pith. sign in

arxiv: 2605.01265 · v1 · submitted 2026-05-02 · ⚛️ physics.med-ph · physics.optics

X-ray dark-field imaging from intensity flow: A Fokker-Planck approach to grating interferometry

Samantha J. Alloo , Florian Schaff , Regine Gradl , Benedikt Gunther , Franz Pfeiffer , Kaye S. Morgan This is my paper

Pith reviewed 2026-05-10 14:54 UTC · model grok-4.3

classification ⚛️ physics.med-ph physics.optics
keywords grating interferometrydark-field imagingFokker-Planck equationX-ray imagingimage retrievaltransmission imagingmedical imaging
0
0 comments X

The pith

A Fokker-Planck equation method retrieves transmission and dark-field X-ray images with fewer artifacts than conventional fitting.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The authors introduce a retrieval algorithm for grating interferometry that uses the X-ray Fokker-Planck equation to model intensity changes and separate transmission from dark-field signals. They test it on experimental data from a test object and a mouse chest scan under different noise and exposure conditions. The new method gives results matching the standard sinusoidal approach but reduces artifacts from grating defects and performs better with short exposures and high noise. This matters for practical medical applications where equipment imperfections and dose limits are common concerns.

Core claim

By treating the intensity variations across the gratings as governed by the Fokker-Planck equation, the authors derive new formulas to extract transmission and dark-field information. Experimental validation shows the images are consistent with conventional retrieval while being smoother and less affected by grating perturbations or reduced flux, with particular benefits for noisy, fast-acquisition data.

What carries the argument

The X-ray Fokker-Planck equation modeling the propagation of intensity through the interferometer setup to derive signal separation formulas.

Load-bearing premise

Intensity measurements in the grating setup follow the Fokker-Planck equation closely enough that the derived formulas cleanly isolate transmission and dark-field without interference from unmodeled setup-specific effects.

What would settle it

Experimental data from a grating with known scratches where the Fokker-Planck images show no reduction in artifacts compared to the conventional method would indicate the claim does not hold.

Figures

Figures reproduced from arXiv: 2605.01265 by Benedikt Gunther, Florian Schaff, Franz Pfeiffer, Kaye S. Morgan, Regine Gradl, Samantha J. Alloo.

Figure 1
Figure 1. Figure 1: Grating interferometry (GI) imaging setup established at the Technical University of Munich (TUM) for imaging of the test sample. This sample comprised a piece of wood (brown) and scrunched tissue paper (white) placed on a spirit level (black with yellow interior). 2 Methods 2.1 Image Retrieval via the X-ray Fokker–Planck-Equation Step 1: Interlace grating interferometry phase-stepping scans In order to ap… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the interlacing procedure used to convert GI data into a form suitable for X-ray Fokker–Planck image retrieval. Columns from each scan image (7 different G2 positions) are stacked sequentially: the first column from all 7 images side-by-side, then the second column, and so on, forming a single interlaced image. This image is resized to a square by duplicating each row 7 times until its height m… view at source ↗
Figure 3
Figure 3. Figure 3: Retrieved (a) transmission and (b) dark-field images of the test sample using the conventional pixel-wise sinusoidal fitting method. Retrieved (c) transmission and (d) dark-field images using our new XFPE-based approach. (e) and (f ) are a magnified region from both dark￾field images, highlighting how a grating defect–indicated by the yellow arrow in c–is handled by the two methods. The conventional approa… view at source ↗
Figure 4
Figure 4. Figure 4: Retrieved transmission (1st column) and dark-field images (2nd–5th columns) of a mouse chest using the conventional pixel-wise sinusoidal fitting method (top row) and our new XFPE approach (bottom row). Results are shown for different exposure times, indicated along the top of the figure. The turquoise arrows in panels b) and g) indicate the lung edge, where the conventional retrieval method amplifies nois… view at source ↗
Figure 5
Figure 5. Figure 5: Azimuthally averaged Fourier power spectra of the retrieved a) transmission and b) dark-field images from simulated 20 ms (solid full-color curves) and experimental 140 ms (dashed full-color curves) grating interferometry (GI) data of the mouse chest. The transmission spectra are plotted on the same axis, whereas the dark-field spectra are displayed on separate axes because the approaches retrieve differen… view at source ↗
read the original abstract

