Tolerating uncertainty: photodynamic therapy planning with optical property variation

Treatment planning is of utmost importance in interstitial photodynamic therapy, as it predicts the required light delivery to the target volume in an upcoming treatment. However, planning remains a major challenge due to several uncertainties such as the tissue optical properties and the concentrations of the photosensitizer and oxygen. Any difference in these parameters from the assumed values during planning could significantly affect the outcome of the actual treatment. This work introduces PDT-SPACE, a PDT light source power allocation using a convex optimization engine to minimize damage to organs-at-risk (OAR) with robustness against variation in tissue optical properties. Three power allocation methods are proposed and compared with respect to the resulting standard deviation in the damage to organs-at-risk and their runtime. The proposed approaches are demonstrated for ALA induced PpIX as photosensitizer in a virtual brain tumor that models a glioblastoma multiforme case. Results show that choosing a power allocation to minimize the OAR damage standard deviation under optical property variation tends to also minimize the tumor coverage as there is only one degree of freedom to optimize upon. This motivates simultaneous source position and power allocation optimization.