Comparative Evaluation of Three Automated Lattice Placement Strategies for Lattice Radiation Therapy: Hexagonal, Cubic, and Poisson Disk Packing
Abstract
Purpose
Lattice radiation therapy (LRT) is an innovative radiotherapy technique lacking the standardized approach for the spatial distribution of high-dose regions. Manual placement is time-consuming and subjective. Automated regular geometric lattice arrangements can satisfy most clinical requirements, however, the intra-tumoral heterogeneity is spatially irregular. This study aimed to assess three automated geometric lattice placement algorithms and their effects on dose distribution.
Methods
Retrospective planning was performed on 9 patients (3 NSCLC, 3 prostate, 3 head-and-neck) with large GTVs (>100 cc). Lattices(diameter D) were generated using three automated algorithms: (1) Hexagonal Close Packing (HCP, center-to-center distance≥2D); (2) Simple Cubic Packing (SCP, center-to-center distance=2D); (3) Poisson Disk Packing (PDP, center-to-center distance≥2D). Constraints maintained margins≥10mm from OARs and ≥5mm from the GTV boundaries. Plans were optimized on the UIH treatment planning system using VMAT (20Gy for GTV and 60Gy for lattices in 5 fractions). The same beam setups and optimization objectives were applied across three methods per patient. Lattice number, volume ratio, and Peak-to-Valley Dose Ratio(PVDR) were evaluated using ANOVA with a significance level of P=0.05.
Results
All methods completed placement within 2 seconds. Median lattice numbers were 5(range 2-33), 2(1-12), and 7(3-28) for HCP, SCP, and PDP, respectively (SCP vs. PDP: P=0.004). The Vlattices/VGTV (0.2%-3.3%) was significantly lower for SCP compared to HCP (P=0.004) and PDP (P=0.001). Average PVDRs were 2.06(HCP), 2.21(SCP), and 2.02(PDP). SCP exhibited significantly higher PVDR(P=0.024) and lattice D90(P=0.011) versus PDP, while HCP and PDP showed no statistical differences. Spatially, PDP demonstrated superior adaptability in populating narrow, irregular GTV sub-regions where rigid grids (HCP/SCP) failed to place lattices.
Conclusion
PDP provides high lattice numbers camparable to HCP, while SCP has the highest PVDR. PDP represents a feasible automation strategy with superior geometric adaptability for irregular targets, warranting further validation in larger cohorts.