Illumination Correction in X-Ray Near-Field Holography: From Dynamic Intensities to Full Wave Fields

Abstract

X-ray near-field holography (NFH) enables lensless, high-resolution imaging of weakly absorbing objects via propagation-based phase contrast, combining high penetration depth with low radiation dose. Its accuracy, however, depends critically on precise knowledge of the illuminating wavefront: standard flat-field correction assumes a temporally stable illumination and fails for spatially complex or dynamically varying probes, introducing systematic phase distortions that propagate into three-dimensional holotomographic reconstructions.

This dissertation addresses illumination correction along two complementary lines. For temporally dynamic illuminations, we introduce an adaptive dynamic flat-field correction that partitions holographic datasets into subsets of consistent dynamics and applies a criterion-based selection of the background-describing components, enabling semi-automatic processing without requiring a separate illumination measurement and reducing evaluation time from one week to one day. For static but spatially complex illuminations, we develop refractive divide&update, a phase retrieval algorithm that corrects for a known, complex probe while resolving strong phase shifts. Coupling this reconstruction scheme with tomographic self-consistency further resolves reconstruction artifacts in strongly phase-shifting three-dimensional objects, as demonstrated on both simulated and experimental data.