Angle dependent dose transformer algorithm for fast proton therapy dose calculations

Accurate 3D dose calculation for Pencil Beam Scanning Proton Therapy (PBSPT) is typically performed with Monte Carlo (MC) engines, but their runtimes limit adaptive workflows and repeated evaluations. Current deep-learning proton dose engines often require orthogonality between proton rays and the CT grid, forcing computationally expensive beamlet-wise 3D reinterpolation. We propose the Angle-dependent Dose Transformer Algorithm (ADoTA), which eliminates grid rotation by augmenting the model input with a fast analytical beamlet-shape projection that explicitly encodes beam direction. The model was trained on CT data from 108 patients to predict beamlet dose distributions for initial energies of $70$–$270\mathrm{M}\mathrm{e}\mathrm{V}$ over an $80\times 110{\mathrm{m}\mathrm{m}}^2$ field, and tested on an independent cohort of 50 patients. On the test set, gamma pass rates $(1\mathrm{\%},3\mathrm{m}\mathrm{m})$ were $99.40\pm 0.86\mathrm{\%}$ (thorax) and $99.87\pm 0.23\mathrm{\%}$ (abdomen/pelvis). Single-beamlet inference took $1.72\pm 0.8\mathrm{m}\mathrm{s}$. By avoiding reinterpolation, end-to-end 3D dose computation was reduced by $\approx 86\mathrm{\%}$ relative to the fastest published reinterpolation-based methods. For full treatment plans, gamma pass rates $\mathrm{\Gamma }(2\mathrm{\%},2\mathrm{m}\mathrm{m})$ with a 10% dose cut-off reached $98.4\mathrm{\%}$ (lung) and $98.9\mathrm{\%}$ (prostate). ADoTA provides an angle-aware deep-learning proton dose engine that preserves MC-level accuracy across heterogeneous anatomies while substantially reducing computational overhead.