Slides (.pdf) - EPFL
Slides (.pdf) - EPFL Slides (.pdf) - EPFL
Demonstration of a novel PET concept Chiara Casella, ETH Zurich EPFL Lausanne, March 12 th 2012 . .
- Page 2 and 3: intro axial concept AXPET AXPET per
- Page 4 and 5: intro axial concept AXPET AXPET per
- Page 6 and 7: intro axial concept AXPET AXPET per
- Page 8 and 9: intro axial concept AXPET AXPET per
- Page 10 and 11: intro axial concept AXPET AXPET per
- Page 12 and 13: intro axial concept AXPET AXPET per
- Page 14 and 15: intro axial concept AXPET AXPET per
- Page 16 and 17: intro axial concept AXPET AXPET per
- Page 18 and 19: intro axial concept AXPET AXPET per
- Page 20 and 21: intro axial concept AXPET AXPET per
- Page 22 and 23: intro axial concept AXPET AXPET per
- Page 24 and 25: intro axial concept AXPET AXPET per
- Page 26 and 27: intro axial concept AXPET AXPET per
- Page 28 and 29: intro axial concept AXPET AXPET per
- Page 30 and 31: layer1 clustPPLayCutsum_Zmm_lay1 Z
- Page 32 and 33: Y[mm] intro axial concept AXPET AXP
- Page 34 and 35: intro axial concept AXPET AXPET per
- Page 36 and 37: intro axial concept AXPET AXPET per
- Page 38 and 39: intro axial concept AXPET AXPET per
- Page 40 and 41: intro axial concept AXPET AXPET per
- Page 42 and 43: intro axial concept AXPET AXPET per
- Page 44 and 45: intro axial concept AXPET AXPET per
- Page 46 and 47: intro axial concept AXPET AXPET per
- Page 48 and 49: intro axial concept AXPET AXPET per
- Page 50 and 51: intro axial concept AXPET AXPET per
Demonstration of a<br />
novel PET concept<br />
Chiara Casella, ETH Zurich<br />
<strong>EPFL</strong> Lausanne, March 12 th 2012<br />
. .
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Outline<br />
AX-PET : AXial Positron Emission Tomography (PET)<br />
•Brief introduction about PET (Positron Emission Tomography)<br />
•Axial concept<br />
What is it ? Why?<br />
•AX-PET detector<br />
choice of the detector elements<br />
AX-PET modules<br />
“demonstrator” for a PET scanner<br />
•AX-PET detector performance<br />
from characterization measurements with 22-Na sources<br />
•Simulations<br />
•Tomographic image reconstruction<br />
description of the reconstruction methods<br />
measurements campaigns with phantoms and radiotracers<br />
reconstructed images<br />
•Perspectives<br />
preliminary results with Digital Si-PM as alternative photodetectors<br />
•Conclusions<br />
Chiara Casella, 12/3/2012 1
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
AX-PET in the context of HEP<br />
‣ Long tradition of using technologies from High Energy Physics into other fields,<br />
and particularly medical applications.<br />
‣ AX-PET : small size calorimeter, using scintillating crystals, WLS, photodetectors<br />
“borrowed” from HEP<br />
Selection of most relevant AX-PET papers :<br />
Novel Geometrical Concept of a High Performance Brain PET Scanner<br />
Principle, Design and Performance Estimates<br />
J. Seguinot et al, Nuovo Cimento C Vol 29, No 4, pp 429-463 (2006)<br />
High precision axial coordinate readout for an axial 3-D PET detector<br />
module using a wave length shifter strip matrix<br />
A. Braem et al, NIM A 580 (2007), 1513-1521<br />
Wavelength shifter strips and G-APD arrays for the read-out of the<br />
z-coordinate in axial PET modules<br />
A. Braem et al, NIM A 586 (2008), 300-308<br />
The AX-PETdemonstrator—Design, construction and characterization<br />
P. Beltrame et al, NIM A 654 (2011) 546-559<br />
The AX–PET Concept: New Developments And Tomographic Imaging<br />
P.Beltrame et al, 2011 IEEE NSS Conference Record MIC 22-5<br />
axial concept,<br />
HPD readout<br />
WLS strips<br />
G-APD as photodetectors<br />
Characterization<br />
& performance<br />
Tomographic images<br />
for more details : https://twiki.cern.ch/twiki/bin/view/AXIALPET/WebHome<br />
2
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Positron emission / annihilation<br />
‣ basis of the PET system Positron Emission :<br />
‣ β + decay of different radionuclides<br />
Radionuclide Half life Emax_e+(MeV)<br />
11 C 20.4 min 0,96<br />
15 O 122 sec 1,73<br />
18 F 109,8 min 0,63<br />
22 Na 2.6 years 0,55<br />
‣ Positron Annihilation :<br />
2 photons emitted<br />
‣“back - to - back”<br />
‣ Eγ = 511 keV<br />
Unstable<br />
parent<br />
nucleus<br />
p → n + e + + νe<br />
e + e − → γγ<br />
(Eγ = 511 keV )<br />
p → n + e + + νe<br />
e + e − → γγ<br />
(Eγ = 511 keV )<br />
positron<br />
annihilation<br />
positron emission<br />
Physics of the positron annihilation => fundamental limits to the spatial resolution of PET<br />
‣ Finite positron range (ρ)<br />
annihilation position ≠ emission point<br />
ρ depends on the energy of the positron (i.e. on the radioisotope)<br />
‣ Non-collinearity of the 2 photons<br />
residual momentum of the e+e- at the annihilation<br />
the 2 photons are emitted with a small deviation from 180˚ (Δθ ~ 0.5˚)<br />
blurring of the spatial resolution R_FWHM ~ 0.0022 x D [mm]<br />
3
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Positron Emission Tomography : General principles<br />
PET detector principle :<br />
coincidence of 2 photons of defined energy (511 keV) and emitted on the same line (“back to back”)<br />
(3)<br />
(2)<br />
(1)<br />
(5)<br />
(4)<br />
(1) Inject the radiotracer into the body<br />
(2) Wait for uptaking period<br />
(3) Start the acquisition (i.e. detection of coinc. events)<br />
(4) Feed the data into the reconstruction algorithms<br />
(5) image of the activity concentration<br />
‣ PET : “in-vivo” functional imaging technique<br />
‣ get a (quantitative) image of the radio-tracer concentration<br />
A.Del Guerra, CERN Academic Training 2009<br />
4
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Requirements for an ideal PET<br />
number of counts does matter !!!<br />
50000 counts 100000 counts 200000 counts 500000 counts 1M counts 2M counts<br />
‣ how to improve counting statistics ?<br />
‣ increase radio-tracer concentration<br />
‣ extend the scan duration time<br />
‣ OPTIMIZED DETECTOR SENSITIVITY<br />
( geometry / materials )<br />
‣ HIGH SPATIAL RESOLUTION (=> be able to resolve small structures)<br />
‣ GOOD TIMING RESOLUTION (small coincidence window => reduce random coincidences)<br />
‣ GOOD ENERGY RESOLUTION (=> reject scattering events in the body)<br />
A.Del Guerra, CERN Academic Training 2009<br />
5
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Requirements for an ideal PET<br />
number of counts does matter !!!<br />
2009<br />
Training Academic CERN Guerra,<br />
50000 counts 100000 counts 200000 counts 500000 counts 1M counts 2M counts A.Del<br />
