Phase retrieval beyond the homogeneous object assumption for X-ray in-line holographic imaging

X-ray near-field holography has proven to be a powerful 2D and 3D imaging technique with applications ranging from biomedical research to material sciences. To reconstruct meaningful and quantitative images from the measured intensities, however, it relies on computational phase retrieval, which in many cases assumes the phase-shift and attenuation coefficient of the sample to be proportional. Here, we demonstrate an efficient phase retrieval algorithm that does not rely on this homogeneous-object assumption and is a generalization of the well-established contrast transfer function (CTF) approach. We then investigate its stability and present an experimental study comparing the proposed algorithm with established methods. The algorithm shows superior reconstruction quality compared to the established CTF-based method at a similar computational cost. Our analysis provides a deeper fundamental understanding of the homogeneous-object assumption, and the proposed algorithm will help improve the image quality for near-field holography in biomedical applications.