1. Imaging molecolare ad alta risoluzione spaziale ed alta efficienza
F. Garibaldi CV INFN
- perche’
- come
cosa occorre
richiesta

2. Functional Molecular Imaging Large collaboration is needed
What is needed
Submillimetric spatial resolution
-High efficiency
collimation is a key parameter
standard parallel hole collimator
pinhole
multipinhole
Small animal imaging, important area of bio-medical research
- studying new radiopharmaceuticals
- animal such mice serve as models for many studies (neural function, coronary diseases, cancer, stem cells etc.)
Some example
Challenging detectors
Models of bone dismetabolism(osteoporosys)(h)
Drug metabolism (m)
Brain tumors ( for example, neuroblastoma in children) (h)
Studying specific vs aspecific uptake of radiopharmaceuticals (m)
Studying Annexin V peptide (it is taken up in apopotosys) (m)
Brain activity measurements (m)
Stem cells (d,w)
Techniques
- PET intrinsic limitations
- expensive
- Single Photon Emission
simpler technique
tradeoff spatial resolution vs sensitivity
and FOV
complementarity
Large collaboration is needed
(physicist and M.D.)

3. changing L,d Collimator better res. small FOV lower eff. Pinhole
PET
Thin Line
Collimator
Thin Cone
Pinhole
Coded Aperture
Thin Cones
Compton
Cone Surface
better res. small FOV
lower eff. w
- results similar to pinhole
higher efficiency
> L < d
changing L,d
efficiency vs sp.res.
0 – 20 mm
1 – %

4. parallel hole pin hole spatial resolution efficiency g = de2/16xb2
a = pin hole - detector distance
b = pin hole - object distance
d = hole aperture
a = pin hole angle
efficiency
g = de2/16xb2

5. R3292 (5 inch) or H8500 (Flat panel)
FOV=20 mm
FOV=10 mm
R3292 (5 inch) or H8500 (Flat panel)

6. Scintillator arrays

7. starting point simple desktop detector Understanding limitations
- spatial resolution
- sensitivity
what can be improved ?
intrinsic detector performances
FOV
sensitivity
.pinhole collimator
.array od pixellated scintillator (NaI(Tl))(1.25x1.25x5 mm3) and (1.8 x1.8 x6 mm3)
. PSPMT (R2486 (3”)
(Hamamatsu)
Not independent

9. H8500
H9500
3 mm
6 mm

10. Comparison pin-hole parallel hole collimator
NaI(Tl), 1.8 x 1.8 mm pixel size
57Co source 1 mm diameter at 5 mm distance
140 keV high resolution
parallel hole collimator
FWHM= 2.8 mm
pin hole collimator
I = 3
FWHM = 1.7 mm
efficiency
- Parallel hole : ~ 15.9 counts/mCixs
- Pinhole : ~ 3 counts/mCixs
(1mm aperture)
“only” a factor ~ 5

11. Resolution doesn’t improve, pixels identification not so good
NaI(Tl), 1.8 x 1.8x 5 mm3
Pinhole aperture : 1mm
Source diameter: ~ 1.0 mm
Pinhole aperture: 0.67 mm
d= 17 mm (I=3)
FWHM = 1.1 mm
FWHM = 1.7 mm
d = 7 mm (I=~7)
FWHM = 1.3 mm
NaI(Tl), 1.25x1.25x 5 mm3
I = 7
FWHM = 1.3 mm
Resolution doesn’t improve, pixels identification not so good
--> photodetector limitation

12. Let’s improve the pixel identification
(better sampling at anode level (M16 (4 x 4 mm2 and M64 (2x2 mm2))
Source diameter: ~ 1.0 mm, I ~ 7
NaI(Tl), 1.8 mm pixel size, M64
NaI(Tl), 1.8 mm pixel size, M16
1.0 mm FWHM
1.1 mm FWHM
NaI(Tl), 1.25 mm pixel size, M64
NaI(Tl), 1.25 mm pixel size, M16
1.1 mm FWHM
1.0 mm FWHM

