Scribd
Upload
78 views

PET Reconstruction

PET scanners use scintillating materials like BGO or GSO to detect photons emitted during radioactive decay, converting them to light read by PMTs. Data is organized by photon energy, position, and time to construct lines of response between detector pairs. Images are reconstructed using filtered backprojection or iterative methods to unfold projection data into pixel values representing radiotracer concentration.

Uploaded by

Shahzad Ali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
78 views

PET Reconstruction

PET scanners use scintillating materials like BGO or GSO to detect photons emitted during radioactive decay, converting them to light read by PMTs. Data is organized by photon energy, position, and time to construct lines of response between detector pairs. Images are reconstructed using filtered backprojection or iterative methods to unfold projection data into pixel values representing radiotracer concentration.

Uploaded by

Shahzad Ali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

PET-Scanner Hardware

In PET the emitted Photons are detected using a scintillating material like BGO(bismuth
germanate) or more recently GSO (gadolinium oxyorthosilicate) and LSO (lutetium
oxyorthosilicate).

The photon deposits energy, mainly by scattering, in the scintillator which is converted to
scintillation light that is read out and converted to an electric signal using PMTs
(photomultiplier tubes).
Data Organisation in PET

The quantities measured in PET are photon energy, position and time.
A line of response (LoR) is constructed if both photons have been detected in two different detectors within
a certain time and energy window.
The information content is that a PET event occurred somewhere along this line(LOR).
Listmode and Matrix representation are the two approaches to stored the Data.
List mode and Matrix representation

In List mode format all detected events are written to an event list.
List mode format is mainly used in a research context and for experimental scanner systems.
In matrix representation only a binary information about the event is kept, that is, that an event fulfilling the
imposed criteria occurred in a given LOR.
Matrix representation is used in clinical routine.
Image Reconstruction

A Sinogram stored Projection data acquired in two-dimensional (2D) mode or three-dimensional


(3D) mode.
The Data is stored in the form rows and columns representing angular and radial samplings,
respectively.
Acquired data in each row are compressed (summed) along the depth of the object.
Reconstruction performed an unfolding operation to reconstruct the image.
Simple Backprojection

A reconstruction matrix of a definite size (e.g., 128 128 pixels) is chosen.


The image matrix is in (x, y) coordinates, the sinogram data are in polar coordinates.
An image in(x, y) position is related to polar coordinates (r, )
r = x sin + y cos
The measured counts in the projection sinogram corresponding to the calculated r are added to the (x, y) pixel in the
reconstruction matrix.
A(x,y)= 1/
=1 p(r, )

where p(r, ) is the count density in the sinogram element in the acquired matrix and N is the number
of projection angles.

When all pixels are computed, a reconstructed image results from this simple backprojection
Filtered Backprojection

Blurring of the object is the main problem with Simple Back projection.
Since the blurring effect decreases with distance (r) from the object of interest, it can be described by a 1/r
function.
It can be considered as a spillover of some counts from a pixel of interest to the neighboring pixels,
and the spillover decreases from the nearest pixels to the farthest pixels of different frequencies.
The blurring effect is minimized by applying a filter to the acquisition data.
Filtered projection data are then backprojected to produce an image that is more representative of
the original object.
The Fourier Method

The filtered backprojections are accomplished by the Fourier method.


In the Fourier method A(x, y) in the object and p(r, ) in a sinogram are related by by the Fourier transformation.
The projection data obtained in the spatial domain (Fig. 4.2a) can be expressed in terms of a Fourier series in
the frequency domain as the sum of a series of sinusoidal waves of different amplitudes, spatial frequencies.\

F (vx, vy) = F f(x, y)


where F(x, y) is the Fourier transform of f(x, y) and F denotes the Fourier
transformation.
A filter H(), in the frequency domain is applied to each projection, i.e.,

F1(v) = H(v) F(v)

where F() is the filtered Fourier projection which is obtained as the product of H() and F().

Finally, the inverse Fourier transformation is performed to obtain filtered projection data in the
spatial domain

Which are then backprojected in the same manner as in the simple backprojection.
Iterative Reconstruction

In iteration methods of image reconstruction, an initial estimate of image activity is stored in a matrix identical in size to
the acquisition matrix. Usually all pixels in the estimate matrix are assigned an initial constant value.
A set of simulated projections are unfolded from this estimate for comparison with the acquired projection.
After all projections are unfolded, each estimated projection is then compared with the corresponding measured
projection.
If there is a difference between the two, a weighted correction is applied to all pixels in each
estimated projection.
All corrected projections are backprojected to obtain an updated image, which is then entered into the algorithm
as the estimated image for the next iteration.

qi =
=1 aijCj
Where qi is the simulated projection Cj is the counts (activity) in the jth pixel, aij is the probability that
an emission from pixel j is recorded, and N is the number of pixels in the ith LOR.
Conclusion

The PET scanner itself has to be set up properly and the individual detector responses have to be
determined and quantified correctly.
The image quality strongly depends on the number of acquired PET events.
Iterative algorithms, like OSEM, are the preferred choice in case of limited statistics.
The most important parameters in iterative reconstruction are the number of iterations and the choice of
the number of subsets.
If the number of iterations performed is too small, the image will look blurred and constrast is low

You might also like