Section 2 M Vision Geometry Calibration V Mc 062707 V Rjo062807
-
Upload
richard-okeeffe -
Category
Technology
-
view
948 -
download
4
description
Transcript of Section 2 M Vision Geometry Calibration V Mc 062707 V Rjo062807
Page 1 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Section 2MVision Geometry CalibrationSiemens Medical Solutions, Inc.Oncology Care Systems Group4040 Nelson AvenueConcord, CA 95420
The reproduction, transmission or use of this document or its contents is not permitted without express written authority. Offenders will be liable for damages. All rights, including rights created by patent grant or registration of a utility modal or design, are reserved.
MVision Physicist Self-Led Training
81 40 258 Rev. C
Page 2 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Section 2: Table of Contents
Objectives Scale Factor
Overview MVision Image Reconstruction
Geometry Calibration Phantom MVision Protocol
Geometry Calibration Phantom Positioning
Projection Matrices
MVision Gain Videos:
MVision Gain Field Creation Geometry Calibration field creation
Geometry Calibration Phantom positioning
Page 3 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Section 2: Table of Contents
MVision Control Console Display Labs:
Geometry Calibration Window MVision Gain
Geometry Calibration Results MVision Protocol
Video: Geometry Calibration MVision Geometry Calibration
Fail to Pass Geometry Calibration
Videos:
Section 2 Review
Geometry Calibration Phantom
Setup (p. 40)
Section 2 Quiz
MVision Geometry Calibration (p.
50)
Page 4 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Objectives
At the completion of this section, you will be able to:
Identify the fundamentals of MVision Geometry calibration
Understand the function of the Geometry Calibration Phantom
Describe the MVision Reconstruction Process
Describe the MVision Protocol parameters and how they influence the image quality
Perform the MVision gain calibration
Perform the MVision Geometry Calibration
Page 5 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Overview
Figure 2.0 Geometry Calibration phantom
Page 6 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Overview
Figure 2.1 Projection image of the phantom
Page 7 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Phantom
CradleGantry
Side
Figure 2.2
Geometry Calibration Phantom Perspective View
Shell
Page 8 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Phantom
Cradle Base Plate
Figure 2.3
End and Side Views of the Phantom
Small Tungsten BeadsLarge Tungsten Beads
Level screw
Page 9 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Phantom
Phantom Physical Dimensions
Physical Parameter Dimensions
Diameter of the Phantom Cylinder 140 mm
Diameter of the small Ball Bearings 3.2 mm
Diameter of the large Ball Bearings 6 mm
Number of Ball Bearings 108
Table 2.0
Page 10 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
Suggested Calibration Frequency:
Every 6 months or when it is required
Examples of when Geometry Calibration has to re-done:
Every time Flat Panel mechanical alignment is performed
Adjustments to the LINAC mechanical isocenter are performed
XRETIC Calibration is performed
Page 11 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
Figure 2.4PROJECTION MATRICES
PHANTOM MODEL: 3D INPUT PHANTOM IMAGES : 2D INPUT
Page 12 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
2D Projection Image ProcessingEach phantom projection image is processed to determine the
ball bearing (bb’s) 2D position and size regarding the image coordinate system (u,v).
Figure 2.5
u
V
0 0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 1 1 0 0 0 1 1 1 1 1 0 1 0 0 0 1 1 0 1 0 1 0 0 0 0 0 1 1 1 0 0 1 1 0 1 1 1 0 0 0 1 0 0 1 1 1 0 1 0 1 1 0 1 1 0 0 0 0 1 0 1 1 0 01 0 0 1 0 1 0 1 0 1 1 1 1 0 0 1 0 1 1 1 0 1 1 1 1 1 1 1
Encoded ball bearing list
from 2D projection images
Page 13 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
Figure 2.5
u
V
0 0 0 0 0 0 0 1 0 0 0 1 0 1 0 0 1 1 0 0 0 1 1 1 1 1 0 1 0 0 0 1 1 0 1 0 1 0 0 0 0 0 1 1 1 0 0 1 1 0 1 1 1 0 0 0 1 0 0 1 1 1 0 1 0 1 1 0 1 1 0 0 0 0 1 0 1 1 0 01 0 0 1 0 1 0 1 0 1 1 1 1 0 0 1 0 1 1 1 0 1 1 1 1 1 1 1
Encoded ball bearing list
from 2D projection images
2D Projection Image ProcessingEach phantom projection image is processed to determine the
ball bearing (bb’s) 2D position and size regarding the image coordinate system (u,v).
