2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Towards Proton Computed Tomography
L. R. Johnson, B. Keeney, G. Ross, H. F.-W. Sadrozinski, A. Seiden,
D.C. Williams, L. Zhang
Santa Cruz Institute for Particle Physics, UC Santa Cruz, CA 95064
V. Bashkirov, R. W. M. Schulte, K. Shahnazi
Loma Linda University Medical Center, Loma Linda, CA 92354
• Proton Energy Loss in Matter
• Proton Tomography / Proton Transmission Radiography
• Proton Transmission Radiography Data
• Proton Transmission Radiography MC Study
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Computed Tomography (CT)
CT:
•
•
•
•
•
Based on X-ray absorption
Faithful reconstruction of patient’s
anatomy
Stacked 2D maps of linear X-ray
attenuation
Coupled linear equations
Invert matrices and reconstruct zdependent features
Proton CT:
•
•
replaces X-ray absorption with
proton energy loss
reconstruct mass density (r)
distribution instead of electron
distribution
X-ray tube
Detector array
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Radiography: X-rays vs. Protons
Energy Loss of Protons, r
Attenuation of Photons, Z
N(x) = Noe-  x
DE  
X-Ray Absorption Coefficient
dE
dE
dx   r
Dl
dx
dx
Stopping Power for Protons
4
100
10
Bone
Muscle
H2O
Fat
100

Bethe-Bloch
dE/dl
[MeV/cm]
10
[1/cm]
1
Bone
Muscle
H2O
Fat
dE
dE
r
dl
dx
Low Contrast:
Dr = 0.1 for tissue, 0.5 for bone
0.01
1
10
100
X-Ray Energy [keV]
1000
1
10
NIST Data
Measure statistical process of X-ray removal
100
1000
Proton Energy E [MeV]
Measure energy loss on individual protons
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Negative Slope in the Bethe-Bloch Formula
Proton Energy Loss in H O
2
300
10
•Relatively low entrance dose
(plateau)
• Maximum dose at depth
(Bragg peak)
Energy
Deposit
in 1mm
[MeV/mm]
200
Proton
Energy
[MeV]
100
5
E = 250 MeV
E = 130 MeV
• Rapid distal dose fall-off
• RBE close to unity
0
0
10
20
30
40
Water Depth [cm]
Imaging
Treatment
0
50
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Protons vs. X-Rays in Therapy
Protons:
• Energy modulation
spreads the Bragg peak
across the malignancy
X-rays:
• High entrance dose
• Reduced dose at depth
• Slow distal dose fall-off
leads to increased dose
in non-target tissue
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Milestones of Proton Computed Tomography
•
R. R. Wilson (1946)
Points out the Bragg peak, defined range of protons
•
A. M. Cormack (1963)
Tomography
•
M. Goitein (1972)
2-D to 3-D, Simulations
•
A. M. Cormack & Koehler (1976)
Tomography, Dr  0.5 %
•
K. M.Hanson et al. (1982)
Human tissue, Dose advantage
•
U. Schneider et al. (1996)
Calibration of CT values,
Stoichiometric method
•
T. J. Satogata et al. (Poster M10-204)
Reduced Dose of Proton CT compared to X-Ray CT
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
What is new in pCT ?
• Increased # of Facilities with gantries etc.
See the following talk by Stephen G. Peggs)
• 2 Ph.D. Theses at PSI and Harvard Cyclotron
(U. Schneider & P. Zygmanski)
• Existence of high bandwidth detector systems for protons
–
–
–
–
semiconductors
high rate data acquisition ( > MHz)
large-scale (6”wafers)
fine-grained (100’s um pitch)
• Concerted simulation effort
–
–
–
–
Exploitation of angular and energy correlations
Support of data analysis
Optimization of pCT set-up (detector, energy, ..)
Dose calculation
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Potential of Proton CT: Treatment Planning
X-ray CT use in proton cancer therapy
can lead to large uncertainties in
range determination
Range Uncertainties
(measured with PTR)
> 5 mm
> 10 mm
> 15 mm
Schneider U. & Pedroni E. (1995),
“Proton radiography as a tool for
quality control in proton therapy,” Med
Phys. 22, 353.
Alderson Head Phantom
Proton CT can measure the density
distribution needed for range calculation.
There is an expectation (hope?) that with pCT the required dose can be reduced.
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Low Contrast in Proton CT
Sensitivity Study:
Inclusion of 1cm thickness and density r
at midpoint of 20cm H2O
1
Energy Loss in Water
0.8
0.6
Energy
Loss 0.4
[MeV/mm]
0.2
r
Energy
[MeV]
Range
[cm]
TOF
[ps]
1.0
164.1
38.2
1309
1.1
163.6
38.1
1311
1.5
161.5
37.7
1317
Proton Energy
200
Proton Energy
[MeV]
150
0
158.9
1.0
1.1
1.5
2.0
0
250
r*l
[g/cm2]
2.0
r
dE (250)
dE (250)+1.1
dE (250)+1.5
dE (250)+2
37.2
5
10
15
1325
Depth in H O [cm]
2
20
25
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Requirements for pCT Measurements
Tracking of individual Protons requires Measurement of:
•
Proton location to few hundred um
•
Proton angle to much better than a degree
Multiple Coulomb Scattering MCS1o
•
Average Proton Energy <E> to better than %
•
Improve energy determination with statistics
•
Issue: Dose D = Absorbed Energy / Mass
N/A = Fluence
( for Voxel with diameter d = 1mm
105 protons of 200 MeV = 7 [mGy])
N dE
D 
A dx
In order to minimize the dose, the final system
needs the best energy resolution!
Energy straggling is 1- 2 %.
 E  1%
•
CMS
13.6 MeV

