Automatic Selection of Isocenter(s) and Minimization of Normal Tissue Dose for Multi-Target Oligometastatic Extracranial SABR
Abstract
Purpose
To develop an automated framework that determines the number and placement of isocenters and target margins for extracranial stereotactic ablative radiotherapy (SABR) in polymetastatic disease, and evaluate lung dose tradeoffs versus uniform margins.
Methods
Hierarchical clustering (HC) using Ward’s method was performed on a proof-of-concept 12 lung target case. N=1 to N=12 potential isocenter locations were generated such that for the case with N=1 isocenters, all targets were clustered together and assigned to one isocenter; for N=2, all targets were clustered into two groups and assigned to one of two isocenters, etc. For each cluster set, margins were calculated such that, for a rotational error of 2 degrees and a translational error of 5 mm, 95% of the GTV received the prescription dose. For each set of N=1 to N=12 isocenters, the total additional PTV margin was calculated and plotted versus the number of isocenters, and the point of maximum curvature was used to determine the final number of isocenters to use. Three plans were then generated using the final isocenter clusters: one using dynamically determined margins (DM), and two using uniform GTV expansions of 5 mm and 10 mm. Plans were normalized such that targets received ≥30 Gy to 98% of their volume. Robust evaluation was performed to evaluate impact on target coverage.
Results
The twelve targets were assigned to one of four isocenters using the HC method, resulting in a mean target margin of 5.75 mm (range = 5.21-6.63mm). For the Lung-GTV subvolume, the Dmean was 7.24 Gy (DM), 6.76 Gy (5mm), and 10.88 Gy (10mm); the V13.5Gy was 392.09cc (DM), 326.09cc (5mm), and 894.85cc (10mm); and the V20Gy was 4.76% (DM), 3.91% (5mm), and 11.70% (10mm).
Conclusion
HC may be used to efficiently determine isocenter assignment with only minimal increases in target margin and Lung-GTV dose.