Biomedical Engineering Reference
In-Depth Information
FIGURE 5.4
One slice from a CT scan of the test object used to measure voxel dimensions accurately is
shown. The holes are precision drilled at 10 mm intervals in a solid cylinder of perspex
(
1.1g · cm
3
). Fine tubing can be inserted into the holes and filled with radioactive or
paramagnetic contrast.
FIGURE 5.5
Distortion caused by not correcting the projection data for attenuation is shown. The image
on the left is a PET reconstruction of an [
18
F]-deoxyglucose (FDG) scan of the thorax. The patient
has a single pulmonary nodule, seen in the right hilum, with high FDG uptake. The image
in the center is the corresponding PET attenuation scan that is used for correcting the projec-
tions prior to reconstruction. It is, in effect, a low resolution CT scan. The image on the right
shows the effect of applying attenuation correction to the emission data. The uptake in the
nodule is now not distorted, especially in the anterior-posterior axis and the physiological
distribution in the low-attenuation lungs is now more accurate. Both of these effects are
errors caused by not applying attenuation correction to the projection data. (Image courtesy
of Ms. Bernadette Cronin, Clinical PET Centre, Guy's & St Thomas' Hospital, London.)
will artifactually decrease towards the center of the object. Second, the
inconsistency between projections will often cause “streaking” artifacts in
areas of high contrast of radioactivity concentration (see Figure 5.5). This
may cause difficulties for any registration algorithm that optimizes a voxel
similarity measure. Attenuation may be corrected by multiplying the emis-
sion data with a correction factor file based on transmission measurements
