Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
8/2/2017
1
Solid State Digital Photon Counting PET/CT Instrumentation and Technology
Jun Zhang, PhD, DABR
Assistant Professor
Wright Center of Innovation in Biomedical Imaging
Department of Radiology
The Ohio State University Wexner Medical Center
AAPM2017 Scientific Symposium: Emerging and New Generation PET: Instrumentation, Technology, Characteristics and Clinical Practice
Aug 2 Wednesday 10:45am – 11:15pm
Disclosure
Philips Healthcare (Cleveland, OH) provided the pre-commercial investigational PET/CT system (Vereos).
DPC PET/CT (Vereos, Philips Healthcare) at OSU
Pre-commercial release unit of digital photon counting PET/CT system (Vereos, Philips Healthcare) operational at The Ohio State University (PI: Michael V Knopp, MD, PhD) since October 2014. PET detector technology moving from PMT to solid state DPC is a technology leap.
8/2/2017
2
PMTs Light
Spreading
High Energy Photon
Positioning Logic
Digital One-to-One coupling
DPC tile
Scintillation photon
Analog Many-to-many coupling
Crystal
Analog PMT
Scintillation photons
Digital Photon Counter High Energy
Photon
Scintillator Crystals
From PMT to Solid State DPC
Vereos PET Detector Geometry
18
1
2
3
5
6
7
=
=
=
=
} 010101101
=
===
==
==
=
===
==
=
23,040 x 3,200
Biggest contribution to the overall sensor DCR is caused by only a small percentage of the cells such as ~10 % cells responsible for 70-80% DCR.
Disabling high DCR cells on small active sensor area leads to significantly reduced overall sensor DCR.
1. Philips Vereos PET/CT Whitepaper
2. Haemisch Y, et al. Physics Procedia. 2012;37:1546–1560
DPC Data Acquisition Sequence
Digital Photon Counting
Digital Time of Flight
1:1 Coupling
8/2/2017
3
Energy Resolution
0.0%
2.0%
4.0%
6.0%
8.0%
10.0%
12.0%
14.0%
16.0%
18.0%
20.0%
Energy resolution (%)
9.5
10
10.5
11
11.5
12
FWHM=11.2%
460 480 500 520 540 560 580
Energy [keV]
En
erg
y R
es
olu
tio
n
[% F
WH
M]
7
8
9
10
11
12
13
14
15
Ener
gy R
eso
luti
on
(% F
WH
M)A B
-400 -200 0 200 400
Timing [ps]
FWHM = 312 ps
-400 -200 0 200 400
Timing [ps]
324ps FWHM
321
321.5
322
322.5
323
323.5
324
324.5
325
Tim
ing
Re
so
luti
on
[ps
FW
HM
]
322 ps FWHM
310
312
314
316
318
320
322
324
326
328
330
Timing Resolution (ps)
Axial FOV
18F18F
22Na point source
0 20 40 60 80
02
00
40
06
00
80
01
00
0
Avg Singles Rate [Mcps]
Tim
ing
Re
so
luti
on
[ps
FW
HM
]
Vereos PET/CTGemini TF (ref 7)
Experimental setup (ref 7)
JNM 48(2007) 471
TOF Timing Resolution
NEMA NU 2 2012 Spatial Resolution
8/2/2017
4
FWHM FWTM
at 1cm at 10cm at 20cm at 1cm at 10cm at 20cm
Axial 4.31 4.79 6.17 8.51 9.48 10.82
Transverse (Radial) 4.01 4.64 5.86 8.37 8.97 10.46
Transverse (Tangential) 4.01 4.41 4.99 8.37 8.98 10.37
NEMA NU 2 2012 Spatial Resolution
where R is the overall spatial resolution, Kr – a scale factor, Ri the intrinsic resolution determined by crystal width, Rp the error due to positron range, Ra the error from noncollinearity, and Rl is the decoding error (anger logic positioning
localization). For DPC PET, Rl is zero due to that 1:1 coupling eliminates need for position decoding.
