Input workspace for UHPP: a GPU radiotherapy dose calculation and treatment plan optimization engine

This dataset is the accompanying workspace for the repository UHPP (not publicly available), a GPU-accelerated radiotherapy planning framework built on TERMA ray-tracing, Collapsed-Cone Convolution Superposition (CCCS) dose calculation, beam-orientation optimization, and Geant4 Monte Carlo verification. This work is associated with the following publication: Xu Q, Lyu Q, Jiang L, Jing S, Ruan D, Sheng K. An ultra-high performance parallel (UHPP) framework for complex 4π radiotherapy planning. Phys Med Biol. 2025 Aug 6;70(16). doi: 10.1088/1361-6560/adf2f2. PMID: 40695314. It's associated with a refactor of the UHPP/FastDose codebase described in that paper. The workspace contains three folders: KernelGenZenith8/ — CCCS dose-deposition kernel data for a clinical 6 MV photon spectrum. Includes the 6 MV fluence spectrum (spec_6mv.spec), 14 monoenergetic Geant4 Monte Carlo energy-deposition tables (energy_deposition_*.txt, 0.20–6.00 MeV), and the fitted bi-exponential kernel parameters (fitted_kernel_parameters.txt) used directly by the dose engine. Pancreas/ — Preprocessed input data for five pancreatic cancer patients (Patient001–Patient005), derived from the TCIA collection doi:10.7937/TCIA.ESHQ-4D90. Each patient folder contains: CT density volume (density_raw.bin, uint16, Fortran order), reference dose (doseNorm.bin, float32), grid specification (dimension.txt: voxel counts and sizes in cm), per-structure optimization weights (StructureInfo_else.csv), and binary structure masks (InputMask/, one uint8 file per structure). Patient001 is the default Quick Start demo patient. HeadNeck/ — Preprocessed input data for head-and-neck cancer patients (Patient002–Patient190), in the same format as the Pancreas cohort, derived from the TCIA collection doi:10.7937/K9/TCIA.2015.7AKGJUPZ. The Generated/ folder produced by the pipeline (dose matrices, plan outputs, Monte Carlo results) is not included; it is reproduced by the Quick Start instructions in the repository README.