Purpose: To introduce a novel method of dose approximation for use in Gamma Knife (GK) treatment planning.
Methods: A treatment plan was made utilizing the TMR10 calculation at the center of the Radiosurgery Head Phantom CIRS Model 605 for each sector’s individual dose contribution at each collimation size. Once exported, the resulting dose arrays were arranged into a Python object. A parser was created to extract individual shot information for a given patient’s targets, breaking each shot into its sector activations. The appropriate dose arrays were selected from the Python object and superimposed into the parsed shot position. The resulting dose arrays were evaluated based on pairwise comparisons between the original TMR10 calculation and the superposition approximation (SA). These consisted of Dice Similarity Coefficient (DSC), Hausdorff Distance (HD), and the GK deliverability metrics: coverage, selectivity, and gradient index. DSC and HD comparisons were made at threshold levels ranging from 10% to 90% of each target’s maximum dose in increments of 10%, as well as at each target’s prescription isodose used in planning.
Results: A cohort of 30 patients previously treated at MBPCC, consisting of 71 total targets, was evaluated. This consisted of patients treated for a single metastasis, multiple metastases, meningiomas, pituitary adenomas, glioblastomas, and oligodendrogliomas. The pairwise comparisons exhibited strong agreement between the two calculation methods, especially in the ranges commonly used for planning. At the 50-60% threshold levels, DSC was strictly above 85% and HD was strictly below 3mm. The average differences in deliverability metrics (TMR10 minus SA) for coverage, selectivity, and gradient index being -0.013, -0.007, and 0.119, respectively.
Conclusion: The proposed method accurately approximates the TMR10 dose calculation of GammaPlan in the range of clinical planning significance. This method may be used for external evaluation of treatment planning optimizations to the GammaPlan system.
Funding Support, Disclosures, and Conflict of Interest: This research was supported in part by the Dr. Charles M. Smith Medical Physics Research Endowment.