Premium
Using spatial information about recurrence risk for robust optimization of dose‐painting prescription functions
Author(s) -
Bender Edward T.
Publication year - 2012
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4704499
Subject(s) - medical prescription , mathematics , medical imaging , statistics , dosimetry , medicine , mathematical optimization , computer science , nuclear medicine , artificial intelligence , pharmacology
Purpose: To develop a robust method for deriving dose‐painting prescription functions using spatial information about the risk for disease recurrence.Methods: Spatial distributions of radiobiological model parameters are derived from distributions of recurrence risk after uniform irradiation. These model parameters are then used to derive optimal dose‐painting prescription functions given a constant mean biologically effective dose.Results: An estimate for the optimal dose distribution can be derived based on spatial information about recurrence risk. Dose painting based on imaging markers that are moderately or poorly correlated with recurrence risk are predicted to potentially result in inferior disease control when compared the same mean biologically effective dose delivered uniformly. A robust optimization approach may partially mitigate this issue.Conclusions: The methods described here can be used to derive an estimate for a robust, patient‐specific prescription function for use in dose painting. Two approximate scaling relationships were observed: First, the optimal choice for the maximum dose differential when using either a linear or two‐compartment prescription function is proportional to R, where R is the Pearson correlation coefficient between a given imaging marker and recurrence risk after uniform irradiation. Second, the predicted maximum possible gain in tumor control probability for any robust optimization technique is nearly proportional to the square of R.