Tumor-Geometry–Adjusted Lattice SBRT Using PCA-Aligned and Adaptively Sized Vertices for Veterinary Patients
Abstract
Purpose
Spatially fractionated lattice radiotherapy enables dose escalation in bulky tumors by embedding high-dose vertices within a lower-dose background. In veterinary radiation oncology, tumors are frequently relatively large and geometrically irregular, limiting the applicability of fixed, patient-axis–aligned lattice templates. We developed an automated lattice-generation approach for veterinary patients that aligns lattice geometry to tumor shape using principal component analysis (PCA) and adapts vertex size locally to maximize coverage while maintaining geometric and organ-at-risk (OAR) constraints.
Methods
A scripted module implemented within a commercial treatment planning system computed the GTV principal axes via PCA and rotated a three-dimensional Cartesian lattice accordingly. High-dose spherical vertices (nominal radius 5 mm, 20 mm center-to-center spacing) were placed at alternating grid nodes. Near the tumor boundary, vertex diameters were automatically reduced (down to 70% of nominal) to preserve a ≥5 mm internal GTV margin and vertex-to-vertex spacing. Corresponding low-dose “void” structures were generated to produce a spatially fractionated peak–valley dose pattern. A simultaneous integrated boost plan delivering 66.7 Gy to vertices and 20 Gy to the rest of the tumor in five fractions was optimized and evaluated in a large, irregular canine sarcoma treated with stereotactic body radiotherapy (SBRT).
Results
The algorithm automatically generated a fully constrained lattice geometry in under 5 minutes. In a 643 cm³ GTV (PTV=1011 cm³), boost vertices occupied 12.8 cm³ (~2% of GTV). The plan achieved a mean vertex dose of 71.5 Gy (maximum 76.4 Gy) and a mean void dose of 22.2 Gy while satisfying all OAR constraints and maintaining steep dose fall-off. Characteristic lattice peak–valley dose modulation was preserved.
Conclusion
PCA-aligned, adaptively sized lattice can be generated robustly for veterinary SBRT, addressing the geometric complexity of companion-animal tumors, improving vertex packing and planning automation without and providing a translational framework for human lattice radiotherapy applications.