Small Volume Lattice Treatment Planning Automation for Head and Neck Malignancies: A Feasibility Study
Abstract
Purpose
This work introduces a framework to automate treatment planning of small-volume lattice radiotherapy applied to head and neck malignancies.
Methods
Four retrospective CT-datasets of head and neck tumors with contoured GTVs were employed. An ESAPI script (Varian Eclipse v18) was developed to automate treatment planning as follows. Lattice vertices were placed using a deterministic hexagonal close-packed (HCP) 3D grid inside a contracted GTV (lattice volume). To improve the dose gradient, a greedy selection algorithm was applied that penalizes placing vertices within the same axial slices. Candidate spheres were preferentially selected if they overlapped regions of high FDG uptake and excluded if they intersected the neurovascular bundle. Additionally, the script created a VMAT plan using 6 MV-FFF energy with up to four arcs and a series of optimization structures to enforce a sharp gradient from the vertices.
Results
GTV volumes ranged between 16.3 and 87.5cc. Lattice generation included the following parameters: sphere diameter of 5mm, lattice inward margin of 3–5mm, and vertex separation of 15mm. The total number of spheres placed ranged from 2 to 9. The prescription to the lattice spheres was 800 cGy single fraction for oropharynx cases (n=2) and 2700 cGy in 3 fractions for oral cavity cases (n=2). Each individual sphere received D95% ≥ 95% for all cases. The lattice valley (lattice volume minus spheres) received a D2%(near-maximum) of 468 and 486 cGy, and a D98%(near-minimum) of 173 and 176 cGy for oropharynx cases, and a D2% of 1265 and 1270 cGy with D98% of 318 and 591 cGy for oral cavity cases. At least a 45% decrease in vertex-to-valley dose was achieved.
Conclusion
Treatment planning automation for small-volume lattice radiotherapy in head and neck malignancies can provide a valuable tool to standardize sphere placement, optimize vertex-to-valley dose gradients and dose metrics while reducing planning time.