Topic 13: PET & SPECT

Question 1

What is SPECT?

Answer 1

Single photon emission (computed) tomography, substance labelled with radio-nuclide (photon emission), relatively cheap but relatively low resolution and sensitivity

Question 2

What is PET?

Answer 2

Positron emission tomography, substance labelled with radio-nuclide (positron emission), expensive due to scanner and cyclotron.

Question 3

Two other functional tomographic modalities other than PET and SPECT?

Answer 3

MRI spectroscopy and functional MRI.

Question 4

Radionuclides used in SPECT?

Answer 4

Technetium-99m, Thallium-201

Question 5

Example clinical applications of SPECT?

Answer 5

Cardiac perfusion, parkinson’s and blood flow.

Question 6

Positron Emission Tomography

Answer 6

Substance labelled with radio-nuclide. Radioactive decay is a random process Detection of photon-pair, no collimator necessary,

Question 7

PET physics mechanism?

Answer 7

An unstable nucleus emits a positron and a neutrino and the positron hits an electron, and 2 gamma photons are emitted in opposite directions. The two gamma photons are detected at the same time in the detector, with a slight difference (we call this a timing window then we call it a coincidence)

Question 8

PET applications in medicine?

Answer 8

oncology (cancer) neurology psychiatry cardiology What for? - diagnosis, measuring therapy effectiveness and drug development.

Question 9

The advantage of PET?

Answer 9

Sensitive for picomolar concentrations.

Question 10

What factors influence quality of the PET image?

Answer 10

-Biology - Scanner - amount of detected counts - Image reconstruction and processing (including kinetic modelling algorithms)

Question 11

poisson noise process variance and CV calculations?

Answer 11

variance = mean CV = STD/MEAN=1/SQRT(MEAN)

Question 12

if we have more counts what does that mean for noise?

Answer 12

so cv is lower for higher mean. so for more counts we will be happier as we will have less noise.

Question 13

activity

Answer 13

(average) rate of emissions/decays

Question 14

how do you relate half life and lambda?

Answer 14

lambda = ln(2)/half-life

Question 15

what is the decay of radioactive decay?

Answer 15

1-exp(-lamba*Time)

Question 16

the number of decays in time interval [0,T] is _____ distributed.

Answer 16

Poisson distributed

Question 17

what is the mean of counts that are poisson distributed?

Answer 17

Question 18

what is the coefficient of variation for counts with a poisson distribution?

Answer 18

1/sqrt (mean of counts)

Question 19

How do you increase counts?

Answer 19

1) Inject more radioactivity
- Limitations;

patient dose, scanner counting limitations, purity of inject substance

2) Scan longer

limitation:

scanning time available, patient movement

3) Avoid attenuation

limitations:

patient comfort

Question 20

What is the LOR?

Answer 20

Each coincidence event represents a line in space connecting the two detectors along which the positron emission occurred (i.e., the line of response (LOR)).

Question 21

the mean number of _____ ____ will therefore be proportional to the accumulated acitivity on that LOR.

Answer 21

Detected events

Question 22

Two different reconstruction algorithms?

Answer 22

Anlaytic e.g. filter back projection or

statistical e.g. maximum likelihood.

Question 23

What are analytic methods?

Answer 23

They are based on mathematical inversion of the complete set of ‘line integrals’ (X-ray transform)

Question 24

Reconstruction of backprojection?

Answer 24

acquisition : we know the line where the detected photons came from

“Reconstruction”: photon could have originated anywhere along the back-projected line; assign counts along line with equal probability, proportional to detected counts (back projection)

Question 25

How do we solve the fact that back projection gives us a very smooth version of the original

Answer 25

we will need to apply a sharpening filter.

This can be done before or after backprojection.

Before: Filtered back projection (FBP)

After: Back project and filter (BPF)

Question 26

FBP issues?

Answer 26

Need to discretise analytic formulas - how do i discretise fourier transforms??

Analytic methods ignore various effects

• Noise (has to be handled by filtering)
• Finite detector size
• Limited number of angles - in surgery you cannot

(you can try and use different maths to solve some of these problems but not all of them)

Question 27

What is the fundamental trade - off that we have in medical imaging all of the time?

Answer 27

contrast vs noise

Question 28

Noise is all-important in emission tomography. It is determined by

Answer 28

• the amount of detected counts
• the choice of image reconstruction algorithm

Question 29

KEY POINT : FBP pros and cons?

Answer 29

It is fast

but

inflexible and can lead to noisy images