Introduction to the ClearPEM projects
Joao VarelaLIP, LisbonOn behalf of the ClearPEM Collaboration
Jornadas LIP Universidade do Minho, 7-9 January 2010
The ClearPEM ProjectsHigh performance PET scanner for breast cancer detection. Scanner is in clinical tests.
Other applications to brain imaging and PET animal are being pursued
Projects developed by the ClearPEM consortium, in the framework of Crystal Clear at CERN
IP licensed to PETsys, SA
The ClearPEM scanner
Good spatial resolution ( ~1.5 mm in whole FoV)• Fine crystal segmentation (2x2 mm) • DoI measurement with good resolution (FWHM ~2 mm)
• Dual APD readout of individual crystal pixels
High Sensitivity ( ~ 40 cps/kBq in center FoV)• Long LYSO crystals (20 mm)• Two detector plates with large active area
Reduced Random Background (~30%)• Large flux of single photons (up to 10 MHz)
• Coincidence time resolution of ~4 ns FWHM
The ClearPEM scanner
Two detector plates:
• 6144 LYSO 2x2x20 mm3 crystals
• 12288 APD pixel channels
• Dual readout of crystal pixels for DoI measurement
•160x180 mm2 active area
•Water cooling
•75% of detector channels presently installed
The ClearPEM detector
Pulse Shape•Amplifier rise time: 20ns
•Variation of baseline and pulse shape 1-2%
Noise•Pedestal RMS = 2.2 ADC Counts = 5keV
•ENC = 1300 e- r.m.s.
•Inter-channel dispersion ~ 8%
ASIC Performance
Response to test pulse
Noise measurement
Trigger Performance• Events in coincidence up to 1.5MHz (This involves computation of energy
and time, Compton grouping and transmission to the trigger processor)
• Acquisition rate up to 0.8MHz (This involves readout of the event
dataframe after issuing a trigger)
• Disk storage rate ~ 400MB/s
• High performance useful for fast calibration runs
• Needed for other PET applications
Data Acquisition Performance
Ancillary Systems
Cooling• Water cooling of detector plates at 18 oC
• Stability of temperature ± 0.1 oC
Bias Voltages• Regulation of APD bias voltages on the
detector heads (64 regulation channels)
• Long-term stability of HV ~ 28 mV rms
40 hoursr.m.s. ~28 mV
Energy measurements:
• Average energy resolution at 511 keV for the full scanner is 16.0%
• Dispersion of energy resolution of individual crystals is 8.8%
• Good energy linearity (~ up to few percent)
Energy Resolution
Resolution ~12.5%
137Cs
Na-22 spectra summed for all crystals
Photopeak measurements
Time measurement:
• Photon time is extracted from the pulse samples fitted by the function:
• Coincidence time resolution of the whole scanner is 5.2 ns FWHM
Time Resolution
Resolution ~12.5%
137Cs
Typical pulse sampling
Time difference in coincidence events
All scannercoincidences
50 MHz sampling
Asymmetry distributionsfor different impact points in the crystal
DoI Resolution
Energy in bottom vs top APDs
Depth of interaction measurement
• DoI is measured from light asymmetry in crystal dual readout
• From direct measurement (collimated photons):
DoI resolution ~ 2 mm FWHM for light asymmetry ±40%
• Lu-176 background in crystals is used for DoI calibration in whole scanner
• Average light asymmetry is ±59%
Energy is independent of DoI:
E[keV]=Kabs.(Etop+Krel.Ebottom)
Z[mm]=CDOI.((Etopkrel.Ebottom)/(Etop+krel.Ebottom))
• Photopeak position is independent of DoI (up to few percent)
• Energy resolution does not depend on DoI
Energy Dependence on DoI
Resolution ~12.5%
137Cs
Photopeak position as a function of DoI
Energy resolution as a function of DoI
Energy:• Dispersion of channel gains 15.3%
Depth of interaction (Lu-176 background):• Dispersion of DoI calibration constants 7.8%