𝑅(𝐹𝑊𝐻𝑀) = 𝐾𝑟 × 𝑅𝑖2 + 𝑅𝑝
2 + 𝑅𝑎2 + 𝑅𝑙
2
1. WW Moses. Nucl Instrum Methods Phys Res A. 2011; 648 (Supplement 1): S236–S240
2. Jun Zhang, Michael V Knopp. J Nucl Med 2017; 58 (supplement 1): 1322
= 𝐾𝑟 × 𝑅𝑖2 + 𝑅𝑝
2 + 𝑅𝑎2
Compared to conventional photo-multiplier PET systems, spatial resolution of the SiPM digital photon counting PET system is NOT affected by A. Crystal size B. Positron Range C. Noncollinearity / Acollinearity D. Localization decoding
SAM Question
Answer: (D) 1:1 coupling eliminates need for position decoding
0 10 20 30 40 50 60 70 800
50
100
150
Slice Number
Se
ns
itiv
ity c
ps
/MB
q
0cm10cm
A B
0 2 4 6 8 10
05
10
15
20
25
Activity at Measurement Time [mCi]
Eff
ec
tiv
e S
en
sit
ivit
y [k
cp
s/M
Bq
]
Vereos PETAnalog PET Courtesy of Philips
NEMA NU 2 2012 Sensitivity
Average at 0 and 10cm: ~5.6 kcps/MBq
0 10 20 30 40 50 60 70 800
50
100
150
Slice Number
Se
ns
itiv
ity c
ps
/MB
q
0cm10cm
A B
0 2 4 6 8 10
05
10
15
20
25
Activity at Measurement Time [mCi]
Eff
ec
tiv
e S
en
sit
ivit
y [k
cp
s/M
Bq
]
Vereos PETAnalog PET
Axial Sensitivity Profile
𝐒𝐞𝐟𝐟(𝐀) = 𝐒𝐍𝐄𝐌𝐀(𝐀) × 𝑮𝐓𝐎𝐅
𝑮𝐓𝐎𝐅 =𝟐𝑫
𝒄𝚫𝐭= time-of-flight gain
Budinger. JNM 24(1): 73-78
NEMA PET Sensitivity Phantom (Data Spectrum Corporation, Hillsborough, NC)
8/2/2017
5
Activity Concentration (kBq/mL)
NEC
R (
kcp
s)
0
500
1000
1500
2000
2500
0 10 20 30 40 50
Ra
te (
kcp
s)
Activity Concentration (kBq/mL)
Total Randoms TRUE Scatter
0
50
100
150
200
0 10 20 30 40 50R
ate
(k
cps)
Activity Concentration (kBq/mL)
NECR Scatter Fraction
A B
30%
0 10 20 30 40 50 60 70
05
01
00
15
02
00
Activity Concentration [kBq/mL]
NE
CR
(1r)
[kcp
s]
OSU System
3 x V1
V1, 27cm
V1, 35 cm
Peak NECR = 157 kcps @ 51 kBq/mL TOF peak NECR = 641 kcps @ 51kBq/mL
NEMA NU 2 2012 NECR and Scatter
0 10 20 30 40 50 60 70
02
00
40
06
00
80
0
Activity Concentration [kBq/mL]
TO
F E
ffe
ctive
NE
CR
(1r)
[kcp
s]
OSU System
3 x V1
TOF
Effe
ctiv
e N
ECR
(kcp
s)
Measured TOF Gain - OSU = 325 ps G = 4.1 V1 = 316 ps G = 4.2 Analog ~ 500 ps 2.7
NECR = T2 / (T + S + R)
For a 48cm diameter object, an estimated sensitivity gain and the SNR improvement of the digital photon counting TOF PET (320ps, 4.8cm uncertainty) compared to nonTOF PET are about A. 5.0 and 5.0 B. 10.0 and 10.0 C. 10.0 and 3.2 D. 10.0 and 5.0
SAM Question
Answer: (C) 𝑮𝐓𝐎𝐅 =
𝟐𝑫
𝒄𝚫𝐭=
𝑫
𝚫𝒙 = time-of-flight gain
G SNR =
𝑫
𝚫𝒙
40.7
59.1 63.070.1
81.9 86.3
0
20
40
60
80
100
10 13 17 22 28 37
Co
ntr
ast
(%)
Sphere (mm)
9.2
7.45.8
4.63.6
2.7
0
2
4
6
8
10
12
10 13 17 22 28 37
Bac
kgro
un
d V
aria
bili
ty
(%)
