Efficient X-ray dark field contrast simulations using a condensed history approach

Due to the growing interest in X-ray phase contrast imaging (XPCI), the development of XPCI simulators has become an active field of research. Efficient computer simulations are a prime tool for optimizing the XPCI setup and acquisition process. In addition to phase contrast, modern XPCI setups provide insight into the unresolvable sample micro structure distribution through the dark field signal. However, to accurately simulate XPCI dark field signals, the virtual sample model should contain many fine structures, considerably increasing the simulation time. Moreover, accurately modeling micro structure distributions is challenging and the micro structure distribution of one virtual sample cannot be easily transferred to another. In this work, we apply a condensed history approach to dark field simulations, removing the need to explicitly model the micro structures in the virtual sample. Instead, a tunable dark field material parameter, representing a measure of the material micro structure distribution, is attached to the sample. The condensed history simulation model offers a substantial increase in simulation speed and the dark field material parameter is easily transferable between samples. The implementation is validated using the edge illumination XPCI setup. Condensed history simulations are compared to real data, showing good agreement, and the simulation speed increase is presented for increasing sample thickness and field of view. The simulation model is shown to remain valid even for thick samples. Simulated computed tomography data sets, with and without explicit micro structure, are favorably compared to each other.