z l / X0
p
 E 
E
N
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Dose vs. Voxel Size for pCT Measurements
Trade-off between Voxel size and Contrast (Dr) to minimize the Dose
Define voxel of volume d3
Dose in voxel = Dv
Take n =20cm/d settings
Total dose D = n* Dv
pCT: Contrast - Voxel Size - Dose
200 MeV Protons, 3Stdv.
1
1
Dose
[mGy]
0.01
0.01
Voxel Diameter
Require 3 Significance
E
D~ 2
D r d5
2
0.0001
d=5mm, Si
d=2mm, 2%
d=5mm, 2%
d=1cm, 2%
0.0001
-6
-6
10
10
10
0.01
0.1
1
3
Density Difference [g/cm ]
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Studies in Proton Computed Tomography
Collaboration Loma Linda University Medical Center – UC Santa Cruz
• Exploratory Study in Proton Transmission Radiography
– Silicon detector telescope
– Simple phantom in front
– Understand influence of multiple scattering and energy resolution on image
• Theoretical Study (GEANT4 MC simulation)
– Evaluation of MCS, range straggling, and need for angular measurements
– Optimization of energy
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Exploratory Proton Radiography Set-up
Proton Beam from Loma Linda University Medical Ctr @ 250 MeV
Degraded down to 130 MeV by 10” wax block
Object is aluminum annulus 5 cm long, 3 cm OD, 0.67 cm ID
Very large effects expected, x = r*l = 13.5 g/cm2
Traversing protons have 50 MeV, by-passing protons have 130 MeV
Silicon detector telescope with 2 x-y modules:
measure energy and location of exiting protons
Wax block
Air
Air
Object
Beam from Synchrotron
30 cm
27.3 cm
yx
250MeV
130MeV
Si Modules
50 + 130MeV
xy
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Silicon Detector Telescope
Simple 2D Silicon Strip Detector (SSD) Telescope of 2 x-y modules
built for Nanodosimetry
• 2 single-sided SSD / module
measure x-y coordinates
• GLAST Space Mission developed SSD
6.4 cm
• GLAST Readout
1.3 s shaping time
6.4 cm
194 m pitch, 400 m thickness
Binary readout
Time-over-Threshold TOT
Large dynamic range
5 x 64 channels
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Time-Over-Threshold ~ Energy Transfer
Digitization of position (hit channel) and energy deposit (TOT)
Time-over-Threshold TOT
TOT  charge  LET
TOT Measurement vs Charge
120
100
Pulse
Threshold
80
TOT
[us]
60
40
20
0
0
50
100
150
Input Charge [fC]
200
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Calibration of Proton Energy vs. TOT
Good agreement between measurement
and MC simulations
TOT vs Proton Energy
Measurement and Expectation
TOT Saturation
100
ToT measured
TOT expected
TOT
[us]
10
Proton Energy Resolution
0.4
 
0.4
0.3
0.3
0.2
0.2

LLUMC
Synchrotron P Beam
10
100
GLAST
SLAC Test Beam
1000
4
10
Proton Energy [MeV]
0.1
0
10
0.1
100
Proton Energy [MeV]
0
1000
Derive energy resolution from TOT vs. E plot
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Image of Al Annulus
Subdivide SSD area into pixels
1.
2.
Strip x strip 194um x 194um
4 x 4 strips (0.8mm x 0.8mm)
Image corresponds to
average energy in pixel
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Energy Resolution => Position Resolution
Slice of average pixel energy
in 4x4 pixels
(need to apply
off-line calibration!)
Hole “filled in”
“Fuzzy” Edges
Clear profile of pipe,
but interfaces are blurred
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Multiple Scattering: Migration
Image Features:
Washed out image in 2nd plane (30cm downstream)
Energy diluted at interfaces
(Fuzzy edges, Large RMS, Hole filled in partially)
Migration of events
are explained by Multiple Coulomb Scattering MCS
Protons scatter OUT OF target (not INTO).
Scatters have larger energy loss, larger angles, fill hole, dilute energy
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
GEANT4 MC: Use of Angular Information
Si Telescope allows reconstruction of beam divergence and scattering angles
Select 2 Areas in both MC and Data
A = inside annulus : Wax + Al
B = outside annulus : Wax only
Angular distributions well understood
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
GEANT4 MC: Migration
Beam profile in slice
Migration out of object
Energy of protons
entering front face
Protons
entering the object in front face
but leaving it before the rear face
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
GEANT4 MC: Use of Angular Information
Angular Cut at MCS of the Wax
Energy Profile before (after ) Angle Cut
Less Migration
Sharp Edges
(Energy Average)
Sharp Edges
(Energy RMS)
Energy RMS before (after ) Angle Cut
Angular cut improves the contrast at the interfaces
2002 IEEE NSS/MIC
pCT: Hartmut F.-W. Sadrozinski , SCIPP
Conclusions
Present status of pCT:
•
Long tradition, increased interest with many new proton accelerators
(see next talk by Stephen G. Peggs)
•
pCT will be useful for treatment planning
(reconstruction of true density distribution)
•
Potential dose advantage wrt X-rays
( see Poster M10-204 by Satogata et al. )
•
Use of GEANT4 simulation program aids in planning of experiments
(correlation of energy and angle, “migration”)
(see Poster M6-2, L. R. Johnson et al.)
Our future plans:
•
Optimization of beam energy
•
Investigation of optimal energy measurement method
•
Dose – contrast - resolution relationship on realistic phantoms