‣ how to improve counting statistics ?<br />
‣ increase radio-tracer concentration<br />
‣ extend the scan duration time<br />
‣ OPTIMIZED DETECTOR SENSITIVITY<br />
( geometry / materials )<br />
‣ HIGH SPATIAL RESOLUTION (=> be able to resolve small structures)<br />
‣ GOOD TIMING RESOLUTION (small coincidence window => reduce random coincidences)<br />
‣ GOOD ENERGY RESOLUTION (=> reject scattering events in the body)<br />
5
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
δp<br />
High resolution vs high sensitivity<br />
in “conventional”<br />
PET scanners :<br />
θ<br />
‣ always a compromise between<br />
!"<br />
δp = L · sinθ<br />
‣ solution : add DOI (Depth Of Interaction) information<br />
‣ several attempts / different strategies<br />
scintillator based<br />
radial arrangement<br />
ɛ =1− e −µ·L max interaction efficiency,<br />
long L<br />
min parallax error<br />
- deterioration of the spat. resol.<br />
- non uniformity in the field of view<br />
short L<br />
‣ good spatial resolution (small L, small δp)<br />
‣ good detection efficiency (long L)<br />
6
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
DOI attempts :<br />
(a) multiple<br />
crystals/<br />
photodetector<br />
layers<br />
Conventional PET schemes<br />
(b) dual-ended<br />
photodetectors<br />
DOI from the ratio of<br />
the 2 PD responses<br />
only partial DOI information !<br />
(c) Phoswich design<br />
different types of<br />
crystals with diff decay<br />
times<br />
=> DOI from pulse<br />
shape discrimination<br />
(e) dual layer offset<br />
position.<br />
DOI from different light output<br />
profiles<br />
need a precise 3D identification of the photon interaction point<br />
(d) monolithic. Iterative stat.<br />
models with DOI from light output<br />
intensity and/or light spread profiles<br />
Peng, Levin - Current Pharmaceutical<br />
Biotechnology, 2010, Vol 11, No. 6<br />
7
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
The AXIAL concept<br />
AX-PET approach to the DOI problem : change the geometry !<br />
from radial ... ... to axial !<br />
‣ long and axially oriented crystals<br />
‣ DOI information = position of the hit crystal<br />
8
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
axial direction<br />
The detector concept<br />
(1) TRANSAXIAL COORDINATE (x,y)<br />
scintillator crystals<br />
- Transaxial coordinate: from position of the hit crystal<br />
- Transaxial resolution = d/√12 FWHM<br />
Y<br />
- To increase spatial resolution => Reduce crystals size (d)<br />
- To increase sensitivity => Add additional layers<br />
Z<br />
X<br />
d<br />
9
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
axial direction<br />
The detector concept<br />
(1) TRANSAXIAL COORDINATE (x,y)<br />
scintillator crystals<br />
- Transaxial coordinate: from position of the hit crystal<br />
- Transaxial resolution = d/√12 FWHM<br />
Y<br />
- To increase spatial resolution => Reduce crystals size (d)<br />
- To increase sensitivity => Add additional layers<br />
Z<br />
X<br />
d<br />
WLS (Wave Length Shifter) array<br />
w<br />
(2) AXIAL COORDINATE (z)<br />
scintillation<br />
process<br />
wls<br />
photodetector<br />
ʻescaping coneʼ<br />
(outside total<br />
internal reflection)<br />
crystal<br />
reflective Al<br />
coating<br />
- Axial coordinate : center of gravity method<br />
- Axial resolution < w<br />
Z<br />
9
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
axial direction<br />
The detector concept<br />
(1) TRANSAXIAL COORDINATE (x,y)<br />
scintillator crystals<br />
- Transaxial coordinate: from position of the hit crystal<br />
- Transaxial resolution = d/√12 FWHM<br />
Y<br />
- To increase spatial resolution => Reduce crystals size (d)<br />
- To increase sensitivity => Add additional layers<br />
Z<br />
X<br />
3D localization of the photon interaction point<br />
without compromising between spatial resolution and sensitivity<br />
d<br />
WLS (Wave Length Shifter) array<br />
w<br />
(2) AXIAL COORDINATE (z)<br />
scintillation<br />
process<br />
wls<br />
photodetector<br />
ʻescaping coneʼ<br />
(outside total<br />
internal reflection)<br />
Z<br />
crystal<br />
reflective Al<br />
coating<br />
- Axial coordinate : center of gravity method<br />
- Axial resolution < w<br />
9
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
AX-PET Collaboration<br />
10
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
AX-PET GOAL : PET demonstrator<br />
Goal of the AX-PET collaboration:<br />
Build and fully characterize a “demonstrator” for a PET<br />
scanner based on the axial concept. Assess its performances.<br />
Demonstrator =><br />
Two identical AX-PET modules, used in coincidence<br />
Characterization / Performance =><br />
- test each individual module in a dedicated setup<br />
- characterization in the coincidence setup<br />
- reconstruction of the images of extended objects<br />
- simulations<br />
11
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
detect photons, Eγ = 511 keV =><br />
bare scintillating crystals with escaping light =><br />
Detector choice: Scintillator<br />
• inorganic scintillator<br />
• high Z, high ρ<br />
• non hygroscopic<br />
• nude crystals (i.e. unwrapped, uncoated)<br />
• perfectly polished surfaces ; sharp edges<br />
LYSO crystals (Lu1.8Y0.2SiO5: Ce) Prelude 420 from Saint Gobain :<br />
ZLu = 71<br />
ZLu = 71<br />
176-Lu is a naturally radioactive β emitter:<br />
• lyso bars : 3 x 3 x 100 mm (A ~ 39 Bq/g; β-decay followed by a γ-cascade)<br />
useful for the calibration<br />
3 each<br />
• Al coated on the non-readout face (R ~ 80-85%)<br />
• measured : λ_opt = (412 ± 31) mm ; (ΔE/E)_intr = (8.3 ± 0.5) % FWHM , @511 keV<br />
12
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
• requirement : high absorption of the crystal<br />
scintillation photons (blue)<br />
• EJ-280-10x from Eljen Technology<br />
• blue to green WLS<br />
• highly doped (x10 higher dye concentration<br />
than standard) to maximize the absorption<br />
(absorption length _ blue light ~ 0.4 mm)<br />
• each WLS strip : 3 x 0.9 x 40 mm 3<br />
• decay time = 8.5 ns<br />
Detector choice: WLS strips<br />
• measurements : λ_opt = (188 ± 36) mm 4 cm<br />
photodetector side<br />
mirror coated side<br />
13
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
‣ newest frontier in photo-detection:<br />
‣ excellent photon counting capabilities<br />
Detector choice: Photodetectors<br />
‣ array of commonly biased APD used in Geiger mode<br />
‣ output : analogue sum of all the firing APD cells<br />
‣ high gain (10 5 to 10 6 ) at low bias V (~ 70 V)<br />
