a time driven readout scheme for pet and ct… …using apds and sipms
DESCRIPTION
A Time Driven Readout Scheme For PET and CT… …using APDs and SiPMs. NINO. NINO. NINO. NINO. PADJK. PADJK. PADJK. PADJK. APD Array. APD Array. APD Array. APD/SiPM. LSO. LSO. LSO. LSO. Fig. 5: Functional diagram of the photon detection and signal processing layout. - PowerPoint PPT PresentationTRANSCRIPT
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 11
A Time Driven Readout Scheme For PET and
CT…
…using APDs and SiPMs
F. Powolny1), E. Auffray1), G. Condorelli2), S. Brunner1), M. Despeisse1), G. Fallica2), H. Hillemanns1), P. Jarron1), A. Kluge1), P. Lecoq1), M. Mazzillo2), T. C. Meyer1,3), M. Morel1), D. Sanfillipo2), G. Valvo2)
1) CERN, European Organization for Nuclear Research, 1211 Geneva 23, Switzerland, 2) STMicroelectronics, R&D DSG, 95121 Catania, Italy, 3) Presenter
Outline of Presentation
• Why a new readout technique?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 2
2005
Outline of Presentation
• Why a new readout technique?• The “Time Based” approach.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 3
2005
2006
Outline of Presentation
• Why a new readout technique?• The “Time Based” approach.• APD detector & TB readout.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 4
2005
2006
2007
Outline of Presentation
• Why a new readout technique?• The “Time Based” approach.• APD detector & TB readout.• TOF capability?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 5
2005
2006
2007
2008
Outline of Presentation
• Why a new readout technique?• The “Time Based” approach.• APD detector & TB readout.• TOF capability?• SiPMs and TB readout.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 6
2005
2006
2007
2008
2009
Outline of Presentation
• Why a new readout technique?• The “Time Based” approach.• APD detector & TB readout.• TOF capability?• SiPMs and TB readout.• Summary.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 7
2005
2006
2007
2008
2009
Why a New Readout?
• Clinical goals:– Multimodality: CT, PET, MRI, US, ...
• Reduce patient exposure time;• Compensate for patient/organ motion;• Facilitate image fusion (PET/CT);• Simultaneous imaging of tumor response and responsiveness, as well as dose
delivery in vivo.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 8
PHILIPS GEMINI TF™
Why a New Readout?
• Clinical goals:– Multimodality: CT, PET, MRI, US, ...
• Reduce patient exposure time;• Compensate for patient/organ motion;• Facilitate image fusion (PET/CT);• Simultaneous imaging of tumor response and responsiveness, as well as dose
delivery in vivo.
• Technical goals (“From HEP to PET”):– Use technologies & techniques devel’d for HEP
• State-of-the-art electronics;• Compact and reliable data processing;• System integration and cost.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 9
PHILIPS GEMINI TF™
CERN CMS
Why “Time” Based?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 10
• Leads to simple readout architectures:• Only digital pulses simple electronics⇒• No flash ADCs (running at > 500MHz, power)
• Simple timing circuits:• Need only discriminators and TDCs;
• Derive time and energy information from one digital pulse.
• Time over threshold (T.o.T.), pulse width modulation;
• Build on in-house experience from large experiments.
Time Based Readout: How?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 11
i(t)
V(t)
t
Threshold
t
P.A. Discr.FEDC05 NINO
i(t)
V(t)
Time Based Readout: How?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 12
i(t)
V(t)
t
Threshold
t
Time Walk
P.A. Discr.FEDC05 NINOLSO-like Test Pulse
i(t)
V(t)
Time Based Readout: How?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 13
i(t)
V(t)
t
Threshold
t
Time Walk
P.A. Discr.FEDC05 NINOLSO-like Test Pulse
i(t)
V(t)
Time Based Readout: How?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 14
i(t)
V(t)
t
Threshold
t
Time Walk
Pulse Width
Timestamp from leading edge after time walk correction.Photon energy from falling edge (pulse width).Need only discriminator and TDC for both.
Time Walk Correction
Non-Linearity Correction
P.A. Discr.FEDC05 NINOLSO-like Test Pulse
i(t)
V(t)
APD Detector w/ LHC-Electronics
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 15
V(t)i(t)APD
i(t)P.A. Discr.
Hamamatsu S8550 FEDC05 (ATLAS) NINO (ALICE - TOF)
• FEDC05 chip:– Preamplifier developed for ATLAS
Silicon Tracker– 400 electrons ENC (r.m.s.)– Gain = 4mV/fC– 16 channels– 13ns peaking time
• NINO chip:– Very fast discriminator (3GHz BW)– Signal peaking time: <1ns– Output time jitter: ≤ 25ps– 8 channels– Power consumption: 27mW/ch.
