| This research addresses the planning of Gamma Knife radiotherapy, which is an alternative to treating a variety of brain abnormalities with surgery. The principal aim of this work is to develop an automated planning system that will make it simpler, less time consuming and hopefully more effective for the clinical personnel to develop treatment plans. Currently, treatment planning is a time-consuming task that involves an iterative process of shot selection, placement, and adjustment. Our goal is to replace the iterative part of the planning with an optimization-based real-time planner. Our strategy is to: (1) automate initial shot selection and placement using a combined process of skeletonization and bin-covering, (2) optimize the exposure time for each shot to improve the target coverage while minimizing toxicity to the surrounding tissues, and (3) fine-tune the shot configuration by adjusting shot locations, and adding or deleting shots to further improve the balance between target coverage and normal tissue toxicity. The efficiency and effectiveness of our approach is derived from (1) the use of skeletonization and bin-covering to provide good starting point for the development of the plan, (2) the easy-to-solve linear fractional program that explicitly accounts for the dual objectives of maximizing target coverage while minimizing toxicity, at the same time accounting for dose-renormalization and (3) the fine-tuning step that explicitly accounts for shot overlapping, dose renormalization, and target shape to make determining of hot spots and estimating the effects of shot movement, addition and/or deletion possible. The planning system has been implemented on a Windows-based platform and tested using clinical cases from the standard Gamma Knife treatment model as well as the automatic positioning system (APS) model. The planner consistently produces, in 1--2 minutes, plans with dose conformity compatible to manual plans normally created in 1--4 hours. |
