Aperture-collimated proton LATTICE radiotherapy with a GRID-like entrance pattern: a compact and robust delivery strategy.

Objective.Proton LATTICE radiotherapy (LRT), a form of spatially fractionated radiotherapy (SFRT), delivers a high dose to intratumoral vertices while maintaining low valley dose. Single-field optimization (SFO) is robust for preserving peak-to-valley patterns but results in a high entrance dose. Prior evidence suggests that smaller, more closely spaced high-dose regions may improve the therapeutic ratio. In this work, we propose aperture-collimated proton LRT (AC-LATTICE), which reshapes SFO proton LATTICE fields into narrow GRID-like entrance beamlets to reduce entrance dose while preserving vertex coverage, spatial modulation, and robustness.Approach. A multi-field, geometry-constrained semi-face-centered-cubic vertex optimization framework was developed to generate non-overlapping intratumoral vertices (4.4 mm diameter; minimum 3D spacing 1.4 ± 0.2 cm) and map each vertex to a deliverable GRID beamlet with minimal geometric error under spacing/clearance constraints. A fixed GRID entrance pattern was imposed in beam's-eye view (1.0 cm hole pitch; 5 mm openings); this two-dimensional pitch defines aperture-plane sampling and is distinct from the achieved 3D vertex spacing after feasibility constraints. Proton AC-LATTICE plans were retrospectively created in RayStation v2023B for eight DIBH primary liver cancer patients (mean GTV 89.7 cc; range 57.6-160.6 cc) following consensus guidance. Prescription was 18 Gy per vertex and 3 Gy to GTV using SFO. Metrics included GTVD95%, Dmean, gEUD (a= - 10), PVDR (D2%/D50%), skinD1%, and liver-GTVDmean. Robustness evaluations used 3.5% range and 5 mm setup uncertainties (second-worst-case), with CBCT-based analysis across the course. Statistical comparisons were performed using a two-sided paired t-test.Main results. AC-LATTICE preserved spatial modulation (PVDR ⩾3.5) across dosimetric analyses while maintaining stable target and normal-tissue metrics. Nominal plans achieved PVDR 3.6 ± 0.4, GTVD95%3.1 ± 0.08 Gy, GTVDmean5.8 ± 0.6 Gy, gEUD 3.5 ± 0.1 Gy, liver-GTVDmean0.9 ± 0.3 Gy, and skinD1%1.4 ± 0.6 Gy. PVDR decreased to 3.3 ± 0.4 in the robust second-worst-case scenario (p< 0.05), with average changes of 1.7%, 5.8%, and 2.8% in GTVD95%,Dmeanand gEUD (p< 0.05). CBCT evaluation maintained PVDR (3.6 ± 0.6 on the first CBCT; 3.5 ± 0.6 on the last CBCT), with no meaningful degradation in other metrics. For the end-to-end test using a head-and-neck phantom, the gamma passing rate was 99% using 2%/2 mm criteria.Significance and conclusion. AC-LATTICE is a compact, robust, and clinically feasible proton LRT delivery strategy that maintains robust spatial fractionation under treatment uncertainties and anatomical variation, supporting broader applicability of proton SFRT when the entrance dose is of concern. This compact approach may extend SFRT to more patients, including those with smaller or difficult-to-treat targets.