Professor Brian F Hutton

1
Emission Tomography Principles and Reconstruction
Professor Brian F Hutton Institute of Nuclear
Medicine University College London brian.hutton_at_uc
lh.nhs.uk
2
Outline

• imaging in nuclear medicine
• basic principles of SPECT
• basic principles of PET
• factors affecting emission tomography

3
History

• Anger camera 1958
• Positron counting, Brownell 1966
• Tomo reconstruction Kuhl Edwards 1968
• First rotating SPECT camera 1976
• PET Ter-Pogossian, Phelps 1975

4
Anger gamma camera
Detector 400x500mm 9mm thick Energy
resn 10 Intrinsic resn 3-4mm
Radionuclides Tc-99m 140keV, 6hr I -123
159keV, 13hr Ga-68 93-296keV, 3.3dy I-131
360keV, 8dy
Collimator Designed to suit energy HR
hole size 1.4mm length 33mm septa 0.15mm
5
Organ-specific options specialized collimators
for standard cameras
parallel
fanbeam
conebeam
slit-slat
crossed slit
pinhole
6
Single Photon Emission Computed Tomography
(SPECT)
Single Photon Emission Computed Tomography
(SPECT)

• relatively low resolution long acquisition time
  (movement)
• noisy images due to random nature of radioactive
  decay
• tracer remains in body for 24hrs radiation
  dose standard x-ray
• function rather than anatomy

7
SPECT Reconstruction
sinogram for each transaxial slice
Filtered back projection
1 angle
2 angles
4 angles
16 angles
128 angles
8
Organ-specific systems specialised system
designs, with use limited to a specific
application
9
Positron Annihilation
10
Coincidence Detection
No Collimator
detector 1
coincidence window
detector 2
time (ns)
11
PET "Block" Detector
Scintillator array
PMTs
C
BGO (bismuth germanate)
A
B
Histogram
Images courtesy of CTI
12
Attenuation Correction in PET
attenuation for activity in body N N0 e -?x. e
-? (D-x) N0 e -?D attenuation for external
source N N0 e -?D (Dbody thickness)
'Exact' attenuation correction
(for 511 keV ? 0.096/cm attenuation factors
25-50)
13
Coincidence Lines of Response (LoR)
sinogram
parallel
fanbeam
14
PET Reconstruction
sinogram
1 angle
2 angles
4 angles
16 angles
128 angles

• conventional filtered back projection
• iterative reconstruction

15
Understanding iterative reconstruction
detector (measurement)
Y

• Objective
• Find the activity distribution whose estimated
  projections match the measurements.
• Modelling the system (system matrix)
• What is the probability that a photon emitted
  from location X will be detected at detector
  location Y.
• detector geometry, collimators
• attenuation
• scatter, randoms

m
X
estimated projection
object
Y1
m
X
Y2
16
System matrix
pixeli
0 0 0 0 0 0 0 1 0 0
0 0 0 0 0 1 0 0 0 0
0
0
0
0
0
0
1
0
0
0
voxelj
17
ML-EM reconstruction
BP
original estimate
NO CHANGE
update (x ratio)
original projections
patient
FP
current estimate
estimated projections
18
Image courtesy of Bettinardi et al, Milan
19
Noise control

• stop at an early iteration
• use of smoothing between iterations
• post-reconstruction smoothing
• penalise rough solutions (MAP)
• use correct and complete system model

20
Factors affecting quantification
courtesy Ben Tsui, John Hopkins
21
detector
-

without attenuation correction
transmission
with attenuation correction
22
System matrix with attenuation
0 0 0 0 0 0 0 0.2 0 0
0 0 0 0 0 0.5 0 0 0 0
0
0
0
0
0
0
0.9
0
0
0
m
23
Partial volume effects

• effect of resolution and/or motion
• problems for both PET and SPECT
• similar approaches to correction
• scale of problem different due to resolution
• some different motion effects due to timing
• ring versus rotating planar detector

24
Modelling resolution

• Gamma camera resolution
• depends on distance
• SPECT resolution
• need radius of rotation
• PET resolution
• position dependent

25
System matrix including resolution model
0 0 0 0 0 0 0.1 0.2 0.1 0
0 0 0 0 0.2 0.5 0.2 0 0 0
0
0
0
0
0
0.3
0.9
0.3
0
0
m
26
PET resolution
depth of interaction results in asymmetric point
spread function
radial int
radial ext
tangential
27
Modelling resolution
detector (projection)

• potentially improves resolution
• requires many iterations
• slow to compute

m

• stabilises solution
• better noise properties

object
w/o resn model
Courtesy Panin et al IEEE Trans Med Imaging
2006 25907-921
with resn model
28
Can we consider measurements to be quantitative?

• Scatter correction
• multiple energy windows for SPECT PETCT
  standard models
• SPECT local effects PET more distributed

detector
object

• Scatter fraction
• SPECT 35 PET 2D 15 3D 40

29
Scatter
Monte Carlo

• influenced by photon energy, source location,
  scatter medium
• reduces contrast

measured

• scatter models
• analytical, Monte Carlo, approximate models
• measurement
• triple energy window (TEW), multi-energy
• subtract from projections
• measured proj TEW
• or combine with projector in reconstruction
• compare (forward proj TEW) with measured proj

30
3D reconstruction

• Approaches
• rebin data followed by 2D reconstruction
• single slice rebinning (SSRB)
• multi-slice rebinning (MSRB)
• Fourier rebinning (FORE)
• full 3D reconstruction
• 3D OSEM
• 3D RAMLA

limits for FORE
31
VUE Point 3D-OSEM 28subsets 2iter
FORE 2D-OSEM 28subsets 5 iter
FORE 2D-OSEM 28subsets 2 iter
Courtesy V Bettinardi, M Gilardi, Milan
32
Summary

• Emission tomography
• functional rather than anatomical
• single photon versus dual photon (PET)
• main difference is collimation
• Iterative reconstruction
• very similar approach for SPECT and PET
• currently most popular is OSEM (or similar)
• the better the system model the better the
  reconstruction