Speaker
Description
The rapid development of high-brilliance X-ray sources, including diffraction-limited storage rings and X-ray free-electron lasers, is enabling X-ray imaging at unprecedented spatial and temporal resolution. At the same time, these facilities generate increasingly large and often highly undersampled datasets, creating challenges for image reconstruction and analysis. In this talk, I will present recent advances in deep-learning-based approaches for 3D/4D X-ray image reconstruction developed for high-brilliance X-ray imaging experiments. In particular, I will discuss physics-informed self-supervised methods for ultrafast 3D/4D reconstruction for X-ray Multi-Projection Imaging (XMPI)[1-4]. In the second part of the talk, I will briefly introduce the ongoing SLS 2.0 and cSAXS 2.0 upgrade at the Paul Scherrer Institute, where commissioning starts in June 2026, and user operation starts in August 2026.
[1] P. Villanueva-Perez et al., Optica 5, 1521–1524 (2018).
[2] Y. Zhang et al., Commun. Eng. 4, 54 (2025).
[3] Z. Yao et al., Meas. Sci. Technol. 36, 085403 (2025).
[4] Z. Hu et al., Adv. Sci. 13, e11933 (2026).
[5] T. Rosén et al., arXiv:2412.09368 (2024).