Sphere (mm)
NEMA NU 2 2012 Image Quality
8/2/2017
6
Phantoms IQ
39
89
8.5
2.8
10 37
%
SPHERE SIZE (MM)
Sphere Contrast (%) Background Variability (%)
0.0000.1000.2000.3000.4000.5000.6000.7000.8000.9001.0001.100
2 5 8 11 14 17 20 23 26 29 32 35 38 41 44 47 50 53 56 59 62 65 68 71 74 77 80
SUV
me
an
Slice #
Alpha-Bravo Uniformity SUV Axial Profile - Brain 256FOV 2.2035mCi F18 @ 8:45am on 09/19/2014, phantom on the table, acq started at
10:49am, SUVmax=0.98, SUVmin=0.89, SUVmean=0.97, % variance=8.76%
LBR= 16
LBR= 32
Vereos Gemini
Specifications Vereos Gemini TF 64
Detector Design Direct Photon Counting PMT Anger Logic
Ring Diameter (mm) 764 764
Crystal LYSO LYSO
Crystal Size (mm3) 4 x 4 x 19 4 x 4 x 22
# of Detector Elements 23,040 420 PMT
Crystal-Detector Coupling 1:1 multi : multi
Transaxial FOV (mm) 256 / 576 / 676 256 / 576 / 676
Axial FOV (mm) 164 180
Overlap in Axial FOV 39% 53%
Energy resolution (%) 11 12.3
Energy calibration 1:1 multi : multi
Timing Resolution (ps) 320 550
Timing Calibration 1:1 multi : multi
Coincidence window (ns) 2.0 / 4.0 / 4.6 2.0 / 4.0 / 4.6
Reconstruction Algorithm 3D OSEM TOF 3D OSEM TOF, 3D Ramla
Reconstruction Matrix (pixel) up to 676 x 676 up to 338 x 338
Reconstruction Voxel (mm3) 64 / 8 / 1 64 / 8
# Voxels per slice up to 456,976 up to 107,584
Resolution Recovery PSF No PSF
System Specifications: DPC vs PMT
PMT PET vs DPC PET
Vereos Gemini
8/2/2017
7
Non-TOF 325ps TOF
Non-TOF TOF Non-TOF TOF
550ps 4mm PET 325ps 4mm PET
PMT PET vs DPC PET
PET Image Resolution (Recon)
Spacing (mm) 4 x 4 x 4
Volume (mm3) 64
FOV = 576mm 144 x 144 matrix
FOV = 676mm 169 x 169 matrix
FOV = 256mm N/A
# Voxels of V N (e.g., 20,736)
PSF & Filter on/off
Recon 3D TOF OSEM
2 x 2 x 2
8
288 x 288 matrix
338 x 338 matrix
128 x 128 matrix
8N (e.g., 165,888)
on/off
3D TOF OSEM
1 x 1 x 1
1
576 x 576 matrix
676 x 676 matrix
256 x 256 matrix
64N (e.g., 1,327,104)
on/off
3D TOF OSEM
4 4
4
2 2
2
1 1 1
4x4x4 mm3 2x2x2 mm3 1x1x1 mm3
Small Lesion Detectability with Less PVE
8/2/2017
8
Vereos (1mm3) Gemini (64mm3)
Lesion Detectability with Less PVE
A B C
D E F
A B C
D E F
A B C
D E F
A B C
D E F
A B C D E F
PET/CT Gemini TF 64 Gemini TF 64 mCT mCT Vereos Vereos
XY resolution 2mm 2mm 1mm 1mm 2mm 1mm
Slice Thickness 2mm 2mm 3mm 3mm 2mm 1mm
FDG Injected 4.3mCi 5.0mCi 10.0mCi 12.0mCi 13.9mCi 13.9mCi
Uptake time 45min 45min 50min 49min 118min 118min
Acquisition Time 10min 10min 10min 10min 90sec 90sec
A B C D E F
PET/CT Gemini TF 64 Gemini TF 64 mCT mCT Vereos Vereos
