PET

Question 1

3 steps to PET

Answer 1

Administer tracer
Image acquisition
Image reconstruction

Question 2

PET tracers

Answer 2

18F-FDG
18F-DOPA
11C-choline
11C-methionine
11C-acetate
68Ga-DATA-NOC

Question 3

Comparison to Gamma camera

Answer 3

Full ring of detectors rather than floating heads
No collimator - uses coincidence
Rarely stand alone - often compatible with CT or MR
Often use step and shoot approach - have number of bed positions and time per bed position and overlap (although system exists that uses continuous bed motion)

Question 4

LOR

Answer 4

Line of response - determines relationship between annihilation location and detection. PET more sensitive because collimators usually absorb a lot of emissions

Question 5

Types of event

Answer 5

True coincidence
Scattered coincidence
Random coincidence
Attenuation

Question 6

Block detectors

Answer 6

Scanners have multiple rings
Detector rings typically have many blocks.
Scintillator crystal attached to 4 PMTs
Cuts in crystal act as light guides
Interaction position determined by ratio of signal in each PMT
System could have 10000-20000 crystals 4-6mm in size with only 1000 PMTs

Question 7

What do we want from a scintillator?

Answer 7

Same as gamma camera - but at higher energy

(high efficiency at stopping, high probability of PE not CS, high conversion of energy to photons, transparent to own emissions, good matching between light and PMT response, mechanically robust, short scintillation time is more important because of higher count rate in PET)

Question 8

What scintillator do we use?

Answer 8

Now LSO and LYSO (cerium doped).
NaI was used but wasn’t ideal, then BGO due to better stopping efficiency and PE interaction rate.

Question 9

Signal processing - what do we need

Answer 9

Two signals in coincidence - both signals in energy window (which is usually worse than SPECT) gives line or response.

Question 10

Coincidence window

Answer 10

Allowable time window between the two interactions for them to be considered a pair. Only paired events used in data.

Question 11

Amplitude discriminator

Answer 11

Amplitude = energy. Energy resolution in PET worse than gamma cameras 350keV to 650keV. Further scatter correction necessary.

Question 12

Corrections in PET

Answer 12

Randoms correction
Scatter correction
Normalisation
Dead time correction
Decay correction
PET scanner calibration

Question 13

Sinogram

Answer 13

Events categorized by plotting each LOR as function of angular orientation vs displacement from centre of gantry.
List mode also available

Question 14

2D vs 3D

Answer 14

3D has no septa, better sensitivity, more sensitive to quality of scatter correction.
2D has septa, and planes are processed independently by 2D algorithm.

Question 15

Bed position overlap

Answer 15

Sensitivity will vary through the scanner depending on the overlap, no overlap sensitivity will vary from 100-0 but a lot of overlap sensitivity will be more steady - takes longer.

Question 16

Estimation of annihilation position in LOR

Answer 16

Time of flight difference
delta x = (c delta t) /2
delta t = t2 - t1
Allows localisation with probability of point of annihilation

Question 17

Does time of flight PET directly enhance spatial resolution

Answer 17

No - increases information gained from each pair of photons. Increase in SNR. Reconstruction choices determine what image quality metrics improve to give greater lesion detectability.

Question 18

What reconstruction is used in PET and what refinements can be used

Answer 18

Usually iterative. OSEM.
Refinements include use of PSF (point spread function) modelling and ToF data

Question 19

Attenuation correction

Answer 19

As we are using detectors either side of the body, we know the distance both photons have travelled through together - don’t in SPECT

Not simple - need to map mu x-ray to mu 511keV, and consider resolution of this map to PET resolution. Also can get artefacts from motion, metal implants, CT contrast agents and mis-registration.

Question 20

Point spread function modelling

Answer 20

Due to depth of interaction effects, the spatial resolution in PET is worse in the radial projection (example shows top to bottom when photon won’t be travelling parallel to hole) towards the edge of the FoV, modelling this will improve reconstructed spatial resolution.

Artefacts from this can be seen in phantoms but not necessarily clinical images, so not a no brainer.