AbstractID: 6893 Title: Optimization with both dose-volume and biological constraints for lung
IMRT
The lung is considered to be a parallel organ whose response to radiation exhibits a strong
volume dependence. In dose-volume based optimization algorithms, lung protection is usually
accomplished by imposing multiple dose-volume constraints on the volume fraction receiving
more than 20-25 Gy. Another approach is to use the parallel model, for which the complication
probability increases as a function of the volume fraction damaged (Fdam). To assess the
probability of individual voxel damage, the model uses a logistic function, with parameters
determined by fitting to clinical data. The total fractional damage is given by averaging the voxel
probabilities over the entire lung. This Fdam model has been implemented in our quadratic dosevolume based optimization algorithm as an alternative objective function for the lung. In this
work, we compare optimization based on Fdam with optimization based on dose-volume
constraints. In both cases, standard prescription dose and homogeneity constraints are used for
the target and a maximum dose constraint is used for the cord. It is shown that: (a) it is feasible
to optimize using a hybrid objective function that is dose-based for the target and the cord and
Fdam-based for the lung, and (b) the results obtained with Fdam are similar to those obtained
with dose and dose-volume constraints for all three structures. The advantage of using Fdam,
instead of dose-volume constraints, is that we can directly enter a requirement of our dose
escalation protocol (lung fractional damage) as a constraint in the optimization algorithm.