Transcript Energy Loss

Radiography Studies for Proton CT
M. Petterson, N. Blumenkrantz, J. Feldt, J. Heimann, D. Lucia,
H. F.-W. Sadrozinski, A. Seiden, D. C. Williams
SCIPP, UC Santa Cruz, CA 95064 USA
V. Bashkirov, R. Schulte
Loma Linda University Medical Center, CA 92354 USA
M. Bruzzi, D. Menichelli, M. Scaringella, C. Talamonti
INFN and Univ. of Florence, Italy
G.A.P. Cirrone, G. Cuttone, D. Lo Presti, N. Randazzo, V. Sipala
INFN Sezione di Catania, Italy
• Motivation for pCT
• Tracking Study: Most likely Path inside Phantom
• Calorimeter Studies: Imaging using Energy Loss
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Motivation for Proton Computed Tomography (pCT)
Proton radiation therapy is one of the most precise
forms of non-invasive image-guided cancer therapy.
Well defined range of protons in material,
low entrance dose,
dose maximum (“Bragg peak”)
rapid distal dose fall-off,
Limitation to the precision:
The use of x-ray computed tomography (CT) for imaging.
The resulting uncertainties can lead to range errors from several millimeters up
to more than 1 cm depending on the anatomical region treated.
In addition, during the treatment the patient might move and thus one needs to
verify the position of the target tissue in the beam.
In both cases, a CT system using the protons themselves is the solution.
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Proposed pCT System
R. W. Schulte, et al.,, “Conceptual design of a proton computed tomography system for applications in proton radiation therapy”,
IEEE Trans. Nucl. Sci., vol 51, no.3, pp 866 – 875, June 2004.
Challenges for pCT
For pCT we need to determine
for every single proton :
Where did it go?
How much energy did it loose?
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Apparatus
Measure Positions, Angles and Energy Loss of Single Protons
Use Loma Linda U MC proton beam (here 200 MeV)
Single Proton Tracking in 10 Si planes:
Single-sided, 192 strips, 236 mm pitch,
[GLAST 97 B.T.)
Module = x-y pair with 90o rotated strips
Entrance and exit telescopes + 1 “roving”
inside absorber
Energy Measurement in Calorimeter
Common Readout
One CsI crystal 5 cm x 5 cm x 15 cm
of Silicon Strip + Calorimeter into FPGA
Si has binary readout with Time-over-Threshold (ToT)
CsI read out with Photodiode into
Charge-to-Time-Converter (CTC)
Position Resolution: 70 mm
Angular Resolution: 5 mrad
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Tracking Studies
External Tracking of Proton predicts Path inside Absorber (MLP)
N. Blumenkrantz et al.,, “Prototype Tracking Studies for Proton CT ”, to appear in IEEE Trans. Nucl. Sci.,
To verify the “Banana”:
Measure displacement with
“Roving Plane”at 3 depth
“The most likely path of an energetic charged particle
through a uniform medium”
D C Williams Phys. Med. Biol. 49 (2004) 2899–2911
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Tracking Studies
Multiple Coulomb Scattering (MCS) at work
Displacement in Absorber
Displacement in Absorber
Correlation between “Roving plane #2” and exit parameters
Exit Displacement
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Exit Angle
Multiple Coulomb Scattering (MCS) overcome
MLP < 500 mm Localization within Absorber
Displacement [cm]
Displacement from incoming direction in the “Roving planes”
as a function of exit displacement bins of 500 mm (all angles).
0.5
0.4
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.4
-0.5
RMS = 490um
•
MLP width = 380 um
•
•
•
0
2 4
6 8 10 12 14 16 18 20
Analytical calculation of the most likely
path MLP (open symbols: the size of the
symbol is close to the MLP spread).
Good agreement data – MLP (~300 mm)
but systematically growing difference with
larger displacements:
need M.C. Simulation
Resolution inside Absorber better than
500 mm vs. MLP width of 380 mm
Using exit angle improves resolution:
by ~20%
Depth inside Absorber [cm]
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Energy Loss Studies
Image Low-contrast Phantom = voids in PMMA
Phantom = 12 PMMA plates
(each 1.25 cm thick ), 6th has holes
Phantom
p
2 SSD each (x-y)
Loma Linda University
Medical Center
Calorimeter
Hartmut F.-W. Sadrozinski RESMDD06
Calorimeter Calibration and Resolution
Energy Response with Proton Beams without PMMA
Ebeam =
201.1 MeV
Gaussian fit to
the falling slope of the CTC spectrum
( ¼ of protons are useful
in the determination of the mean energy)
CTC Calibration
1000
Ebeam =
100 MeV
CTC #
800
600
cal
12 PMMA+5 Si
12 PMMA + 4Si
11 PMMA + 5 Si
11 PMMA + 4 Si
Poly. (cal)
400
200
0
0
Ebeam =
0 MeV
(Pedestal)
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
50
100
E [MeV]
150
200
Calorimeter: Work in Progress
Energy Loss in PMMA, Energy Resolution and Straggling
Predicted and measured Mean Energy
680
cal
12 PMMA+5 Si
12 PMMA + 4Si
11 PMMA + 5 Si
11 PMMA + 4 Si
Poly. (cal)
CTC #
660
PMMA Data not described by NIST -> Leakage ?
