nuclear medicine: tomographic imaging – spect, spect-ct and pet-ct
DESCRIPTION
Nuclear Medicine: Tomographic Imaging – SPECT, SPECT-CT and PET-CT. Katrina Cockburn Nuclear Medicine Physicist. Image Acquisition Techniques. Static- (Bones, Lungs) Dynamic- (Renography) Gated- (Cardiac) Tomography SPECT PET List Mode- (Cardiac). - PowerPoint PPT PresentationTRANSCRIPT
Nuclear Medicine: Nuclear Medicine: Tomographic Imaging – Tomographic Imaging –
SPECT, SPECT-CT and PET-SPECT, SPECT-CT and PET-CTCT
Katrina CockburnKatrina Cockburn
Nuclear Medicine PhysicistNuclear Medicine Physicist
Image Acquisition TechniquesImage Acquisition Techniques
StaticStatic -- (Bones, Lungs)(Bones, Lungs) DynamicDynamic -- (Renography)(Renography) GatedGated -- (Cardiac)(Cardiac) TomographyTomography
SPECTSPECT PETPET
List ModeList Mode -- (Cardiac)(Cardiac)
Problems with Planar ImagingProblems with Planar Imaging
Planar imagingPlanar imaging 2D representation of 3D 2D representation of 3D
Distribution of activityDistribution of activity No depth informationNo depth information Structures at different Structures at different
depths are superimposeddepths are superimposedLoss of contrastLoss of contrast
Problems with Planar ImagingProblems with Planar Imaging
Planar imagingPlanar imaging 2D representation of 3D Distribution of activity2D representation of 3D Distribution of activity No depth informationNo depth information Structures at different depths are superimposedStructures at different depths are superimposed
Loss of contrastLoss of contrast
Problems with Planar ImagingProblems with Planar Imaging
Planar imagingPlanar imaging 2D representation of 3D 2D representation of 3D
Distribution of activityDistribution of activity No depth informationNo depth information Structures at different Structures at different
depths are superimposeddepths are superimposedLoss of contrastLoss of contrast
Problems with Planar ImagingProblems with Planar Imaging
Planar imagingPlanar imaging 2D representation of 3D 2D representation of 3D
Distribution of activityDistribution of activity No depth informationNo depth information Structures at different Structures at different
depths are superimposeddepths are superimposedLoss of contrastLoss of contrast
Problems with Planar ImagingProblems with Planar Imaging
Planar imagingPlanar imaging 2D representation of 3D 2D representation of 3D
Distribution of activityDistribution of activity No depth informationNo depth information Structures at different Structures at different
depths are superimposeddepths are superimposedLoss of contrastLoss of contrast
Problems with Planar ImagingProblems with Planar Imaging
Planar imagingPlanar imaging 2D representation of 3D 2D representation of 3D
Distribution of activityDistribution of activity No depth informationNo depth information Structures at different Structures at different
depths are superimposeddepths are superimposedLoss of contrastLoss of contrast
Image contrast 2:1
Object Contrast 4:1
Single Photon Emission Computed Single Photon Emission Computed TomographyTomography
Collect multiple planar Collect multiple planar images at several images at several angles around the angles around the patientpatient Typically 64-128 views Typically 64-128 views
over 360over 360°° Can be 32-64 views Can be 32-64 views
over 180over 180°°
Single Photon Emission Computed Single Photon Emission Computed TomographyTomography
Image ReconstructionImage Reconstruction 2D images of selected 2D images of selected
planes within the 3D objectplanes within the 3D object Better ContrastBetter Contrast Lower Spatial ResolutionLower Spatial Resolution
Normal reconstruction Normal reconstruction techniques are Filtered techniques are Filtered Back Projection or Back Projection or Iterative ReconstructionIterative Reconstruction
