Effects of quantum noise and source blurring on dark-field signal retrieval in X-ray speckle-based imaging
Optics Express · 16.11.2026 · 34 (23) · 44812
DOI: 10.1364/oe.614728 ↗Kurz gefasst
Fachartikel in der Zeitschrift Optics Express, online erschienen am 11.11.2026. Verfasst von Hunwoo Lee und 2 weiteren Autor:innen. Die Arbeit nennt 58 Literaturangaben. Gefördert von: National Institute of Biomedical Imaging and Bioengineering, U.S. National Science Foundation, Congressionally Directed Medical Research Programs.
Abstract (Originalsprache)
X-ray speckle-based dark-field imaging provides high sensitivity to sub-resolution microstructures, yet its quantitative translation to laboratory and clinical systems remains constrained by low photon flux and geometric source blurring. How these hardware limitations propagate through fundamentally distinct retrieval architectures to affect quantitative signal fidelity requires systematic characterization. Here, we evaluate dark-field signal linearity, sensitivity, and baseline intercept under controlled photon starvation and source blurring across two mathematically divergent frameworks: patch-wise explicit tracking (X-ray speckle-tracking X-ray speckle-vector-tracking, XST-XSVT) and differential-based intrinsic tracking (low-coherence system, LCS). Under source blurring from an increasing focal spot size, the smoothing of high-frequency speckle patterns causes a 46.4% reduction in signal sensitivity for XST-XSVT, whereas raw LCS exhibits a smaller net reduction of 22.5% but scales non-monotonically with blur. Under increasing photon starvation across the tested flux levels, XST-XSVT experiences a 46.7% loss in sensitivity; however, analytical Poisson variance subtraction mitigates this noise-induced underestimation, recovering sensitivity by 18.6% at the lowest tested flux. In contrast, unregularized pixel-wise LCS suffers an 84.2% loss in sensitivity under photon starvation. Across both source blurring and photon starvation, XST-XSVT preserves high dark-field linearity ( R 2 ≥ 0.9722) with increasing sample thickness and maintains a zero-consistent baseline intercept, accurately estimating zero dark-field signal at zero sample thickness. Conversely, native LCS exhibits reduced linearity and an elevated non-zero baseline intercept, denoting quantitative inaccuracy under both hardware constraints, particularly at the lowest photon flux and largest focal spot. Spatial window averaging significantly improves LCS performance: at the lowest tested flux, linearity is largely restored ( R 2 = 0.9834) and sensitivity improves by 167.3%; under source blurring, linearity is similarly restored ( R 2 ≥ 0.9878) and sensitivity scaling becomes strictly monotonic (falling by 38.6%). Nevertheless, the quantitative inaccuracy reflected by the non-zero baseline intercept persists in LCS for both photon starvation and source blurring. Consequently, for applications requiring larger focal spot sizes, window-averaged LCS is advantageous due to its superior sensitivity retention and restored linearity, whereas for applications constrained by extremely low photon flux, XST-XSVT is preferred for its robust linearity and unbiased quantitative accuracy. This characterization establishes operational boundaries for low-power and low-coherence X-ray systems, guiding algorithm selection and framework optimization for preclinical and clinical applications.
Förderung
- National Institute of Biomedical Imaging and Bioengineering
- U.S. National Science Foundation
- Congressionally Directed Medical Research Programs
Zitieren
APA
Lee, H., Yuan, J., & Das, M. (2026). Effects of quantum noise and source blurring on dark-field signal retrieval in X-ray speckle-based imaging. Optics Express, 34(23), 44812. https://doi.org/10.1364/oe.614728
Vancouver
Lee H, Yuan J, Das M. Effects of quantum noise and source blurring on dark-field signal retrieval in X-ray speckle-based imaging. Optics Express. 2026;34(23):44812. doi:10.1364/oe.614728
BibTeX
@article{Lee2026,
author = {Lee, Hunwoo and Yuan, Jingcheng and Das, Mini},
title = {Effects of quantum noise and source blurring on dark-field signal retrieval in X-ray speckle-based imaging},
journal = {Optics Express},
year = {2026},
volume = {34},
number = {23},
pages = {44812},
doi = {10.1364/oe.614728}
}
EUVIMED