Page 14 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
3D Phantom Model Phantom Model provides the 3D input used in the calculation of the
projection matrices
Bead X,Y,Z Coordinates
Bead SizeBead TypeBead ID
Figure 2.6
Page 15 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
Coordinate System
Figure 2.8
r
Flat Panel
Gantry
X-ray Source
World (IEC)
coordinate system 1
Gantry
coordinate system 2
Camera
coordinate system 3
X-ray receptor
coordinate system4
Pixelised Imaging
coordinate system5
Y
Z
X
V
u
Page 16 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration
Coordinate System
Figure 2.8
r
Flat Panel
Gantry
X-ray Source
World (IEC)
coordinate system 1
Gantry
coordinate system 2
Camera
coordinate system 3
X-ray receptor
coordinate system4
Pixelised Imaging
coordinate system5
Y
Z
X
V
u
Page 17 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Projection Matrix
Projection MatrixHas 12 coefficients representing geometrical parameters such as
rotation angles and translation vectors
Provide the 3D (voxel) to 2D (pixel) geometrical mapping
Provides the input for the Backprojection step
Is computed for each MVision projection angle
When computed:
Allows 3D reconstruction of 2D MVision projection images
Can be used only in the LINAC where it was computed
Page 18 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Projection Matrix
The 2D points are related to the 3D points by the projection matrix P
Figure 2.9
34333231
24232221
14131211
pppp
pppp
pppp
P
),,(),( zyxRvuR
P
Projection matrix coefficients
2D Ball bearing position on
the phantom projection images
3D Ball bearing position on
the phantom model
Page 19 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Projection Matrix
Projection Matrix Naming Convention
The projection matrices are stored in a single file named:
ProjectionMatrices_<SID>_<StartAngle>_<EndAngle>_<AngleInc>_<CW|CCW>_<BAD>.xml
Example filenames:
ProjectionMatrices_1450_2700_1100_10_CW.xml (successful calibration)
ProjectionMatrices_1450_2700_1100_10_CW_BAD.xml (calibration failed)
Page 20 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Projection Matrix
Projection Matrix Naming Convention
The projection matrices are stored in a single file named:
ProjectionMatrices_<SID>_<StartAngle>_<EndAngle>_<AngleInc>_<CW|CCW>_<BAD>.xml
Example filenames:
ProjectionMatrices_1450_2700_1100_10_CW.xml (successful calibration)
ProjectionMatrices_1450_2700_1100_10_CW_BAD.xml (calibration failed)
Page 21 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Projection Matrix
Projection Matrices and Phantom Model Storage
Figure 2.10
Page 22 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Scale Factor
A I0 - MU Scale Factor is computed from the low MU MVision gain and used in the image reconstruction process Once determined is used to obtain non-attenuated beam data (I0) for
any MVision monitor unit setting
The Scale Factor is computed as:
Figure 2.11
MU
I0FactorScale
non-attenuated beam
Intensity from gain images
Monitor units used to
acquire gain images
Page 23 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Image Reconstruction
Reconstruction VolumeIt is a cube made of voxels and centered at the machine isocenter
It is used for the backprojection algorithm as the template to reconstruct MVision images using the projection data
Figure 2.12
Page 24 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Image Reconstruction
The backprojection process uses the projection matrices to reconstruct 3D MVision images
LINAC Radiation Source
Reconstructed
MVision Slice
b)
LINAC Radiation Source
Reconstruction
Volume
a)Figure 2.13
Flat panel
Page 25 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Image Reconstruction
Backprojection Process
Figure 2.14
)),(
ln(),(0I
vuivuN
X-ray source
Projection Image
Reconstruction
Volume
i(u,) - pixel intensity
at u,v positionu
v
Page 26 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Image Reconstruction
Backprojection Process
Figure 2.14
X-ray source
Projection Image
Reconstruction
Volume
i(u,) - pixel intensity
at u,v positionu
v
)),(
ln(),(0I
vuivuN
Page 27 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Image Reconstruction
Image filtering is necessary in the Backprojection process
Figure 2.15
e)
Filtered Image
1 Projection
a)
2 Projections
b)
4 Projections
c)
256 Projections
d)
Page 28 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain
Acquired like in a MVision acquisition
Used to:Correct differences in the Flat Panel diodes behavior (as in 2D
gain)
Applied to MVision imaging only
Compute a Scale Factor used in the image reconstruction process
Suggested Calibration Frequency: Every 2 Weeks
Page 29 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain
Two MVision Gain fields are delivered in free airHigh MU Gain used on MVision Geometry Calibration projection
data
Low MU Gain for regular MVision projection data
Two hundred gain images are acquired and averaged
Two MVision Gain files are createdThe software “knows” whether to save the gain as “High MU
MVision Gain” or “Low MU MVision Gain”
Page 30 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain
Gain Storage location
Figure 2.16
Page 31 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain Field Creation
Log in on the Practice Database and Service SoftwareUnder the Service Patient, the Site ‘Calibration Other’ on TxDelivery task card is selected
Figure 2.17
Page 32 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain Field Creation
Gain is selected as type of MVision field
Figure 2.18
