Geometry-Based Lattice Positioning for Improved Sfrt Planning In Maas-Sfrthelper
Abstract
Purpose
To extend MAAS-SFRThelper with geometry-based lattice positioning tools, completing an integrated workflow from sphere generation, positioning, and VMAT optimization, to dose evaluation for SFRT planning.
Methods
Previously, we introduced MAAS-SFRThelper's sphere generation engine supporting grid-based, sampling-based, and Voronoi tessellation methods, followed by a comprehensive evaluation module providing dose metrics, 1D/2D/3D visualization, and novel onion-layer analysis. We now present an optimization module that uses geometric analysis to guide lattice positioning prior to VMAT optimization. The module implements: (1) sphere center extraction using area-weighted centroid calculation from ESAPI contours; (2) four geometry-based metrics computed across 72 gantry angles—Sphere Isolation Index (SII) quantifying beam's-eye-view sphere separation, Valley Space Index (VSI) measuring inter-sphere spacing, Sphere Spread Index (SSI) evaluating target coverage and alignment, and OAR Sparing Index (OSI) assessing organ-at-risk avoidance; (3) grid search evaluating N×N candidate positions to identify favorable lattice translations; and (4) automated generation of repositioned lattice and valley structures for Eclipse optimization. Initial testing compared baseline versus repositioned configurations in a phantom study.
Results
Sphere extraction achieved <1mm accuracy compared to known phantom geometry. Grid search evaluated 81 positions in <1 second. In phantom testing with a 22-sphere cubic lattice (8mm radius spheres, 62.5mm target radius), grid search identified a repositioned configuration with 3% higher geometric score but 36% fewer spheres (22→14). Subsequent VMAT optimization showed the repositioned configuration achieved 30% higher PVDR, with 8% higher peak mean dose, 17% lower valley mean dose, and 22% lower OAR mean dose compared to baseline, suggesting geometric metrics may help identify favorable lattice positions.
Conclusion
The new positioning tools complete MAAS-SFRThelper's integrated SFRT planning workflow. Preliminary phantom results suggest geometry-based metrics may help identify favorable lattice positions without iterative dose calculation. Further study and validation with clinical cases is ongoing. The complete workflow is freely available on GitHub.