‣ advantages of a Si sensor:<br />
‣ high QE √<br />
‣ compactness √<br />
‣ insensitive to magnetic field (MRI comp.)<br />
‣ temperature dependent √<br />
‣ dark rate (~ 1 MHz @ thr = 0.5 pe)<br />
SiPM (Si photomultipliers)<br />
also called G-APD (Geiger-mode APD)<br />
also called MPPC (Multi Pixel Photon Counter) - from Hamamatsu<br />
√<br />
√<br />
√<br />
√<br />
for crystals for WLS<br />
~ 1000 pe ~ 10 - 50 pe<br />
MPPC S10362‐33‐050C :<br />
• 3x3 mm 2 ac(ve area<br />
• 50 μm x 50 μm pixel<br />
• 3600 pixels<br />
MPPC 3.22×1.19 Octagon‐SMD :<br />
• 1.2 x 3.2 mm 2 ac(ve area<br />
• 70 μm x 70 μm pixel<br />
• 782 pixels<br />
• custom made units<br />
AX-PET operational<br />
parameters<br />
14
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
LYSO crystals<br />
3 mm<br />
10 cm<br />
3.5 cm<br />
2.8 cm<br />
WLS<br />
The AX-PET module<br />
MODULE<br />
• 6 layers<br />
• 8 crystals / layer<br />
• 26 WLS / layer<br />
• 48 crystals + 156 WLS = 204 channels<br />
• staggering in the crystals layout to optimize photon interaction probability<br />
• optical separation between layers<br />
• each crystal and WLS strip individually coupled to its photodetector<br />
• crystals / WLS strips readout on alternate sides (to optimize packing density)<br />
15
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
The AXPET module<br />
16
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Individual analogue readout of MPPC output<br />
Custom designed DAQ system<br />
‣ fully analogue readout chain<br />
‣ not optimized at all for this specific<br />
application<br />
‣ Amplifiers: OPA486 (Lyso) / OPA487<br />
(WLS)<br />
‣ Fast energy sum for all the crystals in<br />
the module<br />
‣ VATA GP5 chip<br />
- 128 ch charge sensitive integrating<br />
- fast (~ 50ns shaper + discriminator) /<br />
slow (~ 250ns shaper) branches<br />
- sparse readout mode: only the<br />
channels above thr are multiplexed into<br />
the output<br />
‣ analogue info processed by custom made<br />
VME ADC<br />
Readout & DAQ<br />
17
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Individual analogue readout of MPPC output<br />
Custom designed DAQ system<br />
‣ fully analogue readout chain<br />
‣ not optimized at all for this specific<br />
application<br />
‣ Amplifiers: OPA486 (Lyso) / OPA487<br />
(WLS)<br />
‣ Fast energy sum for all the crystals in<br />
the module<br />
‣ VATA GP5 chip<br />
- 128 ch charge sensitive integrating<br />
- fast (~ 50ns shaper + discriminator) /<br />
slow (~ 250ns shaper) branches<br />
- sparse readout mode: only the<br />
channels above thr are multiplexed into<br />
the output<br />
‣ analogue info processed by custom made<br />
VME ADC<br />
Readout & DAQ<br />
17
PPC<br />
PPC<br />
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
‣ analogue sum of the whole module<br />
(i.e. total energy over 48 crystals)<br />
Fast energy sum & Trigger<br />
‣ with a proper threshold choice (LL x HL X notHHL)<br />
=> select only events with 511 keV total<br />
energy deposition<br />
TRIGGER<br />
OPA486<br />
OPA486<br />
test<br />
! E (crystals only)<br />
test<br />
ΣE_Mod1<br />
! E (crystals only)<br />
MODULE1<br />
fast<br />
ΣE_Mod2<br />
LYSO SUM<br />
MODULE2<br />
LYSO SUM<br />
fast<br />
‘slow’<br />
VATAGP5 (128 ch.)<br />
‘slow’<br />
VATAGP5 (128 ch.)<br />
First estimate of the time resolution:<br />
•! measure delay of coincidence wrt Module2<br />
•! measurement from the scope [Lecroy Waverunner LT584 L 1GHz]<br />
S&H<br />
S&H<br />
Channels<br />
above threshold<br />
Analog Channels output:<br />
above Light threshold per<br />
crystal / WLS<br />
Analog output:<br />
Light per<br />
crystal / WLS<br />
LL<br />
HL<br />
HHL<br />
Mod 1, energy sum (scope measurement), 22 Na source<br />
TRIGGER = 2 modules<br />
- each one discriminated @ 511 keV energy sum<br />
- used in coincidence<br />
=> Selection of the good events<br />
18
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Compton scattering in the detector<br />
‣ fast energy sum & trigger<br />
‣ high granularity in the module<br />
‣ good energy resolution<br />
Compton scattering<br />
E1 + E2 = 511 keV<br />
if the event is :<br />
• correctly identifiable (e.g. Compt. kinematics)<br />
=> the event can be included in the reconstruction => improved sensitivity<br />
• ambiguous (1-2 OR 2-1 ?)<br />
=> the event is rejected => improved resolution<br />
}<br />
Compton scattering in the body<br />
Events killed by the trigger !<br />
Possibility to tag Compton scattered events<br />
in the detector<br />
“Inter-Crystal Scattering” events (ICS)<br />
E1 + E2 = 511 keV<br />
win - win situation !<br />
Esum = 511 keV Esum < 511 keV<br />
19
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
AX-PET inspired other developments<br />
COMPET :<br />
University of Oslo, Norway - E. Bolle et al.<br />
research project for a pre-clinical PET scanner with high sensitivity,<br />
high resolution. MRI compatible<br />
- no axial geometry<br />
- 3D reco of photon interaction point with LYSO + WLS + G-APD<br />
E. Bolle et al, NIM A 648(2011) S93-S95<br />
Tampere University (Finland) :<br />
build a small specific scanner based on AX-PET (toward possible commercialization...)<br />
Low cost planar detector for PET<br />
Triumph, Canada - F. Retiere et al.<br />
F. Retiere, NIM A (2011) doi:10.1016/j.nima.2011.12.084<br />
20
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
single module<br />
characterization<br />
Characterization measurements<br />
with point-like 22 Na source (diam = 0.25 mm, A~900 kBq) , @ CERN<br />
2-D<br />
moving<br />
station<br />
tagger<br />
PMT<br />
22Na source<br />
module<br />
- two different taggers<br />
- different distances tagger / source / module<br />
=> both uniform illumination of the module and precisely collimated beam spot<br />
module<br />
22Na source<br />
2-D<br />
moving<br />
station<br />
module<br />
characterization<br />
of modules<br />
in coincidence<br />
AX-PET performance:<br />
NIM A 654 (2011) 546-559<br />
1. Energy measurement / energy calibration => LYSO response<br />
2. Position measurement / spatial resolution => WLS response<br />
21
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
LYSO No. 21 - 22Na coinc. trigger<br />
450<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
0<br />
Lu X-ray<br />
photopeak<br />
Compton<br />
0 100 200 300 400 500 600 700<br />
ADC counts<br />
typical energy spectrum of one LYSO inside the<br />
module :<br />
‣ photopeak (511 keV)<br />
‣ Compton continuum (0 - 340 keV)<br />
‣ Lu X-ray peak (~ 55 keV)<br />
LYSO energy response<br />
