Gaussian sample model in in-line imaging

We investigate the gain in Shannon information that can be extracted from an X-ray image obtained after coherent free-space propagation of the transmitted beam and subsequent digital processing of the detected image. We show that simulated digital forward free-space propagation can produce a much higher formal information gain, both in projection imaging and in phase-contrast computed tomography, compared to conventional phase retrieval based on the Transport of Intensity equation. However, it appears that the extra information gained in the simulated free-space propagation may be due in part to superficial high-frequency content in the obtained images, rather than due to a genuine improvement of the spatial resolution. This points to the need to critically evaluate the performance of different types of image quality metrics and their relationship to the information content of the images.