Probabilistic treatment of uncertainties in the computational modeling of BNCT for brain tumors

Christopher N Culbertson, Purdue University


Treatment planning in BNCT utilizes treatment parameters to represent therapy conditions which are subject to considerable uncertainty. Among these uncertain parameters are the tumor to blood ratio, the blood boron concentration, and the cell-kill chance. Calculations of tumor cell survival and tumor control, two treatment outcome indices, utilize parametric averages of these values. Because the survival function for high LET radiation and the control function as defined by the Porter function are highly non-linear, estimates of average survival and control based on average parameter values are in significant error. A formally rigorous procedure that accounts for parameter uncertainties is developed that can be applied for BNCT treatment planning. From experimental data and observations, probabilistic distributions are constructed to describe parameter uncertainty. By integrating the survival and control functions with these probability distributions as weighting functions, the correctly-weighted averages are found. The formally rigorous average values differ significantly from values based on single-valued parameters. To facilitate routine application of the probabilistic method, a technique is developed that allows the rigorous average values to be calculated without integration by using “effective” parameter values. The effective parameter values, formed from the product of the parametric averages and an f-factor, reproduce the probabilistic average when applied in single-value calculations. The variations in survival and control behavior resulting from probabilistic treatment of parameter uncertainty have implications on anticipated outcomes and on the common planning practice employed in BNCT.




Ott, Purdue University.

Subject Area

Nuclear physics|Surgery|Biomedical research

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server