Combining deterministic and Monte-Carlo methods in the simulation of X-ray attenuation and scatter

Abstract

Purpose: A new hybrid simulation for medical imaging which combines deterministic computation of X-ray attenuation and single-scatter contributions with estimates of higher-order scattering via Monte-Carlo (MC) simulation. Method and materials: We employed mathematically defined objects in simulations of various scanner geometries. Additional features such as the use of form filters or anti-scatter grids were included. For the primary signal, X-ray attenuation was computed along the direct rays from the focus to each detector element. In addition, by summing single Compton or Rayleigh scattering processes inside the object over all paths from the focus to the all detector elements, we calculated the corresponding single-scatter contribution to the signal. Standard MC simulations were used to validate the deterministic method and to compare the contribution from single scattering to the full scatter signal. Differences were determined in the hybrid simulation by a time-optimized MC estimate. The simulations compared to data measured on a CT scanner and a C-arm system using phantoms of various shapes and sizes. Results: Results for the deterministic calculated primary and single-scatter signal were in agreement to within 1 -3 % with the MC simulations. Depending on the material and size of the object, the relative contribution of the single to the full scatter signal varied considerably. The structure in the scatter images was mainly determined by the single scatter signal in all simulations; the difference images therefore show little or no structure and can thus be estimated efficiently by an adapted coarse MC simulation. The hybrid method considerably reduced the computational costs as compared to standard MC. The images from our simulations were in good agreement with the measured data. Conclusions: The new hybrid simulation method allows to assess scatter contributions at low computational expense as compared to standard MC. Thereby it provides a potential basis for systematic scatter corrections in X-ray tomography.