511 keV<br />
all events<br />
N_lyso = 1<br />
fit: gaussian<br />
55 keV<br />
511<br />
keV 511<br />
keV<br />
Light yield at 511 keV ~ 1000 pe<br />
(from independent calibration measurements)<br />
< 511 keV<br />
R_FWHM (@511 keV) [%]<br />
Energy resolution<br />
14<br />
13.5<br />
13<br />
12.5<br />
12<br />
11.5<br />
11<br />
10.5<br />
Energy resolution<br />
‣ from gaussian fit of the photopeak<br />
‣ AFTER ENERGY CALIBRATION<br />
Module 1<br />
Module 2<br />
10<br />
0 20 40 60 80 100<br />
LYSO_ID<br />
< R_FWHM > ~ 11.8% @511 keV<br />
(averaged on 96 LYSO crystals)<br />
22
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
LYSO energy calibration<br />
Photopeak + Intrinsic Lu radioactivity: very good tool for the energy calibration<br />
‣ LYSO contains Lu-176<br />
‣ A ~ 39 cps/g<br />
=> ~ 250 Bq / bar<br />
=> ~ 12 kHz / module<br />
LYSO No. 21 - 22Na coinc. trigger<br />
450<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
0<br />
0 100 200 300 400 500 600 700<br />
LYSO No. 21 - intrinsic radioactivity<br />
[511 keV]<br />
ADC counts<br />
600<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
with source,<br />
coincidence<br />
0 100 200 300 400 500 600 700<br />
[303 keV]<br />
ADC counts<br />
[55 keV]<br />
[202 keV]<br />
all events<br />
no source,<br />
internal trg.<br />
N_lyso = 1<br />
fit: gaussian<br />
Peak position [ADC counts]<br />
Energy Calibration ( LYSO No. 21)<br />
500<br />
450<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
same procedure applied<br />
identically for every channel<br />
deviation from linearity<br />
(~ 5% effect)<br />
<br />
En(ADC) =E0 − a × ln 1 − ADC<br />
<br />
b<br />
0<br />
0 100 200 300 400 500<br />
Energy [ keV ]<br />
MPPC saturation. Due to:<br />
- limited nr of cells in the MPPC (3200)<br />
- important light yield in the scintillator (~1000).<br />
23
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
WLS response<br />
typical integrated raw spectra of few WLS strips<br />
• beam spot collimated at the center of the module (WLS 13)<br />
• 511 keV energy deposition in the LYSO<br />
Collimated beam spot, WLS response<br />
1800 WLS Nr. 13 (central)<br />
1600<br />
1400<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
WLS Nr. 14 (side)<br />
WLS Nr. 24 (peripheral)<br />
0<br />
0 100 200 300 400 500 600 700 800<br />
ADC counts<br />
• more than1 WLS participate to the event (typically 2-4)<br />
• noise should not be included<br />
WLS cluster: Summed ADC counts<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
0 100 200 300 400 500 600 700 800<br />
ADC counts<br />
Light yield in WLS cluster ~ 100 pe<br />
@511 keV LYSO energy deposition<br />
(from independent calibration measurements: 1pe ~ 4 ADC)<br />
axial coordinate :<br />
derived from center of gravity method<br />
from all the WLS participating to the cluster<br />
cluster<br />
noise<br />
24
layer1<br />
clustPPLayCutsum_Zmm_lay1<br />
Z c.grav. in layer 1 (PPeak samelay 1!cutsum)<br />
Entries 4032<br />
400<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
0<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Mean 48.4<br />
RMS 3.356<br />
2<br />
! / ndf 253.2 / 10<br />
Prob 0<br />
p0 304.8 ± 7.5<br />
p1 48.13 ± 0.06<br />
p2 2.213 ± 0.066<br />
35 40 45 50 55 60<br />
z [mm]<br />
clustPPLayCutsum_Zmm_lay4<br />
Z c.grav. in layer 4 (PPeak samelay 1!cutsum)<br />
Entries 9623<br />
Mean 48.08<br />
RMS 2.868<br />
2<br />
! / ndf 10.71 / 7<br />
Prob 0.152<br />
p0 1056 ± 16.9<br />
p1 47.91 ± 0.02<br />
p2 1.501 ± 0.026<br />
35 40 45 50 55 60<br />
z [mm]<br />
layer2<br />
layer4 layer5<br />
z coordinate (COG) :<br />
clustPPLayCutsum_Zmm_lay2<br />
Z c.grav. in layer 2 (PPeak samelay 1!cutsum)<br />
Entries 5716<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
Mean 48.47<br />
RMS 3.017<br />
2<br />
! / ndf 17.16 / 8<br />
Prob 0.02848<br />
p0 561.3 ± 11.8<br />
p1 48.4 ± 0.0<br />
p2 1.744 ± 0.043<br />
35 40 45 50 55 60<br />
z [mm]<br />
clustPPLayCutsum_Zmm_lay5<br />
Z c.grav. in layer 5 (PPeak samelay 1!cutsum)<br />
Entries 10762<br />
Mean 48.32<br />
RMS 3.37<br />
2<br />
! / ndf 30.27 / 6<br />
Prob 3.489e-05<br />
p0 1218 ± 19.0<br />
p1 48.24 ± 0.02<br />
p2 -1.391 ± 0.024<br />
35 40 45 50 55 60<br />
z [mm]<br />
layer3<br />
500<br />
400<br />
300<br />
200<br />
100<br />
0<br />
Axial spatial resolution<br />
clustPPLayCutsum_Zmm_lay3<br />
Z c.grav. in layer 3 (PPeak samelay 1!cutsum)<br />
900<br />
Entries<br />
Mean<br />
7442<br />
48.36<br />
800<br />
RMS<br />
2<br />
! / ndf<br />
3.099<br />
73.49 / 8<br />
700<br />
Prob<br />
p0<br />
9.906e-13<br />
765.8 ± 13.9<br />
600<br />
p1 48.17 ± 0.03<br />
p2 1.587 ± 0.029<br />
3000<br />
2500<br />
2000<br />
1500<br />
1000<br />
500<br />
0<br />
35 40 45 50 55 60<br />
z [mm]<br />
layer6<br />
clustPPLayCutsum_Zmm_lay6<br />
Z c.grav. in layer 6 (PPeak samelay 1!cutsum)<br />
Entries 17215<br />
each measured σz includes :<br />
• intrinsic spatial resolution<br />
• positron range (ρ ~ 0.54 mm)<br />
• non collinearity (÷d)<br />
• beam divergency (÷d)<br />
• (source dimensions; ø=250μm)<br />
Mean 48.08<br />
RMS 3.356<br />
2<br />
! / ndf 39.68 / 3<br />
Prob 1.248e-08<br />
p0 2854 ± 36.7<br />
p1 48.07 ± 0.01<br />
p2 -0.876 ± 0.012<br />
35 40 45 50 55 60<br />
z [mm]<br />
Intrinsic axial spatial resolution<br />
Mod 1 : 1.75 mm FWHM<br />
Mod 2 : 1.83 mm FWHM<br />
Trans-axial spatial resolution<br />
• = d/√12 = 0.87 mm (digital) ; FWHM ~ 2 mm<br />
• = d/√12 = 0.87 mm (digital) ; FWHM ~ 2 mm<br />
z<br />
x<br />
y<br />
layer6<br />
layer1<br />
25
Y[mm]<br />
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
TOP View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
N_coinc_evts=1<br />
-100 -50 0 50 100<br />
X[mm]<br />
Two modules coincidences<br />
Z[mm]<br />
SIDE View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
AX-PET very first coincidence event !<br />
N_coinc_evts=1<br />
-100 -50 0 50 100<br />
X[mm]<br />
26
Y[mm]<br />
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
TOP View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
-100 -50 0 50 100<br />
X[mm]<br />
Two modules coincidences<br />
N_coinc_evts=10<br />
Z[mm]<br />
SIDE View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
N_coinc_evts=10<br />
-100 -50 0 50 100<br />
X[mm]<br />
26
Y[mm]<br />
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
TOP View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
-100 -50 0 50 100<br />
X[mm]<br />
Two modules coincidences<br />
N_coinc_evts=100<br />
Z[mm]<br />
SIDE View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
