A proposed alternative to phase-space recycling using the adaptive kernel density estimator method
A schematic showing the cylindrical coordinate geometry used for the preprocessing of the phase space variable. is the particle direction, is the component, is the - (planar) component, and is the azimuthal angle of w.r.t. the radial vector.
Overlay of scatter plots of various PS parameters: (a) radius vector versus polar angle and (b) Azimuthal angle of the planar component of the particle's direction versus radius vector .
In-air fluence analysis for the original PS distribution and AKDE generated distributions. Angular distributions and energy distributions were generated using a radius around the central axis. (a) Relative fluence versus radii (cm). (b) Central axis angular distribution as a function of angle (degrees). (c) Central axis energy distribution as a function of energy (MeV).
A comparison between benchmark calculation and ion-chamber (CC13) data for three field sizes , , and : (a) Transverse profiles. (b) Depth dose profiles. The relative uncertainty in the high dose region for the Monte Carlo calculation was less than in each voxel.
Transverse profiles showing PS recycling, AKDE, and a benchmark calculation. % dose differences for PS recycling and AKDE with respect to the benchmark calculation has been shown for three field sizes: (a) , (b) , and (c) . The relative uncertainties in the high dose region for both AKDE and PS calculations were less than 1.2% , 0.9% , and 0.8% .
Depth dose comparisons for PS recycling and AKDE with respect to the benchmark calculation. (a) The relative dose as a function of depth for a , , and field size. (b) % dose difference between the PS recycled run and the benchmark for a , , and field size. (c) % dose difference between AKDE and the benchmark for a , , and field size.
Global bandwidths for AKDE variables.
Correlation coefficients from the original PS distribution and KDE transformed PS distribution.
Article metrics loading...
Full text loading...