Semi-classical Monte Carlo algorithm for the simulation of X-ray grating interferometry

Abstract

Traditional simulation techniques such as wave optics methods and Monte Carlo (MC) particle transport cannot model both interference and inelastic scattering phenomena within one framework. Based on the rules of quantum mechanics to calculate probabilities, we propose a new semi-classical MC algorithm for efficient and simultaneous modeling of scattering and interference processes. The similarities to MC particle transport allow the implementation as a flexible c++ object oriented extension of EGSnrc—a well-established MC toolkit. In addition to previously proposed Huygens principle based transport through optics components, new variance reduction techniques for the transport through gratings are presented as transport options to achieve the required improvement in speed and memory costs necessary for an efficient exploration (system design—dose estimations) of the medical implementation of X-ray grating interferometry (GI), an emerging imaging technique currently subject of tremendous efforts towards clinical translation. The feasibility of simulation of interference effects is confirmed in four academic cases and an experimental table-top GI setup. Comparison with conventional MC transport show that deposited energy features of EGSnrc are conserved.

Towards virtual histology with X-ray grating interferometry

“…The same detector as for TLI was used. A mean visibility of 16% was measured and well agrees with theoretical simulations 55 (see supplementary materials).…”

Virtual grating approach for Monte Carlo simulations of edge illumination-based x-ray phase contrast imaging

“…This has not been discussed here, as the speedup would be the same for explicit and virtual grating simulations, leading to the same conclusion regarding the relative simulation time reduction. Alternatively, variance reduction techniques are sometimes implemented to achieve a reduced simulation time [20].…”

“…In general, modelling the effect of the gratings is a crucial aspect of the simulation. Different approaches exist, which can be roughly divided in three categories: (1) modelling the grating as a plane (2D) [18,26,27,44,23,14,15,25,24,32,20],…”

“…(2) modelling the grating as a volume with bars extending along the direction of the optical axis (3D) [21,22,40,41], or (3) simulation frameworks where the effect of the grating is accounted for implicitly, without introducing the gratings as components in the simulation [16,33,31]. Two-dimensional (flat) grating models are common in, but not restricted to, wave-optics simulations and involve either an idealised binary representation of the grating [18,26,27,44] or application of a complex transmission function (projection approximation) [23,14,15,25,24,32,20]. However, assuming the gratings are infinitesimally thin is not always a valid approximation.…”

“…XPCI simulators are often either based on wave optics calculations [13,14,15,16], Monte Carlo (MC) code [17,18,19,20,21,22], or a combination thereof [23,24,25,26,27,28]. In addition, ray tracing-based [29,30,31,32] and empirical methods [33] have been reported.…”