Corrections for Physical Factors - Correction Techniques in Emission Tomography - page 85

FIGURE 5.13: Typical streak artifacts due to metallic hip prosthetics in
CT (left). These artifacts heavily influence quantitation in regions close to
the prosthetics in the attenuation-corrected PET image (right).
plants such as hip prosthetics [33], dental implants [41], or implanted pace-
maker leads [27] are present. Due to high photoelectric absorption at CT
energies in metals, CT images with such implants often show typical streak-
like artifacts of both over- and underestimated X-ray density which in turn
propagate into the attenuation-corrected emission tomography images leading
to diculties in determining quantitative image data in regions close to the
metal (see Figure 5.13). Numerous algorithms have been developed to suc-
cessfully reduce these CT image artifacts (see [53] and [61] for examples), and
this preprocessing of CT data has also been applied to CT data for use in
PET attenuation correction [79]. Nevertheless, one study showed that in the
case of dental implants, correcting the CT data with one such algorithm did
not result in significantly different PET uptake values even in regions affected
by this type of artifact [67].
One of the major causes of attenuation correction-induced loss of PET and
SPECT image quality is based on the different time scales on which CT and
emission tomography techniques operate. While a CT scan can be performed
within a few seconds and can thus be taken at breath hold, both PET and
SPECT scans usually require several minutes of acquisition time per bed posi-
tion to achieve satisfying image statistics. Thus emission tomography images
usually comprise data of several respiratory cycles, whereas CT images usu-
ally represent just a snapshot of a cycle. If the acquired CT data was acquired
during a phase that does not match the average PET or SPECT phase, a
spatial misalignment between CT and PET/SPECT data will result, which
besides hampered image fusion may potentially lead to artifacts due to an erro-
neous assignment of attenuation correction factors. This problem is of special
importance in imaging of lung, liver and heart due to the large gradients in at-
tenuation coecients between liver/heart and lung tissue, possibly resulting in
erroneous tracer quantification or even wrong diagnoses. Clinically, the most
frequently seen artifacts of this kind are cold areas at the lung base, indicating
that the CT was done in too deep inspiration to match the mean respiratory
phase during PET (see Figure 5.14). Several studies have demonstrated the