Page 33 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain Field Creation
The Table eccentric is set to either 90˚ or 270˚ to prevent field override message
Figure 2.19
Page 34 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Gain Field Creation
MVision protocol selection
The 8MU or 15MU protocol is used for low MU gain
The 60MU protocol is used for High MU gain
Figure 2.20
Page 35 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Used to set the MVision acquisition and reconstruction parameters
Protocol creation application is launched from Coherence RTT
Acquisition Parameters
Only the Monitor Units can be edited/changed
Arc Angle, Gantry direction, Sampling and SID are fixed
Reconstruction Parameters
Slice Size (pixels)
Slice thickness (mm)
Kernels (filters)
Page 36 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Creating a MVision ProtocolUser must be logged on the Service Software prior to create/modify MVision protocols
Figure 2.21
Cone Beam Protocol icon
Page 37 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Protocol configuration
Figure 2.22
Reconstruction KernelSmoothing-Pelvis (Default)
Protocol Monitor Units
Slice Size (pixels)128 x 128
256 x 256 (default)
512 x 512
Slice Thickness (default 1mm)
MVision protocol name
Page 38 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the Slice Size on the image quality
Figure 2.23
Page 39 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the Slice Thickness on the image quality
Figure 2.24
Page 40 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the Kernels on the image quality
Figure 2.25
Page 41 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the Kernels on the image quality
Figure 2.26
Page 42 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the Kernels on the image quality
Figure 2.27
Page 43 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the amount of monitor units on the image quality
Figure 2.28
Page 44 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Protocol
Effect of the amount of monitor units on the image quality
Figure 2.29
Page 45 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MPR Thickness
Effect of the Mutiplanar Reformat Thickness (MPR) on the image quality
Figure 2.30
Page 46 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Phantom Positioning
The Geometry Calibration Phantom is aligned with the room lasers
Figure 2.31
a)
c)b)
Page 47 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Video: Geometry Calibration Phantom Positioning
Page 48 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Field Creation
Geometry Calibration is selected as type of MVision field
Figure 2.32
Page 49 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Field Creation
The 60MU protocol is selected for the Geometry Calibration
Figure 2.33
Page 50 July 2007 Copyright © Siemens AG 2007. All rights reserved.
MVision Control Console Display
Figure 2.34
Page 51 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Window
The Geometry Calibration is a sub task under the Calibration task card
Figure 2.35
Acquired Projection Images are displayed
here
Images that have a valid
Projection Matrix after
interpolation are displayed here
Images that have an invalid
Projection Matrices are displayed here
Images that have
Valid Projection Matrices are
displayed here
Page 52 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Results
Successful Calibration display
Figure 2.36
Dog Ear
Page 53 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Results
Successful Calibration display
Figure 2.37
Page 54 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Geometry Calibration Results
Calibration Fails due Invalid projection matrices
Figure 2.38
Page 55 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Video: Geometry Calibration
Page 56 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Troubleshooting
Errors during Geometry Calibration Geometry Calibration failed due missing projection
Figure 2.39
Page 57 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Troubleshooting
Errors during Geometry Calibration Mismatch between expected and actual gantry position Gantry position and delivered dose of a projection image were not
received by the control console
Figure 2.40
Page 58 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Troubleshooting
Geometry Calibration not performed
Figure 2.41
Page 59 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Troubleshooting
Fail to Pass the Geometry Calibration
Phantom misalignment: Align the phantom properly and re-acquire the images Wrong Phantom Orientation: Align the phantom with label “Gantry Side” toward the Gantry Phantom position in the table: Place the phantom following the recommendations in the Lab 4 MVision
Geometry Calibration Object in the Image: Make sure there is no object other than the phantom in the beam path Wrong MVision Protocol: MVision calibration images should be acquired using the 60MU protocol
Page 60 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Section 2 Review
Now that you have completed this section, you will be able to: Identify the fundamentals of MVision Geometry calibration
Understand the function of the Geometry Calibration Phantom
Describe the MVision Reconstruction Process
Describe the MVision Protocol parameter and how they
influence the image quality
Perform the MVision gain calibration
Perform the MVision Geometry Calibration
Page 61 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Section 2 Quiz
Page 62 July 2007 Copyright © Siemens AG 2007. All rights reserved.