MC underway.
640
620
90
95
100
105
E [MeV]
110
115
CsI Calorimeter Resolution
5
Energy resolution E in CsI
(measured and corrected for pedestal)
Include ~15 cm of PMMA
SigmaE [MeV]
4
3
2
SigmaE
SigmaE corr
1
0
0
50
100
E [MeV]
150
200
Energy resolution in PMMA
factor 2 to large to be caused
by energy straggling.
Leakage? MC underway.
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Proton Images
Energy Loss ~ Integrated Stopping Power ~ Dr*d
Resolve phantom into 2D pixels of size d and fit for the energy mean.
When adding the depth of the target voids d, construct 3D voxels used in CT
Energy contrast ~ density difference* voxel size Dr*d
Reconstructed energy
Phantom Holes
Diam. 1.0 cm, depth 1.25 cm (D)
Diam. 0.6 cm, depth 0.6 cm (B, F)
Loma Linda University
Medical Center
(pixel size 1.2 mm x 1.2 mm).
Black boxes
= target and control pixel
Hartmut F.-W. Sadrozinski RESMDD06
Reconstructed energy
(pixel size 2.4 mm x 2.4 mm)
Required Dose for Image Reconstruction
Determination of Fluence Limit by Data Reduction by factor n = 2,…, 64
Pixel size:
1.2 mm x 1.2 mm
n=2
n=16
n=4
n=8
n=32
n=64
Fluence limit : n = 8-16 (blurred image and many white pixels = no valid fit)
1rst Fluence Limit: number of protons in pixel > 10!
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Determination of Dose for the ~10 mm Voxel
Data Reduction by factor n = 1,…, 200
n =
2
4
n =
32
64
8
16
Pixel size:
8 mm x 8 mm
128
Pattern Recognition relies
on Significance
(energy contrast/resolution)
S  DEm /  m
Fluence limit is reached at about n=64
Required significance S >2
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
200
Dose –Contrast – Voxel Size
Dose ~ Fluence = Number of Protons / Voxel size
Dose  N / Area  N / d 2
Energy Contrast : DE  Dr * d
Significan ce S  DE /  m
S  2    DE  2 E / N
 E2
DoseD  2
D r d3
Dose to the patient during imaging depends
on the square of the effective energy resolution
(including beam straggling).
Proton energy resolution needs to be better
than energy straggling (~1%)
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06
Error on Energy Mean vs. Energy Resolution
Energy mean error can be improved with larger dose  m   E / N
Error on mean  m vs  E/sqrt(N)
error on mean [CTC #]
10
1
0.1
0
1
 E/sqrt(N) [CTC #]
10
Ratio of errors
Ratio (sm : sE/sqrt(N)) vs Fluence Reduction n
2.5
2.0
1.5
Error on the mean is
approximately equal to the
RMS/sqrt(N)
Revisit after improving calorimeter
leakage .
(Note that suspected leakage will not be a
factor in a larger calorimeter system)
1.0
0.5
0.0
1
Loma Linda University
Medical Center
10
Fluence Reduction Factor n
100
Hartmut F.-W. Sadrozinski RESMDD06
Clinical Interpretation
Dose D – Voxel Size d – Density Variation Dr
E
D~ 2
3
D r d
2
Dose D for two voxel sizes d:
d [cm]
D [mGy]
1.2
2.8*10-5
0.6
2.5*10-4
Ratio
d3 0.13
D 0.1
Dose vs. Voxel Size
1.E-03
Exp 1.2
Dose D [mGray]
Counting Limit
1.E-04
Expect after reduction of
calorimeter leakage:
resolution = straggling
the dose will be reduced by
factor 4.
1.E-05
1.E-06
1
Loma Linda University
Medical Center
10
Diameter d [mm]
100
Hartmut F.-W. Sadrozinski RESMDD06
Conclusions
We performed beam tests with the elements of a prototype pCT system
• High resolution tracker using silicon strips
• Crystal calorimeter
• Fast DAQ
Tracking studies show that the location within the phantom can be determined
using external telescopes to a precision of better than 500 mm.
Calorimeter studies show that imaging is possible using the energy loss of the
protons.
By reducing the number of protons, relative fluence limits are derived, which can
be expressed as minimum doses to image a voxel with a density difference.
They scale well as a function of voxel size d (~1/d3), as predicted.
MC is needed to take into account the finite extension of the prototype detectors
on the energy scale and energy resolution (P. Cirrone’s talk).
Loma Linda University
Medical Center
Hartmut F.-W. Sadrozinski RESMDD06