Back ProjectionBack Projection
Back Project Back Project each planar each planar image onto image onto three three dimensional dimensional image matriximage matrix
3 3
3
3
3
3
33
6
6 6
6
Back ProjectionBack Projection
Back Project Back Project each planar each planar image onto image onto three three dimensional dimensional image matriximage matrix
3 36
1
1
1
1
1
1
2
2
2
Back ProjectionBack Projection
Back Project Back Project each planar each planar image onto image onto three three dimensional dimensional image matriximage matrix
3 36
1
1
1
1
1
1
2
2
2
3
3
6
22 3 2
3 4 3
2 3 2
Back ProjectionBack Projection
Back Project Back Project each planar each planar image onto image onto three three dimensional dimensional image matriximage matrix
3 3
3
3
3
3
33
6
6 6
6
4 4
4 4
6 6
6
6
8
Back ProjectionBack Projection
Back Project Back Project each planar each planar image onto image onto three three dimensional dimensional image matriximage matrix
3 3
3
3
3
3
33
6
6 6
6
4 4
4 4
6 6
6
6
8
Back ProjectionBack Projection
More views – More views – better better reconstructionreconstruction
1/r blurring, even 1/r blurring, even with infinite number with infinite number of viewsof views
Filtered Back ProjectionFiltered Back Projection
Filter planar views Filter planar views prior to back prior to back projectionprojection
Correction of 1/r Correction of 1/r blurring requires blurring requires ‘Ramp’ Filter‘Ramp’ Filter Gives increasing weight to Gives increasing weight to
higher spatial frequencieshigher spatial frequencies Amplifies NoiseAmplifies Noise
SPECT FIlters
-0.2
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
Frequency (fraction of Nyquist)
Ramp
Shepp-Logan
ModifiedShepp_Logan
Hanning
Hamming
Butterworth
Filtered Back ProjectionFiltered Back Projection
In PracticeIn Practice Use modifications Use modifications
of Ramp Filterof Ramp Filter Compromise Compromise
between Noise between Noise and Spatial and Spatial ResolutionResolution
SPECT FIlters
-0.2
0
0.2
0.4
0.6
0.8
1
0 0.2 0.4 0.6 0.8 1
Frequency (fraction of Nyquist)
Ramp
Shepp-Logan
ModifiedShepp_Logan
Hanning
Hamming
Butterworth
Modified Ramp FilterModified Ramp Filter
Multiplication of the Multiplication of the ramp filter by another ramp filter by another functionfunction Often a gaussian Often a gaussian
shape shape Width of the gaussian Width of the gaussian
affects the “roll off” of affects the “roll off” of the rampthe ramp
Filtered Back ProjectionFiltered Back Projection
Problems with Filtered Back Problems with Filtered Back ProjectionProjection
Back projection is mathematically correct, Back projection is mathematically correct, but real life images require Filtered Back but real life images require Filtered Back ProjectionProjection
Back Projection can introduce noise and Back Projection can introduce noise and streaking artefactsstreaking artefacts Not good with attenuation correctionNot good with attenuation correction
Filtered Back Projection can reduce noise Filtered Back Projection can reduce noise and artefacts, but may degrade resolutionand artefacts, but may degrade resolution
Iterative ReconstructionIterative Reconstruction
NOT a new techniqueNOT a new technique Pre-dates Filtered Back ProjectionPre-dates Filtered Back Projection Computationally IntensiveComputationally Intensive
Long Reconstruction TimesLong Reconstruction Times Requires fast computers for reconstructionRequires fast computers for reconstruction Takes around 1 min for a 16-frame gated 128 Takes around 1 min for a 16-frame gated 128
x 128 matrix cardiac scanx 128 matrix cardiac scan
What is Iterative Reconstruction?What is Iterative Reconstruction?
