Biomedical Engineering Reference
In-Depth Information
If one were to plot the dose distributions of the non-target tissues for
these two cases as overlaid DVHs, one would get something like the
histograms seen in Figure 8.7. This figure raises the vexing problem
of crossing DVHs. If a normal
100
100
few fixed fields
360° rotation
few fixed fields
360° rotation
tissue DVH for one plan is
everywhere below that for
another, it is easy to deduce that
it is the better (i.e., will be less
morbid for the patient) and,
equally, the DVH which lies
everywhere above another is the
worse. But, what is one to think
if they cross? In Figure 8.7 the
red DVH (for the 360° rotation plan) shows a large volume receiving
low doses, the blue DVH (for the 4-field box) shows a lesser volume
receiving low doses, but a greater volume receiving higher doses.
Which is one to prefer?
If I knew a definitive answer to this question I would happily reveal
3 fields
3 fields
80
80
rotation
rotation
60
60
40
40
20
20
0
0
0
0
25
25
50
50
75
75
100
100
125
125
Dose (% of the target dose)
Dose (% of the target dose)
Figure 8.7. Overlaid DVHs for the
non-target tissues for two plans.
it. But, the fact is that we have here reached the boundary of our
knowledge - or, at least, our ability to quantitatively evaluate plans;
the clinician's experience is all we have to go on. And, if we cannot
answer this question, then what use are all our models, and what faith
can we put in a computed score?
Having taken this pessimistic position, let me step back a little and
present a modest bit of modeling.
The influence of tissue architecture
Andrzej Niemierko and I some time ago undertook a simple computer
experiment. We created a cylindrical tumor of 8 cm diameter within a
cylindrical patient of 20 cm diameter, much as for the cases sketched
in Figure 8.5, and planned its irradiation with photon beams, in one
case with a 3-beam technique and in the other case with a 360
°
rotation.
The tumor was required to receive essentially the same dose in both
cases. The normal tissue outside the tumor received, of course, a
quite different dose distribution in the two cases. In fact, the DVHs
in Figure 8.7 are the normal tissue DVHs for this experiment. Two
types of normal tissue were considered: serial and parallel. We then
applied two NTCP models (Niemierko and Goitein, 1991, 1993a) to
compute the difference in the NTCP in the two plans, separately for