N_coinc_evts=100<br />
-100 -50 0 50 100<br />
X[mm]<br />
26
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Z[mm]<br />
SIDE View - d(Mod1,Mod2) = 150 mm<br />
100<br />
50<br />
0<br />
-50<br />
-100<br />
-100 -50 0 50 100<br />
X[mm]<br />
Rintr =<br />
N_coinc_evts=100<br />
Z_projection with X= 0.00<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
<br />
R 2 meas − R 2 ρ − R 2 180<br />
(0.54 mm) 2<br />
Axial resolution<br />
Intersection of LOR with central plane<br />
no tomographic reconstruction Z_proj Z_proj !!!<br />
Entries Entries 20000 20000<br />
Mean Mean 0.1968 0.1968<br />
RMS RMS 1.104 1.104<br />
22<br />
! ! / / ndf ndf 53.9 53.9 / / 21 21<br />
Prob Prob 0.0001019 0.0001019<br />
Constant Constant 1125 1125 ± ± 11.7 11.7<br />
Mean Mean 0.1977 0.1977 ± ± 0.0059 0.0059<br />
Sigma Sigma 0.6485 0.6485 ± ± 0.0064 0.0064<br />
0<br />
-10 -5 0 5 10<br />
Z[mm]<br />
≈ 1.35 mm<br />
[0.0022 x Diam = 0.33 mm] 2<br />
2200<br />
2000<br />
1800<br />
1600<br />
1400<br />
(R_FWHM)z 1200 ~ 1.5 mm<br />
Y_projection with X= 0.0<br />
1000<br />
including :<br />
• intrinsic resolution 800<br />
• positron range<br />
600<br />
• non collinearity<br />
• (source dimensions 400 ; ø=250μm)<br />
∼<br />
200<br />
0<br />
-10 -5 0<br />
<br />
<br />
(R F W MH)z SingleMod<br />
<br />
(2)<br />
27
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
δp<br />
F2F<br />
OBL<br />
θ<br />
!"<br />
Parallax free demonstration<br />
parallax error is more<br />
and more important<br />
outside the center of<br />
the FOV<br />
intersection of LORs with the plane containing the source<br />
14000<br />
12000<br />
10000<br />
8000<br />
6000<br />
4000<br />
2000<br />
0<br />
160<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
3<br />
" 10<br />
Z_projection with X'= 0.00<br />
axial trans-axial<br />
Z_proj<br />
Entries 226472<br />
Mean 0.07255<br />
RMS 0.9407<br />
2<br />
! / ndf<br />
438 / 22<br />
Prob 0<br />
Constant 1.264e+04 ± 3.845e+01<br />
Mean 0.07968 ± 0.00172<br />
Sigma 0.6546 ± 0.0018<br />
-4 -2 0 2 4<br />
Z[mm]<br />
50000<br />
40000<br />
30000<br />
20000<br />
10000<br />
Y'_projection with X'= 0.00 and expected Y'= 0.00<br />
0<br />
Y_proj<br />
Entries 226472<br />
Mean -0.079<br />
RMS 0.8928<br />
-4 -2 0 2 4<br />
Y'[mm]<br />
Z_projection with X'= 0.00 Z_proj<br />
Y'_projection 3 with X'= 0.00 and expected Y'=-45.00<br />
" 10<br />
Y_proj<br />
Entries 2875919<br />
Mean 0.1839<br />
RMS 0.9481<br />
2<br />
! / ndf<br />
1540 / 21<br />
Prob 0<br />
Constant 1.62e+05 ± 1.39e+02<br />
Mean 0.1946 ± 0.0005<br />
Sigma 0.6431 ± 0.0005<br />
-4 -2 0 2 4<br />
Z[mm]<br />
350<br />
300<br />
250<br />
200<br />
150<br />
100<br />
50<br />
Entries 2875919<br />
Mean -44.65<br />
RMS 0.9086<br />
-49<br />
0<br />
-48 -47 -46 -45 -44 -43 -42 -41 -40<br />
Y'[mm]<br />
-50<br />
-50 -40 -30 -20 -10 0 10 20 30 40 50<br />
Y'[mm]<br />
Intrinsic resolution is not degraded by parallax effects, even in very oblique configuration !<br />
Z[mm]<br />
Z[mm]<br />
50<br />
40<br />
30<br />
20<br />
10<br />
0<br />
-10<br />
-20<br />
-30<br />
-40<br />
-30<br />
-40<br />
ZY'_projection with X'= 0.00 and expected Y'= 0.00<br />
X_proj<br />
Entries 226472<br />
Mean x<br />
Mean y<br />
-0.08777<br />
8000<br />
0.07623<br />
RMS x<br />
RMS y<br />
1.103<br />
1.818 7000<br />
axial transaxial<br />
F2F σ=0.655 RMS=0.893<br />
ZY'_projection with X'= 0.00 and expected Y'=-45.00<br />
X_proj<br />
50<br />
40<br />
30<br />
Entries<br />
Mean x<br />
Mean y<br />
RMS x<br />
RMS y<br />
2875919<br />
-44.61 45000<br />
0.1884<br />
1.115 40000<br />
1.835<br />
35000<br />
20<br />
10<br />
OBL 0 σ=0.643<br />
30000<br />
25000<br />
RMS=0.909<br />
-10<br />
20000<br />
-20<br />
15000<br />
-50<br />
-50 -40 -30 -20 -10 0 10<br />
Y'[mm]<br />
6000<br />
5000<br />
4000<br />
3000<br />
2000<br />
1000<br />
0<br />
10000<br />
5000<br />
0<br />
28
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
AX-PET is a fully simulated system<br />
Simulations<br />
Dedicated Monte Carlo simulations for AX-PET .<br />
Why? (1) get a better understanding of the detector<br />
(2) train Compton reconstruction algorithms<br />
(3) support image reconstruction<br />
(4) simulate an hypothetical full ring scanner<br />
and estimate its potential performance<br />
GATE simulation package (Geant4 application for tomographic emission)<br />
Non conventional nature of AX-PET<br />
=> several challenges for simulations (standard templates could not be applied)<br />
- geometry (axial; staggering; layered structure)<br />
- WLS modeling<br />
- dedicated sorter for the coincidences<br />
(module sum; trigger logic... up to DAQ rate modeling)<br />
29
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Simulations: some results<br />
Excellent agreement data / simulations :<br />
One AXPET Module<br />
illuminated by a collimated<br />
511 keV gamma beam :<br />
Data and Simulations<br />
Two AXPET Modules in coincidences<br />
!"#$%&'()%*%+,)%-($,(&$.'+-(/).0+)%-($$<br />
LYSO energy spectrum LYSO multiplicity<br />
1%**'.'()$%&'()%*%+,)%-($,23-.%)45/$,.'$)'/)'&$,(&$)4'%.$'**%+%'(+6$%($$!"#$<br />
.'+-(/).0+)%-($%/$'/)%5,)'&$-($/%502,)%-(/7<br />
! 82'%(9:%/4%(, ;,/'& -( 3'-5').6 -. '('.36<<br />
! =,>%505 ?('.36<<br />
! "-5@)-( 8%('5,)%+/ A"8B<<br />
! :'0.,2 :')C-.D7<br />
#%502,)%-($%/$@'.*-.5'&$;6$0/%(3$EFP$'('.36$.'/-20)%-($,)$QEE$D'ST$C%)4$@-%()92%D'$/-<br />
%($)4'$+'().'$-*$)4'$UVST$;,+D9)-9;,+D$3,55,$'5%//%-(T$F$5-&02'/$,)$WQ$55$&%/),(+<br />
• identify ICS events before image reconstruction.<br />
• several identification algorithms tested<br />
axial resolution from WLS strips Inter-Crystal Scattering (ICS)<br />
!"#$%& '()*+(,%-$ -./0,120340," 2/56". 2/+7(683<br />
OEP OQP9OOP OEP9OIP RQP<br />
• identification rate for ICS ~ 60%<br />
very incomplete collection of results here !<br />
simulations paper is in preparation !<br />
EFGHIGHJ #%502,)%-($,(&$%5,3'$.'+-(/).0+)%-($*-.$KL9M?N<br />
30
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Towards a tomographic reconstruction...<br />
How to mimic a full scanner with 2 modules only available?<br />
Central FOV :<br />
rotating the phantom...<br />
θ = 0˚, 20˚, 40˚... 180˚ (9 steps)<br />
1 tomographic acquisition<br />
= 27 steps acquisition<br />
φ = 0˚<br />
Extended FOV :<br />
...and rotating also the module<br />
φ = 20˚<br />
θ = 0˚, 20˚, 40˚... 360˚ (18 steps)<br />
mimics a 18-modules<br />
ring, with coincidences<br />
between face-to-face ±<br />
one adjacent modules<br />
31
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