22Na-Source
(2 x 2 x 10 mm3)
SIXTH FRAMEWORK PROGRAMME PRIORITY 1
FP6-2002 LIFESCIHEALTHProposal 505785
The NINO Chip
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 16
x6 x6x6x6
Vout +
Vout -
Input
stage
threshold
TH+ TH-
4 Cascaded amplifiers Output buffer
Out1 +
Out1 -
Out A1 +
Out A1 -
Out A2 +
Out A2 -
Out A3 +
Out A3 -
Out A4 +
Out A4 -
In +
In -
Out1+Out1 -
Vdd
M5 M0
M4 M1
M3 M2
M6
M7
In+ In-BiasN1
BiasN3BiasN2
RR
Inth1 Inth2
Vdd
M6 M7
M8
InA+ InA-
OutA+ OutA-
BiasN5
Cascade amplifier
RR
Out A1 +
Out A1 -
Out A2 +
Out A2 -
Out A3 +
Out A3 -
Out A4 +
Out A4 -
time [s]5ns 10ns 0
time [s]5ns 10ns 0
time [s]5ns 10ns 0
time [s]5ns 10ns 0
2.60
2.30
2.00
1.70
1.40
2.60
2.30
2.00
1.70
1.40
2.50
2.30
2.10
1.90
1.50
1.70
2.20
2.10
2.00
1.90
“SPICE” of Readout Chain
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 17
LSO-APD Pulses
FEDC05 Output
NINO output (+)
V(t)i(t)APD
i(t)P.A. Discr.
Hamamatsu S8550 FEDC05 (ATLAS) NINO (ALICE - TOF)
Time Stamp Energy
Energy Resolution & Dynamic Range With APDs & NINO
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 18
176Lu Background
122keV
57Co
176Lu Background
122keV 22Na
511keV
1275keV
511keV
1275keV
Raw data[Pulse Width]
Corrected data[Charge/Energy]
Look-up Table
Linearity after correction
57Co
22Na137Cs
22Na
Energy resolution
Timing Resolution With APDs & NINO
• How “fast” are APDs in the time encoded ‘NINO’ setup?
• What are the limitations?• What are the alternatives?• The road to TOF-PET…
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 19
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 20
Timing Resolution With PMT (Reference)
470ps FWHMNo corrections applied;CFD is free of Time Walk.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 21
Timing Resolution With PMT–APD
470ps FWHMPhoto-peak SelectionAnd
Time Walk Correction on APD Setup
1180ps FWHM
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 22
Timing Resolution With Dual APD
1180ps FWHM
470ps FWHMPhoto-peak SelectionAnd
Time Walk Correction on both APD Setups
1600ps FWHM
Raw data
After photo-peak selection
+ Time walk correction
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 23
TOTAL Crystal APD Electronics
480 340 230 220
1130 800 540 517
100 50 30 20€
σTotal = σ crystal2 + σ APD
2 + σ electr2
1. Front end electronics noise together with the APD dark current contribute ~ 20% to the total time jitter.
[ps] rms
[ps] FWHM[%]
Time Jitter of Detector and Readout
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 24
€
σTotal = σ crystal2 + σ APD
2 + σ electr2
1. Front end electronics noise together with the APD dark current contribute ~ 20% to the total time jitter.
2. Non-uniformities in avalanche amplification in the APD account for ~ 30%.
TOTAL Crystal APD Electronics
480 340 230 220
1130 800 540 517
100 50 30 20
[ps] rms
[ps] FWHM[%]
Time Jitter of Detector and Readout
€
N = R 1− e−
tN
τ ⎛
⎝ ⎜
⎞
⎠ ⎟
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 25
€
σTotal = σ crystal2 + σ APD
2 + σ electr2
1. Front end electronics noise together with the APD dark current contribute ~ 20% to the total time jitter.
2. Non-uniformities in avalanche amplification in the APD account for ~ 30%.
3. The principal jitter component of the total jitter is attributed to a) the Poisson-like photon production in LSO within the crystal decay
time of 40ns, and b) the high threshold sensitivity of the APD being N ~ 20 p.e.
(R = 2200 p.e., tN = 340ps, t = 40ns)
TOTAL Crystal APD Electronics
480 340 230 220
1130 800 540 517
100 50 30 20
[ps] rms
[ps] FWHM[%]
Time Jitter of Detector and Readout
TOF in PET: Why?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 26
From HEP we know:• Event patterns congested by background;• “Space” points help to remove confusion and
improve reconstruction efficiency; • Charge division, Stereo view, delay lines, cathode
readout are known methods;In PET similar problems arise:
• Count rate contaminated with scattered and random photons;
• TOF reduces randoms and increases sensitivity.
€
NECR = T
1+ SR
+ 2RT
⎛ ⎝ ⎜
⎞ ⎠ ⎟
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
0 300 600 900 1200 1500 1800
TOF timing (psec)
SNR squared
Data courtesy of J. S. Karp, IEEE, Trans. Med. Imag. Vol. 10(D denotes patient diameter)
D = 27cm
D = 35cmLesion Detectability
APDSiPM
“S”catter“R”andom“T”rue
w/o TOF with TOF
LOR (z)
“Space” coordinate from t2 – t1
100ps 15mm
t1
t2
€
Δz = t2 − t12
× c
Conventional Back-Projection
g
g g
g
SiPMs for TOF-PET?
