A group from Chicago, Argonne and Fermilab are interested in the development of large-area systems to measure the time-of-arrival of relativistic particles with (ultimately) 1 pico-second resolution, and for signals typical of Positron-Emission Tomography (PET), a resolution of 30 pico-seconds (sigma on one channel). These are respectively a factor of 100 and 20 better than the present state-of-the-art. This would involve development in a number of intellectually challenging areas: three-dimensional modeling of photo-optical devices, the design and construction of ultra-fast (200 GHz) electronics, the `end-to-end' (i.e. complete) simulation of large systems, real-time image processing and reconstruction, and the optimization of large detector and analysis systems for medical imaging. In each of these areas there is immense room for creative and innovative thinking, as the underlying technologies have moved faster than the applications. We collectively are an interdisciplinary (High Energy Physics, Radiology, and Electrical Engineering) group working on these problems, and it's interesting and rewarding to cross the knowledge bases of different intellectual disciplines.

http://psec.uchicago.edu/

The picosecond club

Timing Resolution of408nm vs. 635nm Laser

0

20

40

60

80

100

120

0.0 20.0 40.0 60.0 80.0 100.0

Npe

Tim

ing

Re

so

luti

on

(p

s)

635nm

408nm

- 2”x2” Burle/Photonis 85011 1024-anodes 10-micron pore tubes (skew timing over a 2”x2” tube ~100 ps)

- Hamamatsu PLP-10 picosec laser

TESTING A SILICON PHOTOMULTIPLIER TIME-OF FLIGHT(TOF) SYSTEM IN FERMILAB’S TEST BEAM FACILITYAnatoly Ronzhin, Mike Albrow, Erik Ramberg – Fermilab,

Jerry Vavra – SLAC,Henry Frisch, Tyler Natoli, Camden Eartly, Heejong Kim, Andrew Kobach, Fukun Tang,

Scott Wilbur, Jean-Francois Genat – University of Chicago,Ed May, Karen Byrum, John Anderson, Gary Drake – Argonne National Laboratory

12 September, 2008

TESTING A SILICON PHOTOMULTIPLIER TIME-OF FLIGHT(TOF) SYSTEM IN FERMILAB’S TEST BEAM FACILITYAnatoly Ronzhin, Mike Albrow, Erik Ramberg – Fermilab,

Jerry Vavra – SLAC,Henry Frisch, Tyler Natoli, Camden Eartly, Heejong Kim, Andrew Kobach, Fukun Tang,

Scott Wilbur, Jean-Francois Genat – University of Chicago,Ed May, Karen Byrum, John Anderson, Gary Drake – Argonne National Laboratory

12 September, 2008

Optimization of LSOfor Time-of-Flight PETOptimization of LSO

for Time-of-Flight PET

W. W. Moses1, M. Janecek1, M. A. Spurrier2, P. Szupryczynski2,3 ,W.-S. Choong1, C. L. Melcher2, and M. Andreaco3

1Lawrence Berkeley National Laboratory2University of Tennessee, Knoxville

3Siemens Medical Solutions

October 21, 2008

• Motivation• Reflector Optimization• LSO Optimization• PMT Optimization

Outline:

This work was supported by the NIH (NIBIB grant No. R01-EB006085).

Time-of-Flight in PETTime-of-Flight in PET

• Can localize source along line of flight.

• Time of flight information reduces noise in images.

• Variance reduction given by 2D/ct.

• 500 ps timing resolution 5x reduction in variance!

c = 30 cm/ns500 ps timing resolution

7.5 cm localization

•Time of Flight Provides a Huge Performance Increase!•Largest Improvement in Large Patients

•Time of Flight Provides a Huge Performance Increase!•Largest Improvement in Large Patients

D

Commercial TOF PET w/ LSOCommercial TOF PET w/ LSO

~550 ps Coincidence Timing Achieved~550 ps Coincidence Timing Achieved

Our Goal:“Demonstration” TOF PET Camera

Our Goal:“Demonstration” TOF PET Camera

Achieve the Best Timing Possible w/ LSOAchieve the Best Timing Possible w/ LSO

• With better timing resolution (t), huge gains predicted(23x variance reduction for 100 ps timing)

• Measure image improvement vs. timing resolution

• Use LSO scintillator

– Don’t change other factors that influence SNR(efficiency, scatter fraction, etc.)

What Limits Timing Resolution?What Limits Timing Resolution?

CrystalGeometry

326 psPMT

PMT

PMT 422 ps

Light Sharing 454 ps

PMT Array 274 ps

Baseline 160 ps

Non-TOF Block Detector Module

•Many Factors•“Optical Geometry” Particularly Important

•Many Factors•“Optical Geometry” Particularly Important

Proposed Side-Coupled DesignProposed Side-Coupled Design

Proposed Geometry(Side-Coupled Crystal)

ScintillatorCrystal

PMT

PMT

Shorter Optical Path Length & Fewer ReflectionsShorter Optical Path Length & Fewer Reflections

Conventional Geometry(End-Coupled Crystal)

384 ps(543 ps coinc.)

218 ps

Detector Module DesignDetector Module Design

PMT

(HamamatsuR-9800)

Two LSO Crystals(each 6.15 x 6.15 x 25 mm3)

Reflector

(on all five faces of each crystal, including the face between the

two crystals)

Optical Glue

(between lower crystal faces and PMT)

Hole in ReflectorOn Top Face of

Crystals

Two Side-Coupled Scintillator Crystals per PMTTwo Side-Coupled Scintillator Crystals per PMT

Detector Ring GeometryDetector Ring Geometry

Crystals Decoded by Opposing PMTCrystals Decoded by Opposing PMT

Exploded View

• Top face of each crystal (with hole in reflector) is coupled via a small (<1 mm) air gap to the edge of one opposing PMT.

