A library of proton lineal energy spectra spanning the full range of clinically relevant energies

In locations where the proton energy spectrum is broad, lineal energy spectrum-based proton biological effects models may be more accurate than dose-averaged linear energy transfer (LET_d) based models. However, the development of microdosimetric spectrum-based biological effects models is hampered by the extreme computational difficulty of calculating microdosimetric spectra. Given a precomputed library of lineal energy spectra for monoenergetic protons, a weighted summation can be performed which yields the lineal energy spectrum of an arbitrary polyenergetic beam. Using this approach, lineal energy spectra can be rapidly calculated on a voxel-by-voxel level.

Monoenergetic proton tracks generated using Geant4-DNA were imported into SuperTrack, a GPU-accelerated software for calculation of microdosimetric spectra. Libraries of proton lineal energy spectra which span the energy range of 0-300 MeV were computed. The libraries were validated by comparison to Monte Carlo calculations in the literature, as well as lineal energy spectra measured experimentally with a tissue equivalent proportional counter.

The lineal energy libraries have been made available in three data formats, two are plain-text, .csv and .les, and one binary encoded, root. Library files include the lineal energy bin abscissa in keV per micron and the unnormalized number of counts occurring within that bin. A computational technique for summation of the library files to yield the lineal energy of a polyenergetic beam is described in this work.

The lineal energy libraries can be used to rapidly determine the lineal energy spectra at the location of cell cultures for in-vitro experiments and in each voxel of a treatment plan for in-vivo outcome modelling. These libraries have already been incorporated into RayStation 2023B-IonPG for lineal energy spectra calculation and we anticipate they will be incorporated into further dose calculation engines and Monte Carlo toolkits.