Biomedical Engineering Reference
In-Depth Information
which characterize the non-uniform intensity of each beam constitute
the
fluence map
of the beam, which is characterized by an array of at
least 30 x 30 intensity values. Thus, in IMRT, there are far more
some 10
13
in toto
distinguishable possibilities
and it would take
about 1,000 years to assess all possibilities and choose the best. In
light of these numbers,
evaluation of all distinguishable plans is not
feasible
. Instead, one must resort to intelligent search algorithms that
examine only a very small subset of all possible plans.
−
−
It is worth noting that optimization is potentially useful in uniform-
intensity radiation therapy as well as in IMRT (Niemierko, 1992).
There are only two related differences: optimization of uniform-
intensity radiation therapy requires the setting of far fewer variables
than does IMRT; and, for that reason, optimization is essential for
planning IMRT, but not for uniform-intensity radiation therapy.
Scoring vs. Searching
In the past, the attention of those attempting optimization was focused
on the search process. This was for a couple of reasons. On the one
hand, the scale of the problem makes it mathematically challenging
and, therefore, technically interesting. On the other hand, the problem
of searching in a large space of possible solutions to find the optimal
solution is one that has received a great deal of attention from
mathematicians, and there exists a shelf-full of mathematical approa-
ches which can be taken over and tried. However,
no matter how hard
it may be to develop an efficient and reliable search process, it is much
harder to develop a realistic and reliable measure of plan goodness.
In consequence, it is the latter problem which should receive the lion's
share of attention
.
Let us take a peek at
our planner as he
develops an IMRT
plan using the opti-
mization scheme ill-
ustrated in Figure 9.4.
Figure 9.5 is a sneak
view of him. He has
his hands folded and
appears to have noth-
ing to do except to
stare at the meter
Figure 9.5. IMRT being planned
