Speaker
Description
The ability of X-ray radiation to penetrate matter is key to its use as a non-destructive probe for the inner structure of objects, materials and tissues, by ways of computed tomography. For samples with vanishing absorption contrast, lens-less X-ray phase contrast imaging (XPCI) is an unique tool, that is adopted on a growing number of beamlines. Recent cutting edge examples are as diverse as nanoimaging of neuronal tissue for connectomics [3,4], morphological transitions of nanoparticles [5,6], or ultrafast imaging of hydrodynamics at X-ray free electron lasers (XFELs) for cavitation [7]. Rapidly evolving X-ray imaging capabilities by advances on instrumental side (photon counting detectors, fourth-generation sources) as well as on the analysis and reconstruction side (e.g. automated segmentation with machine learning), demand specialized software for efficient and high-quality phase retrieval in XPCI, in particular in the high resolution full-field variant of holographic tomography (holo-tomography).
To this end, we present HoToPy [1], a Python-based toolbox for X-ray holo-tomography. It features a state-of-the-art phase retrieval algorithms for the deeply holographic and direct contrast imaging regimes, including nonlinear approaches and extended choices of regularization and constraint sets [2]. By using recent optimization methods, automatic differentiation, popularized by machine learning, and fully differentiable forward operators, we enable straightforward extensions and modifications to the problem of phase retrieval. Examples are the addition of a (smoothed) total variation regularization or structured illumination, while still being able to apply (non-smooth) objects constraints, such as a compact support or range constraints. Furthermore, HoToPy features auxiliary functions for iterative tomographic alignment, image processing, and simulation of imaging experiments.
We demonstrate HoToPy’s capabilities and discuss current developments and challenges in phase retrieval on data recorded with the ‘GINIX’ instrument of the P10 beamline at the PETRA III storage ring (DESY, Hamburg).
[1] J. Lucht, P. Meyer, L.M. Lohse and T. Salditt, submitted (2025), preprint https://arxiv.org/abs/2506.11567
[2] S. Huhn, L.M. Lohse, J. Lucht and T. Salditt, Optics Express (2022)
[3] J. Livingstone, et al. in Biomedical Optics Express (2025)
[4] A. Azevedo, et al., in Nature (2024)
[5] L. Grote et al., in Nature Communications (2022)
[6] M. Veselý et al., in ChemCatChem (2021)
[7] H. Hoeppe et al., in New Journal of Physics (2024)
