BLUE RIBBON POSTER THERAPY: Understanding & Optimizing Isocenter Placement In Single- and Multi-Isocenter Linear Accelerator-Based Radiosurgery Treatment Planning for Multiple Brain Metastases
Abstract
Purpose
Brain metastases (BM) can be treated with linear accelerator (LINAC)-based stereotactic radiosurgery (SRS). Treatment planning for this modality has evolved over time; plans typically have one, two, or n isocenters, where n is the number of targets. Common isocenter locations include the center of each target, the center of the brain or a region within it, the centroid of the targets, or coordinates determined by an expert clinician. The goals of this work are to 1) investigate the impact of isocenter quantity and location on treatment plan quality, and 2) explore isocenter designation options outside current and historical standards.
Methods
In a simulated phantom, thousands of configurations of targets, organs, and isocenters were optimized, and the resultant analysis served as the framework for advanced isocenter designation models. Next, ten anonymized GammaKnife® patients with 9–11 BMs were selected to be replanned for LINAC-based SRS with 1/2/3/4/5 isocenters. Three different isocenter designation models were utilized, each yielding 50 configurations.
Results
It was demonstrated that increasing isocenter quantity significantly improves integral dose, conformity, and organ sparing. That said, this relationship is non-linear; each new isocenter has diminishing and eventually potentially negative returns. It was also found that plan quality decreases as isocenter distance-to-target increases, and that factors such as target depth, size, and organ proximity influence this relationship—the models that accounted for these factors outperformed the simplest model. Finally, patient-specific quality assurance was performed to confirm plan deliverability.
Conclusion
This work indicates that multi-isocenter solutions, when sufficiently configured and optimized, outperform single-isocenter solutions when treating 9–11 BMs. While the derived models show great promise, future work aims to 1) explore the potential impact of other parameters (e.g., arc configuration), 2) improve the modeling of identified significant phenomena (e.g., organ avoidance), and 3) investigate the models’ handling of other quantities of BMs.