3. IMAGE GUIDANCE
3D image based setup verification and
correction
Knowledge on 3D patient anatomy for
every treatment fraction.
Adaptive corrections of inter fractional
uncertainties of the treatment.
6. KV CBCT
At present, an important tool for setup
verification and correction
It uses Cone beam X-ray source, which
acquires an entire volume ( 14 – 26cm in
length )
Though not of diagnostic quality, generally
adequate for imaging bone and soft tissues
7. FAN BEAM MV CT
Equipped with helical tomotherapy delivery
system
Nominal energy is reduced to 3.5 MV Beam,
and it is collimated to 4mm at isocenter
Eliminates artifacts, but inherently causes
poor subject contrast
9. MV CBCT
It consists of a Si flat panel adapted for MV imaging
attached to linac
It uses Cone Beam X-ray
CBCT image reconstruction and remote controlled
couch movement
Nominal scan dose is comparatively more ( 3 to 10
cGy)
11. XVI Basics
XVI is an electronic imaging device (EID) consists of a kV X-ray
source and an amorphous silicon (a-Si) radiation image
detector panel.
The kV source generates X-rays that are projected as planar
images on to the plate of the kV detector.
Acquired planer images are reconstructed into volume images
using reconstruction algorithm.
12.
13. X Ray Generator Tube
It has two focal spots
Small 0.4mm
Large 0.8mm
Maximum current in small is 80 mAs and in large is 100 to 500
mAs.
Anode rotation is 16000rpm and air cooling system is used
Total filtration 3.6mm which includes Inherent filtration of
0.9mm Al, a cone shaped Al 0.6mm, Al disc of 2.0mm and
0.1mm Cu.
14. Imaging Panel
It has a matrix of 1024 X 1024 pixels.
The size of the image receptor 409.6mm ×
409.6mm.
Radiation on each pixel is represented by a
number which is known as pixel value.
600 – 650 frames can be taken in volume view.
15. Gantry Arc minimum is 187.5° and maximum is 360°.
Gantry Speed = 3.18°/sec.
Frame rate = 5.75 frames/sec.
16. Image Acquisition
Three different modes available
PlanerViewTM - Radiography
MotionViewTM - Fluoroscopic view
VolumeViewTM - CT like X ray
17. Image Process
Raw Images
Offset and Multi level gain correction
Bad Pixel Correction
Downsize
Invert Image
Rotate 90˚
DB
Multi level Gain
Calibration
Auto Offset
Bad Pixel Map
ONLINE
Images
Reconstruction Online / Offline
1024 / 512
1024
1024
1024
Flex Map
18. Gain is the mean ratio of signal output to the signal input of a
system.
Here the output is the pixel value and input is the incident beam.
Two types
Single Level gain – Response of detectors for a single dose value.
Multi level Gain – To eliminate the nonlinearity between dose and
detector response.
Multi level Gain Calibration
19. It is an image that has a record of all pixels on the panel
that do not respond correctly to a dose of radiation.
This map is then applied to an acquired image to remove
the bad pixels which are replaced with average data from
the surrounding 8 pixels.
Bad Pixel Map
20. The mechanical system of the XVI move under gravity as
the gantry rotates.
This has the effect of causing the radiation field centre to
move with respect to the kV detector.
A map is created of the radiation centre against gantry
rotation and stored as a flex map.
The flex map is used to make corrections to the images as
the data is back projected to create the 3D reconstruction.
Flex Map
22. XVI does a reconstruction process based on Feldkemp-Davis-Kress (FDK)
algorithm.
It is a kind of filtered back projection (FBP) algorithm widely in use.
Reconstruction is exact for any object which is invariant for translational
in the axial direction orthogonal to plane of orbit.
Reconstructed CBCT projections are not truncated.
It is popular for its simplicity and ease of implementation and its good
robustness to CB artefacts.
Reconstruction
23. Data Filtering in FBP
It is to smooth out the statistical noise
Some typical filters used are the Hanning filter,
Butterworth filter, low pass cosine filter, Weiner
filter, etc.
XVI reconstruction is based on Feldkemp variant
and weiner filter.
x
y
27. It checks accuracy of KV and MV isocenter
coincidence
It also evaluates the accuracy of remote controlled
couch with a known shifts
Image registration accuracy can be determined
using Penta guide phantom
29. Uniformity
To Verify the change in the average signal (pixel value) in
different portions of the homogeneous phantom(CTP 600)
image.
Tolerance ≤ 1.5%.
30. Low Contrast Visibility
To verify the visibility of low dense materials in XVI image.
Using the mean value and standard deviation of the pixel
values in both polystyrene and LDPE low contrast visibility is
computed.
Tolerance ≤ 1.5%.
31. Spatial Resolution
To find the spatial resolution on patterns of high contrast
resolution.
Tolerance minimum 10 lp/cm.
32. Image Registration Accuracy
To check the accuracy of kV images to
the MV isocentre and make sure that
it is within specification for
registration.
This test done using Ball Bearing
phantom made with long plastic tube
and a steel ball of diameter 8mm at
the tip.
Image registration accuracy is ± 1mm.
33. Image Registration
It is the movement of one reference point in a 3D coordinates
system.
Three types of image registration method available in XVI.
Grey Value
Bone
Manual
Manual: The translational and rotational movement is done
manually until a good match between CBCT and CT is
achieved.
35. It is also based on the cross correlation method.
The algorithm matches the edges or outlines of the bone
(high value pixel) by minimizing a generalized distance
between them.
It uses fast simulated annealing.
Reduces computational load
Reasonable, insensitive to noise
36. Grey Value Registration
It refers to pixel value
It is the cross correlation of the pixel value on the whole
volume (clip box) of the both study sets.
The algorithm used is ‘Grey level correlation ratio’ technique.
37.
38.
39. Here the cost function is the summation of RMS difference of
the pixel value of the reference image and localized image.
Here the variables are
40. QA for the X-Ray generator
X-Ray output consistency
KVp accuracy
mA accuracy
KVp linearity
mA linearity
41. References
AAPM TG 142
AAPM TG 179
AAPM TG 101
AAPM TG 66
Practical cone beam algorithm, Fred kemp et al
New grey scale template image matching algorithm, Ryo Takei
et al
Hierarchical Chamfer matching , parametric edge matching,
Gunilla Borgefors
Filtered Back Projection algorithm