13. CsI(Tl) arrays Hamamatsu PSPMT’s C8,M16,M64 (different anode)
4.2x4.2 mm2
2.5x2.5 mm2
1.5x1.5 mm2

14. Improving sampling -> better pixel identification
(more pixel in the image)
M16, 1.8 mm pixel
R2486, 1.8 mm pixel
M64, 1.8 mm pixel
M16, 1.2 mm pixel
M64, 1.25 mm pixel
R2486, 1.2 mm pixel

15. small anode pixel ->better sampling ->better performances
- pin hole aperture dominates the spatial resolution
- apertures = 0.5 mm (or 0.3 mm) would improve spatial resolution but lowering counting rates
small anode pixel ->better sampling ->better performances
M16 and M64 have small area (~20 x 20 mm2)
arrays possible solution
but dead area --> not the best
H8500 (50 x 50 mm2, 64 channels (anode sampling
6 x 6 mm2)
- H9500 (50 x 50 mm2, 256 channels (anode sampling
3x3 mm2) available

17. catene resistive vs multiwire

18. To be done - fixing FOV according to the particular research
- fixing the area (50 x 50 mm2 to 100 x 100 mm2)
seems to best solution
--->
FOV = ~ 15 x 15 mm2 to 50 x 50 mm2
(according to the detector area and magnification)
- maximizing the number of pixel
- trying to improve sensitivity
- reading out all the channels

19. How
Pin hole (tungsten) collimator (0.3,0.5,0.7,1) mm
better photodetector
NaI 1.8 x 1.8 mm2 *(and 1.25 x 1.25 mm2)-H channels(6x6 mm2 anode pixel)
Improving detector peformances (more pixel, better indentification, improved spatial resolution)
NaI x 1.25 mm H channels (3x3 mm2 anode pixel)
Improving the efficiency
- bigger detector area
- better collimation (coded aperture)
Next step: smaller scintillator pixel size? CsI (Tl) (or Na? ) or NaI(Tl)

20. (Pin hole 0.7 mm tungsten, H8500 64 ch) Flood Field irradiation
New Detector
(Pin hole 0.7 mm tungsten, H ch)
(1.8 x 1.8 mm2) (1.25 x 1.25 mm2)
Good pixel identification For 1.8 x 1.8 not so good for 1.25 x 1. 25
-> better anode sampling is needed --> H channels
Flood Field irradiation
Eg = 122 keV

22. Read-out electronics

23. High resolution preserving high SNR ? coded apertures
Ideal pinhole
+perfect resolution
-zero transmitted power
Real pinhole
+some signal through
-degraded resolution
Coded Aperture
+signal of finite pinhole
+resolution of ideal pinhole

24. Coded apertures I (Image) = O (object) x A (aperture) A  G = d then
Figure adapted from:Fenimore and Cannon, Optical Engineering, 19, 3, , 1980.
Example of apertures with
known decoding pattern
I (Image) = O (object) x A (aperture)
There are decoding patterns G allowing:
A  G = d then
A  G = Ô, in fact
Ô = R  G = ( O × A )  G = O * (A  G) = O * PSF

25. coded aperture collimators
simulation for our desktop detector
10Ci in 10 s
4444 pixels
1.25 x 1.25 mm2
FoV 22 cm2.
Mask NTHT MURA 2222, =2, 1% transparent, thickness 1.5 mm W. Pitch 0.68 mm.
Line source
10Ci in 10 s
2D source
10Ci in 10 s
sensitivity improved by a factor 30!
coded aperture collimators

26. Submillimeter spatial resolution
Reconstruction of a 122 keV point-like source using the coded apertures
Submillimeter spatial resolution
High sensitivity
(factor ~ 30)
FWHM=0.93 mm
Sensitivity=145 cps /MBq
5 cps/MBq with pinhole

27. Image of a mouse head
Top view of a mouse injected with 3 mCi of 99Tc-MDP (image time 25 min; FOV = 16 x16 mm2)

28. Conclusioni Rilevanza Perche’ Come Cosa occorre
Imaging funzionale, mediante radionuclidi ad alta risoluzione ed efficienza, signle photon (planare e tomografico)
Come
Scintillatore(i) pixellato, fototubo(i), DAQ “veloce” per ~ 1000 ch, collimatori pinhole e/o aperture codificate
Cosa occorre