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 27
SiPMs (MPPCs, etc.)…• combine the advantages of conventional PMTs: fast and high gain, and• those of solid state devices: compact, insensitive to magn. fields, low cost;• are sensitive to single photo-electrons;• are binary devices (photon counting)
But have …• low fill factors;• high thermal (dark current-) count rates;• high terminal capacitance.
Pixel structure of a ST-Microelectronics 1 x 1mm2 SiPM*) SPAD = Single Photon Avalanche Diode
SiPM pulse degradation for several terminal capacitances
SPAD*)
Time Response With Laser Pulses
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 28
Non-commercial 1 x 1mm2 (400 pixels) SiPM test structure of STM – Catania:
Laser pulse: 50ps, 405nm
Test Setup
σ1 = 181psσ2 = 130psσ3 = 109ps
Dark noise (1p.e.)
“Cross talk” (2p.e.)Laser events
Del
ay [n
s]D
elay
[ns]
Pulse height [µA]
Pulse height [µA]Noise
1 SPAD
2 SPADs
3 SPADs4 SPADs
FWHM:425ps (1p.e.)306ps (2p.e.)256ps (3p.e.)
€
σ jitter =σ ele
2 + σ SPAD2
dI /dt
≈ σ SPAD
dI /dt
= 4.6μA35μA /1ns
=130ps
σ n = n ×σ SPAD
n × dI /dt=
σ jitter
n
Time Jitter vs. Np.e.
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 29
€
σ jitter =σ ele
2 + σ SPAD2
dI /dt
≈ σ SPAD
dI /dt
= 4.6μA35μA /1ns
=130ps
σ n = n ×σ SPAD
n × dI /dt=
σ jitter
n
Laser Timing With NINO
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 30
Without time walk corrections:
Use look-up table for correcting data.
Non-commercial 1 x 1mm2 (400 pixels) SiPM test structure of STM – Catania:
Laser Timing With NINO
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 31
NINO has little if any influence on timing precision.
FWHM:430ps (1p.e.)320ps (2p.e.)275ps (3p.e.)221ps (4p.e.)
Non-commercial 1 x 1mm2 (400 pixels) SiPM test structure of STM – Catania:
With time walk corrections:
SiPM: Energy Resolution
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 32
NINO spectrum normalized from pulse width to energy(algorithm derived from SPICE simulations).
LSO 3 x 3 x 20 mm3 crystal
Energy x 105 [eV]
SiPM output
FWHM: 30%
NINO output
FWHM: 29%
Energy x 106 [eV]
22Na-Source
(3 x 3 x 20 mm3)
Hamamatsu S 10931-050 P: 3 x 3mm2, 3600 pixels
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 33
22Na SiPMHV
NINO
LSO 3x3x20 mm3
PMT
CFD
Hamamatsu S 10931-050 P: 3 x 3mm2, 3600 pixels
SiPM: Timing Resolution
Preliminary conclusion: • SiPM timing resolution
better than that of fast PMTs.
• SiPM resolution still contaminated by jitter from time walk (corrections to come).• Time walk in energy window, expected from SPICE, to be σ ~ 120ps (280ps FWHM)
€
σ = σPMT2 + σ SiPM
2 = 251psσ PMT =185psσ SiPM =170ps
SiPM 71V
Selection from 90ns to 110ns
Width [ns]
Del
ay [n
s]
After photopeakselection:
FWHMSiPM = 400ps
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 34
Summary
• Experience from BioCare (FP6) Work:– Validation of Time based Readout Scheme (patented).– Understand limitations in sensor performance (PMTs & APDs).– Timing resolution intrinsically limited by photon statistics and
insufficient APD gain.
• CERN time-based electronics well adapted to SiPMs:– Simple architecture around NINO discriminator (no additional
amplification needed);– Single detector time resolution of ≤400ps FWHM (σ = 170ps) achieved.– Time walk corrections (~280ps FWHM) still pending;– Resolution intrinsically limited by photon velocity in crystal (~200µm/ps)
100ps maximum in 20mm 50ps FWHM in chosen LSO crystal.
“Timing Precision” vs. Np.e. (Q)
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 35
(Westcott, Knoll, Lynch, Wright)
APD
(σ).34
Time Evolution of SiPM Signal
2009 Pisa Meeting on Instrumentation La Biodola, Italy – May 24 - 30, 2009
Thomas C. Meyer/CERN-PH 36
Io = 34 μANpe = 2000τcrystal = 40 nsτRC = 5 ns
I SiPM
(t) [u
A]
Time [ns]
1 p.e. response