Corrections for Physical Factors - Correction Techniques in Emission Tomography - page 83

While these points are very valuable especially in clinical context, there are
also some specific disadvantages when using CT data for attenuation correc-
tion. One of these problems is based on the aforementioned basic assumption
that from an X-ray point of view, human tissue can be divided into mixtures
of just three basic materials (air, water, and bone). In some scans, this as-
sumption cannot be maintained because other materials that do not behave
like air, water or bone are present. One such instance that is very frequently
seen in clinical scans is the usage of contrast-enhanced CT image data for at-
tenuation correction. Shortly before contrast-enhanced CT scans, a contrast
agent is given to the patient. This can be done either by intravenous injec-
tion into the vascular system to delineate blood vessels from other soft tissue
structures or by oral ingestion where the contrast agent is concentrated in
the intestine. Chemically, these contrast agents contain atoms of high Z ele-
ments like barium and iodine. This ensures an exceedingly high photoelectric
absorption, resulting in high attenuation coecients at X-ray energies and
thus good contrast to soft tissue in CT images. Transformation algorithms
such as the one described above will treat regions containing a contrast agent
as bone-like tissue due to their high X-ray attenuation. However, for these
contrast agents, the influence of photoelectric absorption on total attenuation
coecient decreases when going to PET energies in a much stronger manner
than a mixture of bone and water. Thus, the determined attenuation coe-
cients in those regions will be overestimated which in turn should lead to an
increase in tracer uptake as compared to attenuation-corrected images using
native non-enhanced CT data. Indeed, this type of artifact can be observed
in clinical scans (see Figure 5.12). Additionally, the case of bolus injections
right before or even during the CT scan raises the question whether transient
concentrations of contrast agent in the vascular system will represent a valid
attenuation map even if a correct scaling to PET energies can be found.
The question whether contrast agent-induced overestimation in tracer
quantification has a significant impact on clinical diagnoses and treatment
planning has been the subject of many studies. One study demonstrated PET
image artifacts in veins caused by intravenous injection of an iodine-containing
agent in 4 out of 30 oncology PET/CT scans [4]. These artifacts were caused
by agent concentrations that had significantly higher CT values than in those
cases without artifacts. Another study demonstrated significant overestima-
tion of myocardial FDG uptake quantification when using contrast-enhanced
CT data for attenuation correction in cardiac PET/CT [17]. However, numer-
ous authors suggest that CT contrast agents intravenously given actually do
not result in clinically significant image artifacts or lesion uptake overestima-
tion [8] [64] [89]. This problem seems to be of greater significance when an
oral contrast agent is given prior to the CT scan due to a high range of pos-
sible concentrations and large distribution volumes [23]. Besides having two
CT scans (one without a contrast agent for attenuation correction, another
contrast-enhanced one for diagnosis) [84], correction methods addressing this
problem have been developed. These mainly include different segmentation