Biomedical Engineering Reference
In-Depth Information
the tumor and normal tissues needs to be done so far as is possible
with the patient positioned in a reproducible manner, as discussed in
Chapter 7; otherwise the anatomy at the time of treatment may well
have shifted relative to where it was at the time of imaging.
Until about the mid-1970s, the principal type of imaging available
was X-ray radiography - enhanced by various forms of externally
introduced contrast media to image vessels, lymph nodes, body cavi-
ties, and so forth. The target volume was typically defined by infe-
rence from such radiographs, underpinned by anatomic knowledge
and an appreciation of the typical patterns of disease spread. Dif-
ferent parts of the target volume were often defined by different
means. Thus, for example, the presence of an abnormally straightened
cerebral vessel could indicate the presence of a tumor distending,
and hence presumed to abut, the vessel in one region. Elsewhere,
the tumor might be presumed to extend up to, but not into, some bony
structure thought, on the basis of a bone film, to be uninvolved by
disease - and, still elsewhere, by knowledge of a high probability of
extension of disease along some anatomic pathway.
Normal (i.e., presumed uninvolved) anatomy was likewise im-
perfectly determined from radiographs which show bone-tissue
and tissue-air interfaces well, but the boundaries between soft tissues
at best very poorly. The images were often supplemented by pub-
lished atlases of normal anatomy derived, for example, by meticulous
drawings of cross sections of frozen cadavers, and scaled to match
the patient's outer body contour, measured using lead wire.
All this changed dramatically with the clinical availability of com-
puted tomography (CT) - in the mid-1970s for head scans, and in
the early 1980s for scanning throughout the body - and, now,
magnetic resonance imaging (MRI). Nevertheless, the
principles
for
mapping the patient's body remain to this day the same, namely:
to identify disease where it can be directly imaged;
to infer the presence or absence of disease from normal tissue
abnormalities, or lack thereof;
to combine information gleaned from multiple imaging tech-
niques;
to apply knowledge of the known patterns of disease spread; and
possibly, by marking the anatomy with surgical clips.
The enormous difficulty - indeed, to date, the virtual impossibility -
of delineating the target volume by automatic means is due to the
