EUVIMEDEUVIMED entdecken
Crossref

Deep learning–based automatic aberration compensation and reconstruction of the wrapped phase in digital holographic microscopy

Yuzirui Zhang, Siyu Wei, Demian Wang, Chao Wang, Jie Chen, Xin Tang, Yong Kong

Applied Optics · 10.07.2026 · 65 (20) · 6725

DOI: 10.1364/ao.601755 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Applied Optics, online erschienen am 01.07.2026. Verfasst von Yuzirui Zhang und 6 weiteren Autor:innen. Die Arbeit nennt 37 Literaturangaben. Gefördert von: Natural Science Foundation of Shanghai, Shanghai Municipal Education Commission.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In digital holographic microscopy (DHM), the reconstructed wrapped phase is often severely affected by background aberrations arising from optical path imperfections, device inaccuracies, and environmental disturbances. Existing aberration compensation methods generally rely on phase unwrapping procedures, which are prone to error propagation and instability under complex aberration conditions. To address these limitations, this study proposes a deep learning–based non-parametric automatic aberration compensation framework for wrapped phase reconstruction. The proposed method adopts an adversarial reconstruction architecture, employing HRET-UNet as the generator and PatchGAN as the discriminator. The generator integrates residual dense blocks, an efficient channel attention mechanism, and Transformer modules to jointly model local interference fringe details and global aberration distributions. In addition, a joint optimization loss framework is introduced to enhance physical consistency, structural fidelity, and optimization stability during wrapped phase reconstruction. Experimental results demonstrate that the proposed method consistently outperforms UNet, UNet3+, and PACUnet3+ under tilt, curvature, mixed, and high-order aberration conditions. The proposed framework effectively suppresses background distortions while preserving local phase continuity, thereby improving the reconstruction accuracy and robustness of wrapped phase compensation. These results indicate that the proposed method provides a stable and efficient solution for quantitative phase imaging in digital holographic microscopy.

Förderung

  • Natural Science Foundation of Shanghai
  • Shanghai Municipal Education Commission

Zitieren

APA

Zhang, Y., Wei, S., Wang, D., Wang, C., Chen, J., Tang, X., & Kong, Y. (2026). Deep learning–based automatic aberration compensation and reconstruction of the wrapped phase in digital holographic microscopy. Applied Optics, 65(20), 6725. https://doi.org/10.1364/ao.601755

Vancouver

Zhang Y, Wei S, Wang D, Wang C, Chen J, Tang X, et al. Deep learning–based automatic aberration compensation and reconstruction of the wrapped phase in digital holographic microscopy. Applied Optics. 2026;65(20):6725. doi:10.1364/ao.601755

BibTeX
@article{Zhang2026,
  author = {Zhang, Yuzirui and Wei, Siyu and Wang, Demian and Wang, Chao and Chen, Jie and Tang, Xin and Kong, Yong},
  title = {Deep learning–based automatic aberration compensation and reconstruction of the wrapped phase in digital holographic microscopy},
  journal = {Applied Optics},
  year = {2026},
  volume = {65},
  number = {20},
  pages = {6725},
  doi = {10.1364/ao.601755}
}
Kostenlos starten →