Time calibration:• Two pulse shape parameters per channel• Dispersion of time calibration constants 2%
Scanner Calibration
~4600 crystals
Distribution of energy calibrations
Distribution of pulse peak time
~4600 crystals
14
Sensitivity measurement:
• Na-22 source A(22Na) = 2.73 µCi ± 0.3% (101kBq)
• Sensitivity at center of FoV for 10 cm Detector Head opening is 1.0%
(350-700 keV, 20 ns)
• Correction factors:• Incomplete (75%) detector: 1.3• In-detector Comptons: ~2
• Corrected sensitivity: ~2.6%
• Monte Carlo estimation: 3.1%
Count rate scan
Sensitivity (100 mm ): 1.0%
ClearPEM Sensitivity
ClearPEM Spatial Resolution
5 mm
1 mm
Point source imaging• Na-22 point source in a grid with 5mm pitch• Energy window 400-600 keV• Sinograms of 16 source positions are added• Reconstruction with 3D-OSEM / STIR
Spatial resolution• Transaxial 1.2 mm FWHM(corrected for source size ~1mm)
DoI effect• Images without using DoI information show
considerable blurring
With DoI
Without DoI
ClearPEM Image Uniformity
Images of uniform Ge-68 source
• Reconstruction with 4 orientations of the detector plates
• Absorption, scatter and corrections are not applied
• Image artifacts due to detector effects are corrected
•Good image uniformity
Cylinder filled with positron emitter Ge-68
EW=400-700keVTW=4 ns6 iterations 3D-OSEM
Scanner Installation• Hospital IPO, Porto
Phase 1 (present)• Patients indicated for PET/CT (other
disease)• Negative breast exams• Tuning the image reconstruction with
real cases• 11 exams done
Phase 2• Patients with positive indication from
x-rays mammography• Assessment of PEM sensitivity /
specificity • Comparison to mammography and MRI
Clinical Tests Program
ClearPEM scanner at IPO Porto
Initial clinical examsExample of typical exam:• dose 8 mCi• 150 mm detector plate opening• 4 angular orientations• coincidence window ±4 ns• energy window 400-650 keV• fraction of randoms in FoV is 35%
Reconstruction:• 3D-OSEM• randoms, attenuation and scatter
corrections not applied• simple normalization correction • on-going work to reduce statistical noise
Dualmodal PET – US
CERIMED , Hospital Marseille, other partners
Ultra-sound probe with elastography capabilities
Cross-reference system and PET-US image fusion
Construction of second ClearPEM machine is well advanced
ClearPEM and Ultrasound
ClearPEM-Sonic
Detectormodules
Cooling plates
35 cm
60 cm
4.5 cm
Version with DoI
PET Animal
Pre-clinical studies with small and medium size animals
PET/MR insert Texas Institute of Preclinical Studies Application to brain imaging /EPFL
High performance PET
Improvements to Electronics/DAQ• Optical S-link (under test)• ASICv4 (in design)• Intelligent Front End Board (in project)• PET Ring Trigger (hw ready)
PET Time-of-Flight (FWHM ~200 ps)• SiPM and TOF ASIC (collaboration with Torino and CERN)
• SPAD single-photon detectors integrated with TDCs
(collaboration with TUDelft and CERN)
E. Albuquerque1, F. G. Almeida2,13, P. Almeida3, E. Auffray10, J. Barbosa2, A. L. Bastos9, V. Bexiga1, R. Bugalho4, C. Cardoso4, S. Carmona8, J.F. Carneiro2, B. Carriço4, C. S. Ferreira4, N. C. Ferreira5, M.
Ferreira4, M. Frade4, F. Gonçalves1, C. Guerreiro5, P. Lecoq10, C. Leong1, P. Lousã6, P. Machado1, M. V. Martins3, N. Matela3, R. Moura4, J. Neves4, P. Neves6, N. Oliveira3, C. Ortigão4, F. Piedade6, J. F. Pinheiro4, P. Relvas6, A. Rivetti , P. Rodrigues4, I. Rolo4, A. I. Santos8, J. Santos2, M. M. Silva1, S.
Tavernier11, I. C. Teixeira1,9, J. P. Teixeira1,9, J. C. Silva4,10, R. Silva4, A. Trindade4, J. Varela4, 12
1 INESC-ID, 2 INEGI, 3 IBEB/FCUL, 4 LIP, 5 IBILI/FMUC, 6 INOV, 8 HGO, 9 IPO, 10CERN, 11VUB
The ClearPEM Collaboration
Funded by
CERN