Iteration is process of successively better Iteration is process of successively better “guesses”“guesses”
The image processing computer creates The image processing computer creates an image by refining the expected an image by refining the expected projections in comparison to those projections in comparison to those recordedrecorded
This form of IR is known as “Maximum This form of IR is known as “Maximum Likelyhood Expectation Maximisation” Likelyhood Expectation Maximisation” (MLEM)(MLEM)
MLEM AlgorithmMLEM Algorithm
Benefits of IRBenefits of IR
More accurate More accurate modelling of modelling of emission/detectionemission/detection
Can include Can include attenuation attenuation correction and other correction and other information from information from MR, CT etcMR, CT etc
Lower noiseLower noise
Image FusionImage Fusion
““Unclear Medicine” Unclear Medicine” images can be images can be registered to CTregistered to CT Reduces attenuation Reduces attenuation
artefactsartefacts Allows localisation of Allows localisation of
“fuzzy blob” images“fuzzy blob” images Can improve Can improve
diagnostic accuracydiagnostic accuracy
Attenuation CorrectionAttenuation Correction
X-Ray imaging essentially provides an X-Ray imaging essentially provides an attenuation “map”attenuation “map” Images formed by different attenuation Images formed by different attenuation
patterns patterns NM imaging does not need attenuationNM imaging does not need attenuation
In fact do not want it!In fact do not want it! Hybrid imaging (e.g.SPECT-CT) takes Hybrid imaging (e.g.SPECT-CT) takes
attenuation map of CT images and uses to attenuation map of CT images and uses to correct for attenuation in 3D NM images correct for attenuation in 3D NM images
““Jordan”Jordan”
6 x 500ml saline bags 6 x 500ml saline bags strapped to torso strapped to torso phantom (3 each phantom (3 each side) to simulate side) to simulate breast attenuationbreast attenuation
Positioned to cover Positioned to cover anterior LVanterior LV
Normal Perfusion: “Jordan”Normal Perfusion: “Jordan”
FBP FBPSC IR IRSC IRAC IRACSC
Resolution RecoveryResolution Recovery
Resolution worsens Resolution worsens with increasing with increasing distance from the distance from the collimatorcollimator
If we can model how If we can model how this happens, we can this happens, we can build this into our build this into our Iterative projectionsIterative projections
Resolution RecoveryResolution Recovery
Better modelling Better modelling means better means better imagesimages
Fewer counts Fewer counts needed to get needed to get acceptable acceptable imagesimages Shorter Shorter
acquisitionsacquisitions Lower dosesLower doses
NM Imaging: The PET CameraNM Imaging: The PET Camera
PET camera invented in the 1970sPET camera invented in the 1970s Positron Camera 1959Positron Camera 1959
Early positron study (1953)Early positron study (1953)
Why use positron emitters?Why use positron emitters?
Many of the positron emitters occur in biological Many of the positron emitters occur in biological molecules (C, N, O, etc.)molecules (C, N, O, etc.)
Many have small molecular weights relative to Many have small molecular weights relative to the biological molecules they may be used to the biological molecules they may be used to label (e.g., F) even if they aren’t found there label (e.g., F) even if they aren’t found there naturally.naturally.
PET isotopes can be attached to biologically PET isotopes can be attached to biologically interesting molecules with no or minimal impact interesting molecules with no or minimal impact on the behaviour of those molecules in the body.on the behaviour of those molecules in the body.
Positron Emission TomographyPositron Emission Tomography
PET isotopes emit positrons rather than PET isotopes emit positrons rather than gamma raysgamma rays
Coincidence ImagingCoincidence Imaging Better Spatial Resolution (Typically 4mm)Better Spatial Resolution (Typically 4mm)
Requires Dedicated EquipmentRequires Dedicated Equipment Limited AvailabilityLimited Availability
AnnihilationAnnihilationconservation of momentum:
before: system at rest; momentum ~ 0
after: two photons created; must have same energy and travel in opposite direction.
conservation of energy
before: 2 electrons, each with a rest mass of 511keV
after: 2 photons, each with 511keV.
decay via positronemission
electron/positronannihilation
annihilationphoton
annihilationphoton
detector
Coincidence ImagingCoincidence Imaging
line of response(LOR)
Detector ArrayDetector Array
Coincidence Imaging - Detector Coincidence Imaging - Detector RingRing
Types of Coincidence EventTypes of Coincidence Event
2D to 3D Imaging2D to 3D Imaging
Stack multiple rings behind each otherStack multiple rings behind each other Allows for true 3D imaging Allows for true 3D imaging Shorter imaging time so better throughput Shorter imaging time so better throughput
and fewer motion artefactsand fewer motion artefacts
Time of Flight (TOF) PETTime of Flight (TOF) PET
Because we are “timing” Because we are “timing” the arrivals of the photons, the arrivals of the photons, we can tell how far apart we can tell how far apart they arethey are All photons travel at the All photons travel at the
speed of lightspeed of light Simultaneous equation to Simultaneous equation to
work out point of originwork out point of origin Makes “line of response” Makes “line of response”
more like a pointmore like a point
PET Camera CrystalsPET Camera Crystals
NaI has too poor stopping power for 511keVNaI has too poor stopping power for 511keV BGO is main material usedBGO is main material used Siemens patented LSOSiemens patented LSO*this table was provided by Siemens…*this table was provided by Siemens…