FE<br />
G-APD bias<br />
Setup for tomographic reconstruction<br />
Module1<br />
trigger NIM crate<br />
source<br />
rotating<br />
table<br />
setup @ CERN, Blg 304<br />
Module2<br />
power supply box<br />
ADC VME crate<br />
FE<br />
The two modules are mounted on top of a portable<br />
platform, which houses also the electronics, power<br />
supply, etc...<br />
• One rotating motor for<br />
the source / phantom<br />
• One module fixed (Mod1);<br />
the other rotating (Mod2)<br />
32
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Measurements with phantoms<br />
Three different measurement campaigns :<br />
@ ETH Zurich (CH), Radio-Pharmaceutical Institute - Apr 2010<br />
@ AAA (Advanced Acceleration Applications) St. Genis-Pouilly (FR) - July 2010<br />
@ AAA (Advanced Acceleration Applications) St. Genis-Pouilly (FR) - July 2011<br />
in-situ cyclotron production of the radiotracer.<br />
- Phantoms filled with F-18 based tracers (F-18 in water solution)<br />
- t1/2 ~ 110 mins<br />
- Concentration diluted in water; A0: few MBq up to ~ 100 MBq<br />
- Analysis restricted to photoelectric absorption events only (no ICS events for the moment)<br />
33
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Measurements with phantoms<br />
Three different measurement campaigns :<br />
@ ETH Zurich (CH), Radio-Pharmaceutical Institute - Apr 2010<br />
@ AAA (Advanced Acceleration Applications) St. Genis-Pouilly (FR) - July 2010<br />
@ AAA (Advanced Acceleration Applications) St. Genis-Pouilly (FR) - July 2011<br />
in-situ cyclotron production of the radiotracer.<br />
- Phantoms filled with F-18 based tracers (F-18 in water solution)<br />
- t1/2 ~ 110 mins<br />
- Concentration diluted in water; A0: few MBq up to ~ 100 MBq<br />
- very first measurements with high activity and extended objects<br />
- limited to the central FOV (i.e. fixed modules, rotating source)<br />
- extended FOV coverage (one module rotating, rotating source)<br />
- larger phantoms / placed off-centered phantoms<br />
- extended FOV as before<br />
- improved DAQ performance<br />
- improved acquisition methods<br />
33
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
(1) capillaries<br />
L = 3cm ;<br />
Diam = 1.4 mm ;<br />
Pitch = 5 mm<br />
(2) micro Derenzo<br />
H = 1.5 cm ;<br />
Diam = 2 cm;<br />
Rods_Diam = 0.8÷2 mm<br />
(3) mini DeLuxe<br />
H = 5 cm ;<br />
Diam = 7.5 cm;<br />
Rods_Diam = 1.2÷4 mm<br />
Phantoms for reconstruction<br />
(4) homogeneous cylinder<br />
H = 9 cm ;<br />
Diam = 6 cm;<br />
(5) NEMA phantom<br />
H = 6.3 cm ;<br />
Diam = 3.3 cm;<br />
pictures not on scale!<br />
34
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Image reconstruction<br />
Goal :<br />
recover the activity distribution, starting from the acquired data<br />
Data :<br />
LOR of the various coincidence events i.e. projections<br />
(typically organized in “sinograms”)<br />
Angle θ<br />
Two different approaches :<br />
projections<br />
g(s,θ)<br />
Displacement s<br />
ANALYTICAL METHOD ITERATIVE METHOD<br />
• Filtered Back Projection (FBP) :<br />
1) Fourier analysis of the projection data<br />
2) Different weight to different frequencies (“filtering”)<br />
3) “Back-Project”<br />
• simple and fast<br />
☺ ☹<br />
• not extremely accurate<br />
• assumption : measured data are perfectly consistent with the<br />
source object (never true: e.g. noise, gaps between detector...)<br />
• slow and CPU consuming<br />
• accurate model of the emission and<br />
detection processes<br />
• accurate reconstruction<br />
• optimization procedure until the best<br />
estimate is found<br />
• several optimization strategies exist<br />
activity<br />
f(x,y)<br />
☹<br />
☺<br />
35
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Iterative reconstruction method<br />
36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Iterative reconstruction method<br />
(2)<br />
obtain the projection that would<br />
derive from such an object<br />
(1)<br />
initial estimate of the<br />
object count distribution<br />
(typically: uniform)<br />
36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Iterative reconstruction method<br />
(2)<br />
obtain the projection that would<br />
derive from such an object<br />
(1)<br />
initial estimate of the<br />
object count distribution<br />
(typically: uniform)<br />
36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Iterative reconstruction method<br />
(2)<br />
obtain the projection that would<br />
derive from such an object<br />
build up correction coefficients<br />
(1)<br />
initial estimate of the<br />
object count distribution<br />
(typically: uniform)<br />
36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Iterative reconstruction method<br />
36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Iterative reconstruction method<br />
requires a description of the<br />
physical model of emission<br />
and detection processes :<br />
SYSTEM MATRIX<br />
Mij<br />
probability of the activity<br />
in the j-voxel to be<br />
detected by the i-LOR<br />
includes :<br />
- geometry description<br />
- physics (e.g. : attenuation)<br />
- variable fraction of<br />
the voxel contributes to<br />
the counts<br />
- ...<br />
36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
out of the many possible iterative methods<br />
AX-PET uses ML-EM<br />
Max Likelihood<br />
Iterative reconstruction method<br />
Expectation<br />
Maximization<br />
two steps per iteration :<br />
(1) Expectation step : form the<br />
likelihood of any reconstructed image<br />
given the measured data<br />
(2) Maximization step : find the<br />
image with the greatest likelihood<br />
requires a description of the<br />
physical model of emission<br />
and detection processes :<br />
SYSTEM MATRIX<br />
Mij<br />
probability of the activity<br />
in the j-voxel to be<br />
detected by the i-LOR<br />
includes :<br />
- geometry description<br />
- physics (e.g. : attenuation)<br />
- variable fraction of<br />
the voxel contributes to<br />
the counts<br />
- ... 36
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
conclusion<br />
y<br />
y<br />
z<br />
z<br />
x<br />
Simple reconstructed images : capillaries<br />
x<br />
counts [a.u.]<br />
capillaries :<br />
L = 3 cm<br />
∅ = 1.4 mm<br />
pitch = 5 mm<br />
y<br />
y<br />
x<br />
x<br />
z<br />
z<br />
z<br />
z<br />
y<br />
y<br />
x<br />
x<br />
z<br />
z<br />
central FOV<br />
no modules rotation<br />
ETH 2010<br />
Profiling of the reconstructed capillaries (3 different measurements) and resolutions (FWHM) of<br />
