510 / 1544
ESTRO IMRT course 2015
IMRT optimization: algorithms&cost functions
Brussels October 5th, 2015
-33-
• The ‘free fluence’ optimization problem is well behaved. Fast algorithms can
be used.
• The solution space needs to be explored in an interactive manner.
MLC sequencers and precise dose calculation are too slow yet.
Because:
• It is a long way to go from coarse field shapes to the patient-specific
optimum fluence distribution. It is very easy to get trapped in local
minima when directly optimizing field shapes
- see later.
…vs. Direct Aperture Optimization (DAO), Simulated Annealing (Rapid Arc ©)…:
• e.g. ‘Pareto-optimization/navigation’ currently only works for fluence
distributions