Biomedical Engineering Reference
In-Depth Information
the radioactive tracer with subsequent interpretation in a framework of phys-
iological models.
The physical phenomenon underlying the positron emission limits the res-
olution of PET principally to 2 to 3 mm. During practical applications of PET
imaging, the resolution is further reduced due to the limited detector size and
the smoothing applied during image reconstruction. The latter is necessary
to maintain sufficient signal and limit the influence of statistical noise. For
further details the reader is referred to excellent reviews (see, for example,
Eriksson et al.
2
) and to Chapter 5 of this topic for a discussion on the influ-
ence of scanner errors.
Image registration most often relies on specific features detectable in the
images, i.e., characteristic tracer uptake patterns in an organ or part of the
human body. This is especially important when retrospective registration
techniques are employed. PET images are stored with standard image orien-
tation transverse to the patient's head-foot axis. A pixel represents a measure-
ment of radioactivity concentration at a particular position inside the field of
view (i.e., the volume covered by the detector system). These pixel values can
be related to a physiological variable by means of an appropriate model. Usu-
ally, the radioactive tracer does not have a uniform distribution but shows a
tracer-dependent-specific pattern with increased uptake, for instance, in the
brain's gray matter compared to white matter, or in the heart compared to
lung tissue. Different physiological processes in various organs can be inves-
tigated, such as the distribution of blood flow, oxygen utilization, protein
synthesis, receptor binding, or glucose consumption. For patients with
pathologies like tumors or areas of reduced perfusion due to infarction, these
uptake patterns will deviate from those of normal subjects, a fact which is uti-
lized to draw diagnoses. It must be pointed out, however, that pathologies
like metastases may be missed in cases where they are not delineated from
surrounding normal tissue due to similar activity concentration.
3
From the first images produced with PET, it was obvious that PET images
did not show details of anatomical structures with high resolution, if at all.
This is even true for the brain when using
18
18
F]-fluoro-2-deoxy-
D-glucose) to determine the local glucose consumption. With FDG, PET
images show patterns with considerable resemblance to the underlying
structures known from anatomy as shown in Figure 9.1. This fact was also
explored in creating a registered brain atlas based on PET and MR images.
F-FDG (2-[
4
There is pronounced uptake in gray matter structures compared to white
matter. High uptake in the cortex, the prominent structure, with its typical
irregular folding, marks the outline of the brain's surface and hence, with its
easily recognizable substructures, establishes a collection of natural internal
landmarks. Other important landmarks can also be identified in the central
brain and correlated to anatomical structures. Similar observations can be
made with tracers of cerebral blood flow.
The identification of distinct anatomical structures or landmarks can be
much more difficult in cases when the tracer used during PET studies only
exhibits focal uptake in a smaller part of the imaged volume. Figure 9.2 (top)
