
panel
2D
gain
calibra:on
and
dead
pixel
map
• MVCBCT
calibra:on
using
clinically
used
and
custom
CBCT
protocols
• MVCBCT
image
quality
assessment
using
clinically
used
and
custom
CBCT
protocols
*Performed at UIHC RadOnc as Medical Physics Resident

3. Flat
panel
2D
gain
calibra0on
• Correc:on
for
the
differences
in
flat
panel
diodes
response
in
2D
imaging
• Without
correc:on,
an
obvious
banding
paHern
would
be
visible
on
the
image.
• Should
be
done
every
4-­‐6
weeks.
• Coherence
prac:ce
database,
SERVICE
PATIENT
is
used
to
acquire
port
during
gain
fields
at
different
photon
energies,
dose
rates,
clinically
used
SIDs,
monitor
units
and
field
sizes.
• For
each
SID
a
treatment
site
containing
4
gain
fields
is
assigned,
and
each
site
can
be
delivered
in
a
auto-­‐sequence
group.

4. Flat
panel
2D
dead
pixel
map
• Correc:on
for
non-­‐responding
‘dead’
pixels.
• The
grayscale
values
in
the
pixel
surrounding
the
dead
pixel
are
averaged
and
this
value
replaces
the
grayscale
of
the
dead
pixel
• During
ATP
provided
by
the
manufacturer
Dead pixel map limits (table from Siemens ATP)

5. MVCBCT
calibra0on
Using Geometry Calibration Phantom

6. MVCBCT
calibra0on
•
•
Since the Linac rotation suffers with gantry sag, imager’s sag etc, the 3D
projection matrix deviates from a projection matrix model calculated from
transformations between the world and gantry co-ordinates system
The calibration matrices are therefore, obtained from the projection images
of the geometry calibration phantom for each Linac
Geometry calibration co-ordinate system (source: Siemen’s ATP)

7. MVCBCT
calibra0on
•
•
•
Geometry calibration is done every six months or whenever required
A projection imaging dataset of the phantom is acquired using the clinical
CBCT protocol.
On Siemens Oncor machines, a 2000 arc-rotation starting from 2700 to an
end-angle of 1100 is used for MVCBCT acquisition

8. Acquiring
Projec0on
matrices
Posi:on
phantom
using
room
lasers
Fine-­‐tune
posi:oning
using
x-­‐re:c
Spend
sa:sfactory
amount
of
:me
Calibra:on
fails
oWen
due
to
poor
alignment
Take
a
cone-­‐beam
acquisi:on
using
clinical
CBCT
protocol
• AWer
acquisi:on,
each
phantom
projec:on
image
is
processed
to
determine
the
ball-­‐bearings’
posi:ons
and
sizes
rela0ve
to
the
imager’s
co-­‐
ordinate
system
•
•
•
•
•

9. Post
acquisi0on
screens
Arrangement of projection images after calibration projection matrix is fitted.
status message for
successful or failed
calibration is displayed.

10. Failed
Calibra0on:
many
reasons
• Phantom
misalignment
• Incorrect
phantom
orienta:on
(gantry
side
opposite)
• Object
in
image
(e.g.,
level
leW
on
the
phantom
base)
• Incorrect
cone-­‐beam
protocol.
………..
A
very
set-­‐up-­‐sensi:ve
procedure!

11. MV
CBCT
image
quality
MVCBCT Image Quality phantom and sections for image quality tests in IMA
phantom.
•
•
•
•
•
•
Geometry accuracy
Uniformity
Noise
Spatial resolution
Low contrast resolution, and
High contrast resolution

12. MVCBCT: Geometric Accuracy
•
•
•
•
•
Tests the geometric accuracy of the
MVCBCT reconstruction algorithm
Phantom alignment very critical
Check alignment usig x-retic all along
the white engraved axes lines on the
phantom.
Axial, sagittal and coronal views of the
Adaptive Targeting (AT) task-card are
used to locate the beads of interest.
The beads’ x, y and z positions should
be within ± 2 mm of the actual physical
co-ordinates.

13. MVCBCT: Geometric Accuracy
expected ranges for
localization of beads
Localized co-ordinates
from a 270-110 cbct
protocol

14. MVCBCT: Geometric
Accuracy (additional
clinical protocols in
use)

15. MVCBCT: Image quality: Low contrast resolution section 1
(Clinical relevance of the visualization between this range is visualization of bone, air-cavities, and organs)
Passing criteria

16. MVCBCT: Image quality: Low contrast resolution section 2
(Clinical relevance of the visualization between this range is visualization of soft-tissue)
Passing criteria

17. MVCBCT: Image Quality:
Spatial Resolution
Determine smallest visible bar group
Criteria for passing: group 6 (0.3 lp/mm)

18. MVCBCT: Image Uniformity, Noise and Artifact
Performed using
uniform solid water
insert
standard deviation
and across all ROIs
uniformity w.r.t the
central insert

19. Thank
you!!!!!