Corrections for Physical Factors - Correction Techniques in Emission Tomography - page 75

FIGURE 5.4: In PET, the detection probability P detect of an annihilation
on a given line of response between two detectors d 1 and d 2 is independent of
the actual location r 1 (left), r 2 (right) of the annihilation on the line.

makes proper attenuation correction in SPECT a less straightforward task
than in PET. One method of correcting SPECT data for uniform attenuation
characterized by that has found its way into clinical practice is known as the
Chang method [21]. It is a multiplicative post-reconstruction correction per-
formed on SPECT images. If f(x;y) and f AC (x;y) denote the non-corrected
and corrected activity distribution in (x;y), then according to Chang

f(x;y)
M 1 P i=1 exp (s i )

f AC (x;y) = (5.14)

where M is the total number of projection angles and is i is the distance from
(x;y) to the border of the attenuation object in the ith projection. This for-
mula has also been applied in non-uniform attenuation correction, especially
in an iterative approach [44].
Instead of pre-reconstruction (PET) or post-reconstruction (SPECT) at-
tenuation correction approaches, this correction step can also be performed
during reconstruction using iterative algorithms as the maximum likelihood
expectation maximization (MLEM) or the ordered subsets expectation max-
imization (OSEM) algorithm that easily allow modeling of the attenuation
process in the reconstruction. See Chapter 3 for detailed informations on these
approaches.
The basic effect of attenuation in emission tomography images is the un-
derestimation of tracer concentrations in regions close to and especially deep
inside attenuating objects. Examples for PET are given in Figure 5.5 and Fig-
ure 5.6 for a cylinder phantom scan and an FDG patient scan, respectively.