Traditional practice in radiation therapy has been to consider the
problem of tumor control as a static problem even though the time
course of radiation delivery (fractionation) is an important issue.
Recent developments have changed that paradigm and the field is
evolving to include situations where decisions are made in stages. At
one end of the spectrum are the development of functional imaging
methods that may provide information regarding the radiation response
of tissues during the course of treatment. At the other end,
increased survival of patients with metastatic disease results in
patients undergoing a series of treatments over months or years. The
goal is to develop optimal treatments for these situations.
Methods of dynamic programming are appropriate for the solution of
such optimization problems. Examples of problems from these two ends
of the spectrum will be presented. In one set of cases, imaging
provides information that was not previously available; in another
set, decisions are made on the basis of imperfect imaging information.
Issues that might limit this approach are the curse of dimensionality
and the construction of models that integrate the imaging information
in a useful way.
Back to Mathematical Problems, Models and Methods in Biomedical Imaging