reconstructed sources. The resolution still includes the capillary finite size (1.4 mm inner diameter).<br />
Profiling<br />
FWHM [mm]<br />
z<br />
z<br />
Resolution<br />
x<br />
x<br />
y<br />
y
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
extended FOV<br />
2nd module rotation<br />
NEMA phantom<br />
Three regions in the same phantom to<br />
address three different aspects<br />
Hot & Cold rods for contrast<br />
Homogeneous cylinder for assessing the<br />
ability to reconstruct<br />
homogeneous distributions<br />
Series of small rods for resolution<br />
{<br />
{<br />
{<br />
34 mm<br />
63 mm<br />
38
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
extended FOV<br />
2nd module rotation<br />
NEMA phantom uniformly filled - AAA 2010<br />
NEMA phantom hot / cold / warm - AAA 2011<br />
NEMA phantom<br />
Three regions in the same phantom to<br />
address three different aspects<br />
Hot & Cold rods for contrast<br />
Homogeneous cylinder for assessing the<br />
ability to reconstruct<br />
homogeneous distributions<br />
Series of small rods for resolution<br />
{<br />
{<br />
{<br />
34 mm<br />
2 mm<br />
63 mm<br />
38
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
extended FOV<br />
2nd module rotation<br />
NEMA phantom uniformly filled - AAA 2010<br />
NEMA phantom hot / cold / warm - AAA 2011<br />
NEMA phantom<br />
Three regions in the same phantom to<br />
address three different aspects<br />
Hot & Cold rods for contrast<br />
Homogeneous cylinder for assessing the<br />
ability to reconstruct<br />
homogeneous distributions<br />
Series of small rods for resolution<br />
{<br />
{<br />
{<br />
34 mm<br />
2 mm<br />
63 mm<br />
38
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
extended FOV<br />
2nd module rotation<br />
NEMA phantom uniformly filled - AAA 2010<br />
NEMA phantom hot / cold / warm - AAA 2011<br />
NEMA phantom<br />
Three regions in the same phantom to<br />
address three different aspects<br />
Hot & Cold rods for contrast<br />
Homogeneous cylinder for assessing the<br />
ability to reconstruct<br />
homogeneous distributions<br />
Series of small rods for resolution<br />
{<br />
{<br />
{<br />
34 mm<br />
different color scale !<br />
1 mm (i.e. < Resolution)<br />
5 mm<br />
4 mm<br />
3 mm<br />
2 mm<br />
Reconstructed 1 mm rod<br />
=> FWHM ~ 1.6 mm<br />
63 mm<br />
38
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
extended FOV<br />
2nd module rotation<br />
NEMA phantom uniformly filled - AAA 2010<br />
NEMA phantom hot / cold / warm - AAA 2011<br />
NEMA phantom<br />
Three regions in the same phantom to<br />
address three different aspects<br />
Hot & Cold rods for contrast<br />
Homogeneous cylinder for assessing the<br />
ability to reconstruct<br />
homogeneous distributions<br />
Series of small rods for resolution<br />
{<br />
{<br />
{<br />
34 mm<br />
different color scale !<br />
1 mm (i.e. < Resolution)<br />
5 mm<br />
4 mm<br />
3 mm<br />
2 mm<br />
Reconstructed 1 mm rod<br />
=> FWHM ~ 1.6 mm<br />
63 mm<br />
38
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Mini Deluxe phantom<br />
(A)<br />
Rods oriented parallel to Z axis<br />
Resolution phantom : Mini Deluxe<br />
extended FOV<br />
2nd module rotation<br />
AAA 2011<br />
(A) (B)<br />
39
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Mini Deluxe phantom<br />
(A)<br />
Rods oriented parallel to Z axis<br />
Resolution phantom : Mini Deluxe<br />
extended FOV<br />
2nd module rotation<br />
AAA 2011<br />
Rods oriented perpendicular to Z axis<br />
(B)<br />
(A) (B)<br />
Results presented in Valencia, IEEE 2011<br />
39
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
What’s next (1/3) ?<br />
40
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
What’s next (1/3) ?<br />
• improve the reconstruction algorithms<br />
• include ICS (Inter Crystal Scattering) events in the reconstruction<br />
preliminar!<br />
• ICS included:<br />
• only “triple” coinc. (1+2)<br />
• simplest weighting (50% ; 50%)<br />
• Simulated One-Pass List-mode<br />
reconstruction (SOPL)<br />
• No image deterioration when ICS<br />
events are included!<br />
40
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
What’s next (2/3) ?<br />
• full ring simulations:<br />
• performance of (hypothetical) full ring (e.g. resolution, sensitivity, NEC...)<br />
• two different approaches<br />
41
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
What’s next (3/3) ?<br />
• images of (probably dead) small animal<br />
• new measurement campaign<br />
• @ETH Zurich, Radiopharmaceutical Institute<br />
• Spring 2012 ?<br />
• F-18 (bone scan) or FDG ?<br />
Mouse bone scan,<br />
NanoSPECT (Bioscan)<br />
42
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
What’s next (3/3) ?<br />
• images of (probably dead) small animal<br />
• new measurement campaign<br />
• @ETH Zurich, Radiopharmaceutical Institute<br />
• Spring 2012 ?<br />
• F-18 (bone scan) or FDG ?<br />
• test digital Silicon Photomultipliers (from Philips)<br />
as alternative photodetectors<br />
• extremely good timing performance<br />
• GOAL : demonstrate the possibility<br />
a TOF-PET with the axial concept<br />
Mouse bone scan,<br />
NanoSPECT (Bioscan)<br />
42
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
Digital Silicon Photomultiplier (D-SiPM)<br />
D-SiPM :<br />
Single Photon Avalanche<br />
Photodiode (SPAD)<br />
integrated with<br />
conventional CMOS<br />
circuits on the same<br />
substrate.<br />
All SPADs (including<br />
their corresponding<br />
electronics) connected<br />
together<br />
- to photon counter<br />
- to TDC<br />
Thomas Frach,<br />
CERN Detector seminar, Oct 2011<br />
43
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
Digital Silicon Photomultiplier (D-SiPM)<br />
Advantages of D-SiPM for AX-PET:<br />
Intrinsically very fast photodetector (~ 50ps).<br />
Great potential for TOF-PET.<br />
Small size. High level of integration.<br />
Compactness.<br />
Interesting concept of simplifying the readout chain.<br />
Bias supply included.<br />
Early digitization of the cell output; integrated<br />
electronics on chip => low noise.<br />
Digital => Temperature and gain stability is less<br />
critical than in analogue SiPM.<br />
Possibility to disable cells with high Dark Count.<br />
MRI compatible.<br />
1 sensor<br />
full tile,<br />
64 sensors<br />
Technology Evaluation Kit (TEK) :<br />
MPPC :<br />
3200 cells<br />
3x3 mm 2<br />
- 4 tiles :<br />
DLS_6400 (x2 tiles) ; 6400 cells/sensor<br />
DLS_3200 (x2 tiles) ; 3200 cells/sensor<br />
- each tile : 8x8 sensors<br />
