A Software Application to Evaluate the Dosimetric Sensitivity of Individual Stereotactic Radiosurgery Plans to Positional Variations

Abstract

In stereotactic radiosurgery, the dosimetric effects of positional variations depend on patient- and plan-specific factors. This work demonstrates the use of in-house software to provide a statistical analysis of changes due to positional variations for any SRS plan. The software gives 3D dose distributions and DVHs based on data exported from the treatment planning system. We demonstrate its use by comparing the probabilistic dose distributions for an example multimet SRS patient under several treatment scenarios featuring different numbers of isocenters and different positional variation action levels. We developed a software application that considers DICOM files from a commercial treatment planning system. The application was used to sample 1,000 geometric transformations from a continuous, highly-flexible, user-defined, patient motion model. Dosimetric effects of the transformations were simulated by sampling the affected target volume from a static dose cloud. The application was validated using known doses calculated by the treatment planning system under known transformations. Using the validated application, we assessed the dosimetric effects of multiple treatment scenarios on four previously treated patients that varied in number of targets and target proximity to organs at risk. While the application provides 3D dose distributions and DVHs, we evaluated the change in the volume of the target receiving at least the prescription dose (ΔV100%). Presented are example results for one of the patients. We compared the effect of treating this patient’s two mets individually vs. treating them simultaneously, as well as using 1 mm & 1° vs. 2 mm & 2° positional variation action levels. When the two mets of the example patient were treated with a single isocenter and 1 mm & 1° action levels, the minimum, median, and interquartile range of ΔV100% was -28.7%/-23.2%, -6.6%/-6.6%, and 7.4%/6.3%, respectively, for met 1/met 2. When each met was treated with an individual isocenter, the values were -8.2%/-13.3%, -3.0%/-3.8%, and 3.2%/4.0%. When treated with a single isocenter and 2 mm & 2° action levels, the values were -39.7%/-33.8%, -9.6%/-8.7%, and 9.7%/8.5%. The sensitivity of the delivered dose distribution to positional variations was measured using ΔV100% for an example patient under several treatment scenarios. As expected, treating the two mets with a single isocenter and using more tolerant action levels resulted in larger dosimetric changes. While the observed effects are aligned with generalized expectations, our approach statistically considers patient- and plan-specific factors and provides clinicians with a probabilistic delivered dose distribution. If used during treatment planning and delivery, this approach could directly result in SRS plans more robust to positional variations.