MOTION MANAGEMENT - Radiation Oncology: A Physicist's-Eye View - page 151

The downward sloping orange line shows what happens in the

absence of motion. One sees only the consequences of reducing
the dose as the field is widened, in order to maintain the same
level of toxicity. The EUD progressively reduces as the field is
made wider and the dose is reduced.
The hump-shaped purple line shows what happens when both
effects are allowed for. It is, in essence, the product of the two

other curves. For too-small margins, the EUD is low due to

underdosage at the target edge. The EUD then rises as the

margin increases and the target coverage is thereby improved.
Finally, at too-large margins, the EUD falls due to the need to
reduce the dose in order to keep the normal tissue morbidity
constant.

Conclusions from the model

This model is very simplistic and its results should not be taken as
being quantitatively accurate. However, it illustrates a very important
point, namely that there is an optimum margin size, that gives the
highest EUD (and hence TCP) for a given fixed level of normal tissue
toxicity . Smaller or larger margins would be worse, i.e., would lead
to lower EUDs. The model also predicts that the optimum margin is
approximately two times the standard deviation of random motion
(slightly corrected for the penumbra size). Figure 7.7b captures the
essence of the basis for choosing the best safety margin to use.

Random and systematic motion

Motion may be either random or systematic. If random, variations in
position occur during a patient's treatment either between fractions,

or more usually, within a treatment fraction

−

with a Gaussian-like
distribution of values. Systematic motion, on the other hand, is likely
to show up as a consistent error in a patient's treatment, that may or
may not vary from patient to patient. The distinction between these
two types of uncertainty in the location of patient anatomy was
introduced in Rabinowitz et al . (1985) where simulator and port films
were analyzed. Retrospective analysis of day-to-day variations in the
location of anatomic landmarks or metal clips relative to the field
border were seen to be approximately Gaussian in distribution and
were interpreted as random variations. On the other hand, there were
consistent deviations between the port films and the initial simulation
film (that was taken to represent the desired beam placement) and
these were interpreted as systematic variations. Of interest was the

−