Home

TODO

Minimum Viable Product for first release

• Kernels calculated in EDKnrc (water, up to 6 MeV)
• Two-source model
• Fluence calculation
• TERMA calculation
• Voxel blocking (Jaws, MLC)
• Varian Millennium 120 MLC
• Off-axis softening
• Ray-marching convolution
• No-tilt approximation
• CUDA functions for hot code
• Python bindings for CUDA functions
• Load RT plan files (RayStation)
• 3DCRT (Performance: < 5s per beam on RTX3080)
• VMAT (Performance: < 3min per beam on RTX3080)
• Varian Enhanced Dynamic Wedge
• TrueBeam 6MV model (WFF)
• TrueBeam 6MV model (FFF)
• Load CT and structure set from file
• Load CT Number to Mass Density LUT
• Orientation: Head First Supine
• Apply density overrides
• Mask calculation with External/Support structures
• TERMA calculation masking
• Clean overall control (module level)
• Command line runner
• Write dose to RT dose file
• Robust logging
• Decent unit test coverage
• CUDA error handlers
• CUDA profiling
• Sphinx/Doxygen docs (readthedocs.io)

Future Version Features

• Kernel tilting
• Kernel hardening (precomputed kernels)
• Material specific kernels (water, lung, bone)
• Electron contamination
• DDA for voxel tracing
• Cumulative vs. Differential kernel definition
• Adaptive dose interpolation
• Smarter handling of structure border voxels
• Max energy 10 MeV
• Varian HD120 MLC
• Couch rotations
• Feet first, prone, etc.
• Load RT plan files (Eclipse/Monaco/Precision/etc.)
• Model optimisation helpers
• Radixact
• Elekta
• Varian Halcyon