PET imaging

Similar to SPECT, PET is a form of tomographic nuclear imaging. However, PET relies on the near simultaneous detection of the pair of gamma photons that are released from an annihilation of a positron and an electron. 



1. Positron decay

In positron decay a positron (represented as e+, β+ or e) is released, which is the antiparticle of the electron (e-). A positron has the same mass and magnitude of charge except that the charge is positive.

Positron emission


Radionuclides that decay via positron emission typically have a larger proportion of protons compared to the number of neutrons (see Segré chart). These proton-rich radionuclides are typically produced in a cyclotron.

The energy difference between the parent and daughter nuclei must exceed 1.022 MeV (2 x 0.511 MeV) for positron decay to occur.


2. Positron travels through matter

  1. As it travels it collides with atoms losing energy and causing ionisation (main method of radiation dose deposition in patient)
  2. As it collides the positron is deflected and path becomes tortuous. The length of the path depends upon the number of collisions and the starting energy of the positron. This means that the distance between where the positron is emitted and where it annihilates is variable.


3. Annihilation



  1. Shortly after its production a positron will annihilate with an electron
  2. The energy from annihilation is released in the form of two photons with an energy of 511 keV
  3. If the electron and positron are at rest before annihilation (initial momentum is zero) after annihilation the momentum of the photons must remain zero. To achieve this the annihilation photons must travel in opposite directions (final momentum is zero)


PET radiopharmaceuticals

The most commonly used radionuclide is fluorine-18 and most common pharmaceutical label is fluorodeoxyglucose (FDG). FDG is a tracer for glucose metabolism

PET scanner

PET Scanner


Blocks of scintillation crystals (detector blocks) are arranged in a circle mounted on a gantry in one or two rows. The ideal qualities of the scintillation crystal are:

  • High linear attenuation coefficient (LAC) for the 511 keV photons
  • High ratio of photoelectric to Compton interactions
  • High number of light photons produced per gamma photon absorbed
  • Short scintillation light decay time 

Scintillation crystal

  • The NAI scintillation crystal used in SPECT and planar imaging not suitable for PET as LAC not enough for the annihilation photons which have a higher energy of 511 keV
  • Most commonly used scintillator in PET imaging is bismuth germanate (BGO)
  • But the light output and light decay time are inferior to NaI
  • Newer materials, such as lutetium oxyorthosilicate (LSO and gadolinium oxyorthosillicate (GSO) are being developed which have more suitable properties
  • Each scintillation detector block viewed by four photomultilipler tubes (see Gamma camera page for more information on PMTs)
  • Block of crystal subdivided by cutting smaller blocks into the scintillation crystal ("called detector elements") and placing a reflective material in the slits to prevent cross-talk of the light photons between the elements.


Now available to download!
Written by radiologists, for radiologists with plenty of diagrams to explain complicated concepts in an easy-to-follow way. An excellent resource for radiology physics revision.


FRCR Physics Notes: Beautiful revision notes for the First FRCR Physics exam

Forming an image

As annihilation produces two gamma photons that travel in opposite directions, this is used to determine which photons should be used to form the image. Two opposite detector elements must simultaneously detect a gamma photon (to within 1 nanosecond) for those photons to contribute to the image. The simultaneous gamma photon by opposite detector elements is called a coincidence and the line between the two detector elements is called the line of response. The detector elements also encode the total energy deposited by the gamma photons.

Line of response


N.B. this is how collimation is achieved in PET. The lead collimator grids that are used in planar imaging and SPECT are not required.


Data acquisition


2D vs 3D acquisition


2D and 3D imaging


In 2D acquisition only coincidences confined to a single slice of the patient are used to form the image. This is achieved by using a collimator ring made of tungsten (which is highly attenuating) to reject photons that reach the detector at an angle and, therefore, are likely to have originated in a different slice of the patient.

In 3D acquisition no collimator is used and coincidences from a much greater volume of tissue are accepted. This method enables a higher total count rate due to more coincidences by allowed to reach the detector. It is useful when there is relatively little scatter / administered radiation such as in brain or paediatric imaging.


Unwanted coincidence rejection


Coincidence and scatter


Unwanted coincidences cause artefactual lines of response to be calculated which do not correspond to the true location of the annihilations. 

  • Increased scatter coincidence:
    • More material to travel through (e.g. body vs brain imaging)
    • 3D acquisition instead of 2D
  • Solution:
    • Energy discrimination (see Gamma camera page for more information). However the photopeak window is wide due to poor energy resolution of the scintillators so scatter coincidence not eliminated
  • Increased random coincidence:
    • 3D instead of 2D acquisition
    • Increased administered radioactivity
    • Increased duration of coincidence window

Data correction

Attenuation correction


  • Attenuation is greater in PET than in SPECT due to the longer path the photon must travel through the patient.


  1. Assume cross-sectional shape and uniform LAC of tissue at 511 keV 
  2. Measure the transmission of 511 keV photons through the patient for each line of response. A radioactive rod source (gallium-68) that gives rise to annihilation radiation is rotated around inside the detector gantry without the patient and then with the patient. This allows a calculation to be made correcting for the attenuation by the patient.



  • Individual detector elements differ in dimensions and fraction of scintillation light photons that reach the PMTs. Same radiation source may not produce same response in every detector element.


  • Rotating rod source used without object in the scanner to calculate the correction factor required for differences in the individual detector elements
  • Correction factor = measured counts for line of response / average counts for all lines

Dead time


  • Following the detection of a photon a detector element cannot detect another photon for a period of time (dead time)
  • Results in loss of counts especially in 3D scanning


  • Dead time measured and mathematical algorithms that take into account detector behaviour applied to extrapolate from measured counts

Radioactive decay


  • Radioactivity decays as the scanner moves down the patient. The longer the delay from start to finish the more the radioactivity will have decayed 


  • Counts corrected for radioactive decay

Data reconstruction

2D acquisitions are reconstructed using filtered back projection or iterative reconstruction (see Acquiring an image part 2). 


Σ  Summary


  • Decay of radionuclide by positron decay
  • Positron released → travels through body → interacts with electron (annihilation) → release two gamma photons of 511 keV that travel in opposite directions

PET scanner

  • Gadolinium oxyorthoscillicate scintillation blocks arranged in circle around gantry
  • Each block connected to 4 PMTs
  • Blocks sectioned into detector elements with reflective material between them

Forming an image

  • Coincident gamma photons (detected by two detectors along line of response within 1 nanosecond of each other) only are recorded and contribute to image

Data acquisition

  • Dimensions
    • 2D uses collimator to accept only photons from a given slice
    • 3D doesn't use collimator, images larger volume of tissue
  • Unwanted coincidence rejection
    • Scatter coincidence from photons created in same annihilation
    • Random coincidence from photons created in different annihilations

Data correction

  • Attenuation correction
    • Assume cross-section shape and uniform LAC of tissue or
    • Use radioactive rod source (gallium-68) with and without patient to calculate correction required
  • Normalisation
    • Rod source calculates correction factor for individual detector elements (i.e. differences between different lines of response)
  • Dead time
    • Time following detection in which detector insensitive to further incident gamma photons calculated and counts corrected for
  • Radioactive decay
    • Correction made for radioactive decay as scanner travels down the patient

Data reconstruction

  • Uses filtered back projection or iterative reconstruction

Next page: NM Image quality

  Send us your feedback

Pin it

Get our newsletter

Please note: Your email address will never be shared with any 3rd parties. It will only be used for Radiology Cafe communications. Emails are sent approx once a month. Read our Privacy policy for more details.