| Popis: |
Grating-based differential phase-contrast X-ray is a novel imaging modality with excellent soft-tissue contrast. Besides standard X-ray attenuation, it provides complementary information on the differential phase shift and the dark-field signal, which reveals structure variations at (sub-)micron scale. Current experimental setups suffer from a narrow field of view of 2–4cm. Thus, multiple exposures have to be stitched together to image larger objects. However, individual exposures are inherently affected by intensity variations, such that tiling artifacts corrupt the stitched projection. These artifacts are most severe in the differential phase image and highly impact their diagnostic value. To address this issue, we propose a novel optimization-based algorithm for fully compensating these tiling artifacts. Our algorithm estimates a smooth bias field for each individual exposure with a global objective function that minimizes the intensity distortion within and across different tiles in the projection. Compared to a currently widely used heuristic, our algorithm leverages the information available from all exposures to estimate the individual bias fields. The evaluation shows the superiority of the proposed algorithm, as it produces bias-free images. To our knowledge, this is the first bias correction algorithm for differential phase images that yields images with nearly imperceptible transitions between individual exposures. |
