Robust spatiotemporally integrated fractionation in radiotherapy

Spatiotemporally integrated fractionation involves finding a fluence-map and a number of treatment sessions that maximize tumor-kill subject to dose-limits on organs-at-risk (OAR). This problem was recently formulated using the linear-quadratic dose-response model. Owing to the uncertainty in dose-response parameters, however, a solution presumed optimal might be infeasible in practice. We address this via a robust counterpart and its convex reformulation wherein the price of robustness is small and robust solutions are less infeasible than nominal even outside our uncertainty sets.