2365 - Vertices Geometric Efficiency Optimization for Lung Cancer Lattice Therapy Treatment Planning

R. Mansfield¹, J. Ginn², T. C. Mullikin³, Q. J. J. Wu², and Y. Sheng¹; ¹Duke University Medical Center, Durham, NC, ²Duke University, Durham, NC, ³Department of Radiation Oncology, Duke University Medical Center, Durham, NC

Purpose/Objective(s): Spatially fractionated radiation therapy (SFRT) improves the therapeutic effect of radiotherapy (RT) likely through bystander and abscopal effects in tumors and by reducing normal tissue toxicity. A form of SFRT, known as GRID, was introduced to deliver a relatively high but heterogeneous radiation dose through a perforated screen. Traditional 2D GRID RT delivered with an alloy grid may be expanded to 3D in Lattice Radiation Therapy (LRT) in which treatment is illustrated as high dose islands or vertices within a sea of lower dose. LRT treatment planning for locally advanced lung cancer patients asks for intelligent vertices placement to improve the efficiency of radiation dose coverage inside the tumor as well as spare organs-at-risk (OARs). This research seeks to develop a packing algorithm to intelligently place vertices within the tumor volume. Materials/

Methods: This project merges two programming environments based exploratory research pathways. First, the tumor volume was defined. Clinical data was read to produce the position of points and faces. The faces specify the connectivity of the vertex mesh from which a tumor volume is defined. The second objective, for which this research serves, was to maximally pack vertices within the defined tumor volume. Mathematically optimal packing was accomplished through automated fitting of hexagonally close-packed (HCP) dose island vertices. An HCP matrix of vertices was grown to a size sufficient for the defined tumor volume. A perimeter boundary was defined based on the dose island radius to ensure vertices do not exceed the tumor volume. The vertices matrix was then manipulated to obtain geometrically optimal packing through the acceptance and rejection of vertices which maximizes the number of vertices within the tumor volume. One late-stage lung cancer patient was simulated with various vertices settings using the proposed algorithm. Tumor volume coverage and distance from the tumor surface to outer vertices edge or tumor to vertices (DTV) were reported.
Results: Output from optimal packing of the 441 cc tumor is provided in Table 1. The algorithm generates a vertices location array within a minute. Results showed that the packed vertices can efficiently cover the target volume with minimal dead space as demonstrated by the DTV.
Conclusion: In this study, a packing algorithm was developed for LRT. Coupled with a previously developed auto planning tool, it is an important step towards a fully automated LRT treatment planning workflow.

Table 1: Varying Vertices Spacing and Diameter in Packing a 441 cc Tumor

VERTICES PACKED	VERTICES SPACING [CM]	VERTICES DIAMETER [CM]	DISTANCE BETWEEN VERTICES [CM]	DTV [MM]	TUMOR VOLUME COVERED [%]	RUN TIME [MIN:SEC]
64	2	1	1	26	7.6	0:47
43		2	0	12	40.9	0:47
22	3	1	2	62	2.7	0:25
15		2	1	36	14.3	0:21
11	4	1	3	29	1.3	0:14
7		2	2	4	6.7	0:14