Corrections for Physical Factors - Correction Techniques in Emission Tomography - page 74

TABLE 5.1: Approximate -values for 140 keV ( 99m Tc SPECT) and 511
keV (PET) in cm 1

(calculated using values from [38])

Material at 140 keV at 511 keV

Adipose tissue 0.143 0.090

Water 0.155 0.096

Soft tissue 0.157 0.098

Cortical bone 0.285 0.172

direction. Connected to the loss of energy is a decrease in wavelength of the
gamma photon.
A gamma photon may also interact with a (predominantly inner-shell)
electron by transferring all its energy, thereby ceasing to exist and ejecting
the electron from the atom. This process is known as the photoelectric effect.
Both effects contribute to the attenuation of gamma rays in matter, and
the linear attenuation coecient can therefore be written as

(E ; material) = Compton (E ; material) + Photo (E ; material): (5.11)

For a given attenuation material, both contributions generally decrease
when the gamma energy increases. In PET and SPECT scans which use
gamma energies above 100 keV, the contribution of the photoelectric effect
to the total attenuation coecient is negligible for both soft tissue and bone
structures, and by far the predominant effect is Compton scattering. This is
no longer the case in high-Z material such as lead, where the photoelectric
effect is predominant for SPECT energies and roughly equal to the Compton
eect at PET energies. Approximate -values in PET and SPECT imaging
are given in Table 5.1.
In PET, attenuation correction factors are independent of the location of
the annihilation event on the line of response because of the fact that the
whole line of response is traversed by either one of the gamma photons. This
is readily visible in Figure 5.4. If P detect denotes the total detection probability
of the photon pair which is created in r 1 on the line of response defined by
detectors d 1 and d 2 , then obviously
P detect / e R r 1

e R d 2

r 1 (r)dr = e R d 2 d 1 (r)dr d 1 (r)dr

· (5.12)

is independent of the specific location of r 1 on the line connecting d 1 and d 2 .
Therefore, for every line of response in PET, just one attenuation correction
factor has to be known to perform attenuation correction which is thus simply
done by multiplying the measured number of counts gi i on the line of response
i by the attenuation correction factor for this line:
g A i = e R i (r)dr · g i : (5.13)

This is not the case in SPECT, where the attenuation correction depends
on the location of the point of origin along the detection line and therefore