EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

An evaluation framework for sparse 4D (3D + time) imaging reconstruction via bootstrapped cross‐validation

Yuhe Zhang iD, Zisheng Yao, Zhe Hu, Tobias Ritschel, Pablo Villanueva-Perez iD

Journal of Microscopy · 01.11.2026

DOI: 10.1111/jmi.70194 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Microscopy, online erschienen am 01.11.2026. Verfasst von Yuhe Zhang und 4 weiteren Autor:innen. Beteiligte Einrichtung: Division of Synchrotron Radiation Research and NanoLund, Lund University, Lund, Sweden. Die Arbeit nennt 74 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: European Research Council, Wenner-Gren Foundations in Stockholm.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Four‐dimensional (4D; 3D + time) microscopic imaging has emerged as a powerful technique for investigating dynamic phenomena in complex systems, enabling direct visualisation of structural evolution in space and time. However, when pushing the limits of spatiotemporal resolution, most time‐resolved imaging techniques yield inherently sparse 4D datasets. While deep learning‐based reconstruction methods have shown promise in reconstructing 4D from sparse spatiotemporal measurements, a practical approach for evaluating their performance in the absence of a 4D reference has, to the best of our knowledge, been lacking. Here, we present a bootstrapped cross‐validation framework that estimates reconstruction performance by quantifying correlations between reconstructions generated from independently sampled subsets of the acquired data, as inspired by the 3D validation strategy in cryo‐electron microscopy, where reconstructions from split datasets are compared to assess resolutions. This enables both qualitative and quantitative assessment in the absence of ground truth. We investigate two representative scenarios with sparse and ultra‐sparse X‐ray datasets and validate this approach using 4D‐ONIX, a 4D deep‐learning reconstruction method, on simulated water droplet collision experiments. The proposed approach provides a reference‐free framework for performance estimation and support for better‐informed experimental strategies across a wide range of ultrafast imaging applications.

Schlagwörter

Stichwörter der Autor:innen

4D imagingdeep learningimage reconstructionreference‐free validationsparse‐view reconstructiontime‐resolved microscopy

Autor:innen und Institutionen

  • Division of Synchrotron Radiation Research and NanoLund, Lund University, Lund, Sweden.
  • Department of Medical Radiation Physics, Lund University, Lund, Sweden.
  • Department of Computer Science, University College London, London, UK.

ORCID-Profile: Yuhe Zhang, Pablo Villanueva-Perez

Förderung

  • European Research Council
  • Wenner-Gren Foundations in Stockholm

Zitieren

APA

Zhang, Y., Yao, Z., Hu, Z., Ritschel, T., & Villanueva-Perez, P. (2026). An evaluation framework for sparse 4D (3D + time) imaging reconstruction via bootstrapped cross‐validation. Journal of Microscopy. https://doi.org/10.1111/jmi.70194

Vancouver

Zhang Y, Yao Z, Hu Z, Ritschel T, Villanueva-Perez P. An evaluation framework for sparse 4D (3D + time) imaging reconstruction via bootstrapped cross‐validation. Journal of Microscopy. 2026. doi:10.1111/jmi.70194

BibTeX
@article{Zhang2026,
  author = {Zhang, Yuhe and Yao, Zisheng and Hu, Zhe and Ritschel, Tobias and Villanueva-Perez, Pablo},
  title = {An evaluation framework for sparse 4D (3D + time) imaging reconstruction via bootstrapped cross‐validation},
  journal = {Journal of Microscopy},
  year = {2026},
  doi = {10.1111/jmi.70194}
}
Kostenlos starten →