- 1 sensor : 3.9 x 3.3 mm 2<br />
received at CERN on Jan 12 th , 2012<br />
44
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
22 Na<br />
source<br />
• 2 LYSO scintillator crystals<br />
• non AX-PET standard<br />
• (2x2x12) mm 3 and (2x2x15) mm 3<br />
• wrapped with teflon<br />
• coupling done with optical grease<br />
• LYSO crystal placed in the center of the pixel<br />
(no precise mechanical alignment)<br />
D-SiPM: first preliminary results<br />
time difference between the two tiles for coincident events<br />
(coincidence window = 2 ns)<br />
#Events<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
0<br />
PRELIMINARY !!!<br />
DLS 3200<br />
T ° C<br />
=5<br />
Peltier<br />
Entries 10964<br />
Constant 930.5 ± 13.1<br />
Mean 0.2449 ± 0.0011<br />
Sigma 0.08534 ± 0.00123<br />
Time resolution = 200 ps FWHM<br />
-0.4 -0.2 0 0.2 0.4 0.6 0.8 1<br />
Δt<br />
[ns]<br />
FWHM ~ 200 ps<br />
(cutting on events at the photopeak)<br />
45
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
D-SiPM: first preliminary results<br />
Minimal AX-PET like set-up<br />
WLS strip<br />
(3 x 0.9 x 40 mm 3 )<br />
DLS_3200<br />
(one pixel enabled)<br />
22-Na source<br />
(~ 3 MBq)<br />
LYSO crystal<br />
(3 x 3 x100 mm 3 )<br />
DLS_3200<br />
(1/2 pixel enabled)<br />
46
intro axial concept AXPET simulations 800 image reconstructions<br />
Raw LYSO spectrum, Na-22 source<br />
1400<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
Raw spectrum<br />
600<br />
D-SiPM: first preliminary results<br />
400<br />
200<br />
511 keV<br />
0<br />
0 500 1000 1500 2000 2500 3000<br />
Nr photons<br />
Energy spectrum<br />
(calibrated, corrected<br />
for saturation)<br />
300<br />
250<br />
200<br />
150<br />
100<br />
PRELIMINARY !!!<br />
50<br />
Nr photons<br />
1000<br />
AXPET performance perspectives<br />
D-SiPM<br />
conclusion<br />
LYSO light yield<br />
1.27 MeV<br />
LYSO energy spectrum, Na-22 source<br />
2500<br />
2000<br />
1500<br />
1000<br />
500<br />
0<br />
0 500 1000 1500 2000 2500 3000<br />
Energy Calibration [0]*(1-TMath::Exp(-x*[1]/[0]))<br />
! / ndf<br />
5.868e-11 / 1<br />
p0 2767 ± 2.05e-05<br />
p1 4.453 ± 4.72e-08<br />
0<br />
0 200 400 600 800 1000 1200 1400<br />
300<br />
250<br />
200<br />
150<br />
100<br />
PRELIMINARY !!!<br />
0<br />
0 200 400 600 800 1000 1200 1400<br />
50<br />
ENERGY CALIBRATED SPECTRUM<br />
2<br />
Energy [keV]<br />
Entries<br />
Mean<br />
RMS 2<br />
! / ndf<br />
Constant<br />
Mean<br />
Sigma<br />
SpectTile4<br />
0<br />
0 200 400 600 800 1000 1200 1400<br />
Energy [keV]<br />
Energy [keV]<br />
R_FWHM ~ 12.3%<br />
47
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
• 2 tiles operated in coincidence<br />
Correlation<br />
LYSO / WLS response<br />
Still to do :<br />
- add more WLS strips<br />
- full WLS cluster<br />
- extract axial coordinate<br />
D-SiPM: first preliminary results<br />
LYSO Energy [keV]<br />
Minimal AX-PET Setup<br />
2000<br />
1800<br />
1600<br />
1400<br />
1200<br />
1000<br />
800<br />
600<br />
400<br />
200<br />
PRELIMINARY !!!<br />
0<br />
0 20 40 60 80 100 120 140 160 180 200<br />
WLS [Nr photons]<br />
48
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives<br />
D-SiPM<br />
conclusion<br />
two needed arrangements<br />
Questions to be answered :<br />
- light yield and ΔE/E from LYSO crystals ?<br />
- axial coordinate reconstruction: does it work ?<br />
- axial resolution through WLS readout ?<br />
- time resolution (long crystals) ?<br />
- cooling for DC reduction ?<br />
D-SiPM: Outlook<br />
AX-PET performance demonstration<br />
timing performance demonstration<br />
49
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Conclusions<br />
Axial concept for a PET scanner :<br />
i.e. long and axially oriented scintillation crystals<br />
Intrinsically parallax free system (DOI information directly from the axial geometry)<br />
Spatial resolution and sensitivity could both be optimized<br />
AX-PET implementation :<br />
3D spatial information of the photon interaction point with :<br />
matrix of LYSO crystals and WLS strips<br />
individual readout of each channel (Si-PM)<br />
Two modules built (i.e. AX-PET demonstrator)<br />
Energy resolution ~ 12% FWHM,@ 511 keV<br />
Spatial resolution ~ 1.35 mm FWHM<br />
(competitive with state of the art PET)<br />
Fully simulated device<br />
Simulations - fully validated on the demonstrator - will assess the final performance of an hypothetical full<br />
ring scanner. Flexible design: scalable in size / dimensions / nr. layers....<br />
=> flexibility in the final target of AX-PET (small animal PET / brain PET)<br />
AX-PET demonstrator :<br />
Extensively tested with sources and successfully used with phantoms !<br />
‣ calorimeter with tracking<br />
capabilities (granularity)<br />
‣ novelty as a PET detector :<br />
- geometry<br />
- WLS implementation<br />
- module sum / trigger<br />
- Compton scattering reconstruction<br />
Currently under test :<br />
Digital SiPM as alternative for photodetector for AX-PET (TOF capabilities)<br />
50
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Conclusions<br />
50
intro axial concept AXPET AXPET performance simulations image reconstructions perspectives conclusion<br />
Conclusions<br />
50
han s!<br />
51
AX-PET collaboration<br />
A. Braem, M. Heller, C. Joram, T. Schneider and J. Séguinot<br />
CERN, PH Department, CH-1211 Geneva, Switzerland<br />
V. Fanti<br />
Università e Sezione INFN di Cagliari, Italy.<br />
C. Casella, G. Dissertori, L. Djambazov, W. Lustermann, F. Nessi-Tedaldi, F. Pauss, D. Renker 1 , D. Schinzel 2<br />
ETH Zurich, CH-8092 Zurich, Switzerland<br />
1 Currently with Technical University München , D-80333 München, Germany<br />
2 Currently with Massachussetts Institute of Technology, Cambridge 02139-4307, USA.<br />
J.E. Gillam, J. F. Oliver, M. Rafecas, P. Solevi<br />
IFIC (CSIC / Universidad de Valencia), E-46071 Valencia, Spain<br />
R. De Leo, E. Nappi<br />
INFN, Sezione di Bari, I-70122 Bari, Italy<br />
E. Chesi, A. Rudge, P. Weilhammer<br />
Ohio State University, Columbus, Ohio 43210, USA<br />
E. Bolle, S. Stapnes<br />
University of Oslo, NO-0317 Oslo, Norway<br />
U. Ruotsalainen, U. Tuna<br />
Tampere University of Technology, FI-33100 Tampere, Finland<br />
Many thanks for the appreciated technical support from :<br />
Volker Commichau, Michael Droge, Hanspeter Von Gunten, Christian Haller,<br />
Urs Horisberger, Bozidar Panev (ETH Zurich)<br />
Jerome Bendotti, Bernard Cantin, Claude David, Niel Dixon, Didier Gabriele,<br />
Markus Joos, Miranda Van Stenis (CERN)<br />
52