XY resolution 2mm 2mm 1mm 1mm 2mm 1mm
Slice Thickness 2mm 2mm 3mm 3mm 2mm 1mm
FDG Injected 4.3mCi 5.0mCi 10.0mCi 12.0mCi 13.9mCi 13.9mCi
Uptake time 45min 45min 50min 49min 118min 118min
Acquisition Time 10min 10min 10min 10min 90sec 90sec
A B C D E F
PET/CT Gemini TF 64 Gemini TF 64 mCT mCT Vereos Vereos
XY resolution 2mm 2mm 1mm 1mm 2mm 1mm
Slice Thickness 2mm 2mm 3mm 3mm 2mm 1mm
FDG Injected 4.3mCi 5.0mCi 10.0mCi 12.0mCi 13.9mCi 13.9mCi
Uptake time 45min 45min 50min 49min 118min 118min
Acquisition Time 10min 10min 10min 10min 90sec 90sec
Vendor A
A B C D E F
PET/CT Gemini TF 64 Gemini TF 64 mCT mCT Vereos Vereos
XY resolution 2mm 2mm 1mm 1mm 2mm 1mm
Slice Thickness 2mm 2mm 3mm 3mm 2mm 1mm
FDG Injected 4.3mCi 5.0mCi 10.0mCi 12.0mCi 13.9mCi 13.9mCi
Uptake time 45min 45min 50min 49min 118min 118min
Acquisition Time 10min 10min 10min 10min 90sec 90sec
Vendor B
Brain PET
90s/bed FDG
2mm Brain of 256FOV
Wholebody 576FOV
4mm Brain of WB 2mm Brain of WB
Brain PET Redefinition – Speed, Convenience, Dose, IQ
10min FDG
1mm Brain of WB
8/2/2017
9
WB PET Redefinition
30s/bed 60s/bed 75s/bed 90s/bed
WB NaF PET
120s/bed 120s/bed 120s/bed
12s/bed 12s/bed 12s/bed
Clinical Ultrafast PET
90s/bed 15min
30s/bed 5min
9s/bed 1.5min
8/2/2017
10
Low Dose FDG PET
NIH R01 PIs: Knopp, Zhang
Clinical Post 90Y Microsphere Radioembolization
Bremsstrahlung SPECT/CT DPC PET
A – 10min/bed; B - 7min/bed
A
B
8/2/2017
11
39% (9bed) 27% (8bed) 20% (7bed) 0% (6bed)
2.5mCi; NaF; 90s/bed; 110cm
39% 27% 20% 0%
• 1:1 coupling eliminates need for position decoding • Timing and energy calibrations per crystal • Improved timing resolution • Reduced dead time • Increased count rate performance • Early digitization eliminates as much as possible
analog noise and distortion.
Highlights
8/2/2017
12
18
1
2
3
5
6
7
18
1
2
3
5
6
7
18
1
2
3
5
6
7
18
1
2
3
5
6
7
18
1
2
3
5
6
7
18
1
2
3
5
6
7
A B C D E F
How much SiPM DPC Detector Coverage Do We Really Need?
A B C
D E F
Full ring detectors
50% detector reduction
33% detector reduction
25% detector reduction
40% tile-ring reduction
50% detector reduction
18
1
2
3
5
6
7
Clinical Evaluation
• This research was supported by the Ohio Third Frontier OSDA TECH 09-028 (PI: Michael V Knopp, MD PhD) and NCI R01CA195513 (PI: Michael V Knopp, MD PhD; Jun Zhang, PhD)
• Philips Healthcare provided the pre-commercial investigational PET/CT system (Vereos) hardware and technical support
• Wright Center of Innovation in Biomedical Imaging (PI: Michael V Knopp, MD PhD)
Acknowledgement
8/2/2017
13
Thank You