• Light seen by the opposing PMT is used to decode the crystal of interaction.

Crystal ofInteraction

Camera GeometryCamera GeometrySection of Detector Ring

• Detector ring is 825 mm diameter, 6.15 mm axial• 192 detector modules, 384 LSO scintillator crystals• Adjustable gap (6 – 150 mm) between lead shields allows

“scatter-free” and “3-D” shielding geometries

Lead Shielding

Modules

“Real” Single-Ring PET Camera for Humans & Phantoms“Real” Single-Ring PET Camera for Humans & Phantoms

Surface & Reflector Optimization MethodSurface & Reflector Optimization Method

Measure Percentage Change in TimingMeasure Percentage Change in Timing

• Measure Timing of “Raw” Crystal(saw cut finish, Teflon tape reflector)

• Apply Surface Treatment

• Apply Reflector

• Re-Measure Timing

• Compute Percent Change

• Repeat for 5 Crystals & Average Results

• Do for All Surface / Reflector Combinations(>100 crystals, each measured twice)

R-9800

• 6.15 x 6.15 x 25 mm3

• Reflector on 5 Sides• Optical Grease• No Hole on Top

Same PMT forall measurements

Surface & Reflector ResultsSurface & Reflector Results

Reflector Saw Cut Chemically MechanicallyEtched Polished

Air GapTeflon 1.00 ± 0.17 0.94 ± 0.10 1.06 ± 0.09ESR 1.01 ± 0.08 0.96 ± 0.16 1.08 ± 0.08Lumirror 1.03 ± 0.13 0.96 ± 0.06 1.04 ± 0.12

GluedESR 0.99 ± 0.09 0.98 ± 0.03 1.01 ± 0.18Lumirror 1.04 ± 0.10 0.97 ± 0.10 0.98 ± 0.22Melinex 1.01 ± 0.16 0.99 ± 0.06 1.01 ± 0.20Epoxy 1.00 ± 0.16 0.95 ± 0.09 1.00 ± 0.15

Paint 0.96 ± 0.03

Average

1.001.021.01

0.991.001.000.98

Average 1.01 0.96 1.03

Optimization: LSO CompositionOptimization: LSO Composition

• Predicted Timing Resolution 1/sqrt(I0)• Want High Total Light Output & Short Decay Time

• Possible By Co-Doping LSO With Calcium

• Predicted Timing Resolution 1/sqrt(I0)• Want High Total Light Output & Short Decay Time

• Possible By Co-Doping LSO With Calcium

0

20

40

60

80

100

120

0 50 100 150 200

25 ns decay time50 ns decay time

Inte

nsity

Time (ns)

• Both Scintillators Have Same Light Output (photons/MeV)• Red Decay Time is 2x Longer Than Blue Decay Time

I(t) = I0 exp(-t/)

Light Output = I0

I0

25

30

35

40

45

50

0.95 1 1.05 1.1 1.15 1.2 1.25 1.3

Dec

ay T

ime

(ns)

Relative Light Output

Optimization: LSO CompositionOptimization: LSO Composition

Ca-Doping Gives High Light Output & Short Ca-Doping Gives High Light Output & Short

Normal LSO High Light Out

Short The Good Stuff!

= Ca-doped

0.1%0.2%0.4%

0.3%

• Ca-Doping Gives Good Timing Resolution• ~15% Improvement Over Normal LSO

• Ca-Doping Gives Good Timing Resolution• ~15% Improvement Over Normal LSO

150

160

170

180

190

200

210

220

230

80 100 120 140 160 180 200Tim

e R

eso

luti

on

(p

s fw

hm

)

Initial Intensity (I0)

Normal LSO

Scaled by1/sqrt(I0)

Measured Results: LSO CompositionMeasured Results: LSO Composition

= Ca-doped

0.1%

0.2%

0.4%0.3%

Optimization: Photomultiplier TubeOptimization: Photomultiplier Tube

• Predicted Timing Resolution 1/sqrt(QE)• Want High Quantum Efficiency Version of PMT

• Predicted Timing Resolution 1/sqrt(QE)• Want High Quantum Efficiency Version of PMT

Peak QE

Blue Sensitivity Index

190

200

210

220

230

240

250

10 11 12 13 14 15Tim

e R

eso

luti

on

(p

s fw

hm

)

Blue Sensitivity Index

• Increased QE Improves Timing Resolution by 7%• Expect 10% Improvement with 35% SBA PMT

• Increased QE Improves Timing Resolution by 7%• Expect 10% Improvement with 35% SBA PMT

Normal (“28% QE”) PMTs

Measured Results: High QE PMTsMeasured Results: High QE PMTs

Scaled by1/sqrt(Blue Index)

= “32% QE” PMTs

SummarySummary

• TOF PET with Significantly Better Timing is Possible• To Achieve, We Must “Think Outside the Block Detector”

• TOF PET with Significantly Better Timing is Possible• To Achieve, We Must “Think Outside the Block Detector”

Hardware Single Coinc. TOF (ps fwhm) (ps fwhm) GainEnd-Coupled Crystal 384 544 4.3

Side-Coupled Crystal 218 309 7.6

Etched, Reflector Paint 227 321 7.3

Ca-Doped LSO 182 258 9.1

32% QE PMT 155 219 10.6

35% QE “SBA” PMT 148 209 11.1

•Depth of Interaction & 150 ps Timing Resolution•11x Reduction in Variance in Practical Geometry•Depth of Interaction & 150 ps Timing Resolution•11x Reduction in Variance in Practical Geometry

ScintillatorArray

ThinnedSiPM Array

Future TOF-PET? (one layer for SPECT) Future TOF-PET? (one layer for SPECT)