1) The MVision Geometry Calibration is required in order to:
A) Correct flat panel sag
B) Map the geometry between a 3D object and its 2D
projections
C) Correct image offset caused by flat panel sag
D) A and C are correct
Page 63 July 2007 Copyright © Siemens AG 2007. All rights reserved.
2) In order to perform the Geometry Calibration one should:
A) Place the Geometry Calibration phantom at the
LINAC isocenter
B) Acquire MVision projection images of the phantom
C) Calibrate the XRETIC
D) A and B are correct
E) A, B and C are correct
Page 64 July 2007 Copyright © Siemens AG 2007. All rights reserved.
3) The Geometry Calibration phantom model and the phantom projection images provide the input necessary to:
A) Reconstruct the MVision images
B) Find out the LINAC isocenter coordinates
C) Compute the projection matrices
D) A and C are correct
Page 65 July 2007 Copyright © Siemens AG 2007. All rights reserved.
4) With regards the projection matrices it is correct to state that :
A) Provide voxel to pixel geometrical mapping
B) Do not depend on the flat panel alignment
C) Disable 3D reconstruction
D) All the above
Page 66 July 2007 Copyright © Siemens AG 2007. All rights reserved.
5) In the MVision Protocol configuration window the protocol parameters that can be changed are:
A) Monitor units, SID and Kernels
B) Kernels, Slice Size and Slice Thickness
C) Arc length, SID and Slice Size
D) Matrix size, monitor units and sampling
E) A and D are correct
Page 67 July 2007 Copyright © Siemens AG 2007. All rights reserved.
6) Increasing a slice size of a MVision image will cause the image spatial resolution, image noise and low contrast resolution respectively to:
A) Increase, decrease, decrease
B) Decrease, increase, decrease
C) Increase, increase, decrease
D) Decrease, decrease, increase
Page 68 July 2007 Copyright © Siemens AG 2007. All rights reserved.
7) A 5mm increase in the MVision slice thickness will:
A) Increase reconstruction time
B) Reconstruct images 5mm thick
C) Decrease image spatial resolution
D) Reconstruct images every 5mm
E) All the above is correct
Page 69 July 2007 Copyright © Siemens AG 2007. All rights reserved.
A) Smoothing Head-Neck kernel and 5mm MPR
B) 1mm MPR and Smoothing kernel
C) Edge Enhancing Head Neck kernel and 2mm MPR
D) 1mm MPR and Edge Preserving kernel
8) A Head and Neck patient of standard size will be imaged with MVision. The Kernel and MPR combination that will provide better soft tissue contrast and less cupping artifact:
Page 70 July 2007 Copyright © Siemens AG 2007. All rights reserved.
9)Overall an increase on the monitor in a MVision acquisition will:
A) Increase the acquisition time and image noise
B) Decrease image noise and increase soft tissue
contrast
C) Increase the visibility of high density objects
D) A and C are correct
Page 71 July 2007 Copyright © Siemens AG 2007. All rights reserved.
10) Regarding the Geometry Calibration which one of the following is correct:
A) The 60MU protocol should be used
B) Phantom alignment is critical to get accurate projection
matrices
C) Image reconstruction is not possible without Geometry
calibration
D) A and C are correct
Page 72 July 2007 Copyright © Siemens AG 2007. All rights reserved.
11) Which of the following is correct about the MVision gain calibration:
A) Is acquired in the same way as the 2D gains
B) Should always be acquired at 15MU
C) Table should be moved out of the beam path
D) The 200 gain images are averaged
E) C and D are correct
Page 73 July 2007 Copyright © Siemens AG 2007. All rights reserved.
Answer Key
1. B2. D3. C4. A5. B6. C7. C8. A9. B10. D11. E