|
| |
Possible Research Topics
(1) Application of the Boland et al technique directly to the
prescription, rather than going through the intermediate step of
appoximating the prescription by a set of pencil beams and then
using this technique to do the leaf sequencing. We could investigate
enhancements of this approach to minimize the number of shapes,
rather than the total dose (subject to a limited degradation of the
deviation from the prescription.) We could then investigate use of
this approach in solving the step-and-shoot problem in [1].
(2) Modification of the Boland et al "network" formulation to the IMAT
case in which additional constraints are needed to relate valid
aperture shapes at adjacent angles.
(3) Conventional planning: Techniques to select the 5 (say) best beam
angles from a list of 36 (say) candidates, developing a treatment
plan by optimizing weights, beam energies, and wedge
placement/orientations at the selected angles. (Assume that the
aperture shape is fixed to conform to the target shape). These
treatment plans will be as easy to implement as those used in
current practice, and conform closely enough to the "culture" that
we can reasonably expect them to be used. Branch and bound
techniques, with reasonable heuristics, can be used to select the
best angles, and continuous optimization techniques of the types
already investigated can be used to select weights, deal with
wedges, etc.
|
