Clinical Instrumentation
Iterative Reconstruction Method:
In recent years, is increasingly common in
commercial equipment, the use of an algorithm that performs an iterative
process.
The reconstruction begins with data projections and an estimate of the first
image
by
filtered
backprojection.
Doing
a
"reprojection"
(i.e. to
create new projections from the
just
estimated
image)
the method compares the reprojected projections with
the
original
ones
(the acquired)
and from there
on a decision is made
as
how good is the approximation.
If it is good enough it
will
stop
otherwise will keep
iterating.
Therefore, it
is necessary to make the decision when to stop iteration.
This is not trivial because the noise is present in the original data as in
the estimates.
To calculate the new data
set of profiles
it is necessary to
incorporate
in the data
all physical processes
involved in the generation of the images: the variation of the resolution in the detector
field of view, the variation of attenuation and scatter radiation with
the
geometry
of
the object and the distribution of radioactivity.
Mathematically a matrix containing the information about the probability of contribution of each tomographic pixel slice to the pixel of the profile of the projection is constructed. This implies the solution of a system of algebraic equations with many unknown variables, which thus it is difficult to solve. Therefore it is employed an approximate method, such as the iterative mentioned. Today, this process has been made possible by the existence of modern computers with faster microprocessors. Some years ago this approach was impossible to implement routinely because it required hours of processing to generate the images.
Not only there are faster computers today, but those with new algorithms that reduce the time of the iterative process. One is the OSEM reconstruction. This acronym (Ordered-Subsets Expectation Maximisation) is based on an algorithm based on Maximum Likelihood techniques (EM-ML). Previously, all projections were used to perform each iteration. OSEM, the new method, uses only a portion of these projections in each new iteration. The order is carefully chosen dividing the data before starting the process in N = 2, 4, 8, 16, 32, etc., projections. In each iteration it is used 1 / N of the total data. OSEM has been successfully applied to reconstruct images of SPECT and PET. The data acquisition process can be tailored more accurately, leading to better images than those obtained with the analytical reconstruction. OSEM made it possible that iterative methods could be applied in clinical nuclear medicine.
Acquisition and Processing of Images
The acquisition conditions
which
are under the direct control of the operator,
comprises two aspects: the patient and data acquisition.
1)
The patient should be properly
positioned
and not moving
during the study.
Regarding the movement of the patient, some modern instruments bring
algorithms
to correct for
motion.
Wherever
possible the table should be parallel to the head.
2)
During acquisition and data processing the following parameters
must be
taken care of
:
- Collimator (usually high resolution is chosen to avoid degradation of image quality with distance from source to collimator).
- Energy window or windows, according to the isotopes to detect or the corrections to be carried out (attenuation and scattering).
- Selection of the orbit: circular or elliptical depending on the desired resolution. Organs like brain require circular orbit.
- Orientation of the patient.
Page 5 Home Index Clinical Nuclear Medicine Therapeutic Methods Previous Page Next Page