Grating interferometry is a promising diagnostic technique that enables simultaneous acquisition of three complementary, synergistic X-ray images: transmission, differential phase, and dark-field. Its key advantage over other setups is its ability to use large pixels and, hence, large-area detectors, as well as its compatibility with low-coherence, compact X-ray sources, both of which are key factors for human-scale imaging. It has already demonstrated strong potential for chest imaging applications, including the diagnosis of pulmonary emphysema, fibrosis, and cancer. To retrieve transmission, differential phase, and dark-field images from data, an algorithm is required to separate the distinct mechanisms contributing to measured contrast. Since its realization, this image-retrieval step has remained fundamentally unchanged. In this work, we develop a novel transmission- and dark-field retrieval algorithm for grating-interferometry derived from the X-ray Fokker-Planck equation. To demonstrate and validate our Fokker-Planck algorithm, we apply it to experimental measurements of a test sample and to data from a mouse chest acquired with varying exposure times and added Poisson noise. The retrieved images were qualitatively and quantitatively compared with those retrieved using a conventional sinusoidal-fitting approach. Across both samples, the Fokker--Planck method produced images consistent with conventional retrieval, with a comparable signal-to-noise ratio. Notably, our Fokker-Planck method suppresses artefacts arising in the conventional approach under grating perturbations (e.g., structural defects like scratches) and reduced flux or visibility, yielding smoother and more reproducible images. Additionally, we demonstrate that our Fokker-Planck method has an advantage over the conventional dark-field retrieval method for fast sample imaging with short exposure times and high noise.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The paper develops a novel retrieval algorithm for transmission and dark-field images in X-ray grating interferometry by solving the Fokker-Planck equation for intensity flow. It reports that the method yields images consistent with the conventional sinusoidal-fitting approach on both a test sample and mouse chest data, with comparable SNR, while suppressing artifacts from grating defects, reduced flux/visibility, and high noise (short exposures), offering advantages for fast imaging.

Significance. If the central claim holds, the approach could improve robustness in clinical grating interferometry applications such as pulmonary imaging, where grating imperfections and low-dose constraints are common. The experimental validation under controlled perturbations and noise is a concrete strength, as is the focus on practical advantages over the long-standard retrieval method.

major comments (2)
  1. [Abstract and Results] Abstract and results (comparison with conventional method): The claims of 'qualitative and quantitative agreement' and 'comparable signal-to-noise ratio' are not supported by any reported numerical error metrics, difference maps, statistical tests, or explicit SNR values; without these, it is impossible to evaluate whether the observed smoothness reflects improved fidelity or implicit regularization.
  2. [Theory] Theory section (Fokker-Planck derivation): The manuscript does not demonstrate that the derived retrieval expressions reduce, by the paper's own equations, to the transmission and dark-field coefficients that would be obtained by fitting the same measured intensities; this leaves open whether unmodeled grating-specific propagation or higher-order scattering terms bias the separation.
minor comments (1)
  1. [Methods] Methods: Specify the precise form of the Fokker-Planck equation employed, including any boundary conditions or approximations for the grating interferometer geometry.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thorough review and constructive feedback on our manuscript. We address each of the major comments below and have made revisions to strengthen the paper as suggested.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and results (comparison with conventional method): The claims of 'qualitative and quantitative agreement' and 'comparable signal-to-noise ratio' are not supported by any reported numerical error metrics, difference maps, statistical tests, or explicit SNR values; without these, it is impossible to evaluate whether the observed smoothness reflects improved fidelity or implicit regularization.

    Authors: We agree with the referee that additional quantitative metrics would enhance the rigor of the comparison. In the revised manuscript, we now include explicit SNR values computed for regions of interest in both the test sample and mouse chest data, difference maps showing the pixel-wise differences between the Fokker-Planck and conventional retrieval methods, and root-mean-square deviation metrics. These additions confirm the qualitative and quantitative agreement while demonstrating that the reduced artifacts and smoothness arise from the method's robustness to noise and perturbations rather than over-regularization. revision: yes

  2. Referee: [Theory] Theory section (Fokker-Planck derivation): The manuscript does not demonstrate that the derived retrieval expressions reduce, by the paper's own equations, to the transmission and dark-field coefficients that would be obtained by fitting the same measured intensities; this leaves open whether unmodeled grating-specific propagation or higher-order scattering terms bias the separation.

    Authors: We thank the referee for highlighting this important point regarding the theoretical consistency. Upon review, we recognize that an explicit limiting case was not presented. In the revised version, we have added a new paragraph in the Theory section that derives the reduction of our Fokker-Planck-based expressions to the conventional sinusoidal fitting coefficients under the assumptions of the standard grating interferometry model (small phase gradients and negligible higher-order scattering). This shows that our method is consistent with the conventional approach in the appropriate limit, with differences arising only from the handling of perturbations and noise. revision: yes

Circularity Check

0 steps flagged

Fokker-Planck derivation yields independent retrieval formulas

full rationale

The paper derives its transmission and dark-field retrieval algorithm directly by solving the X-ray Fokker-Planck equation for intensity flow through the grating interferometer. The resulting closed-form expressions are obtained from the differential equation under stated assumptions rather than by fitting parameters to the measured intensities or by re-expressing fitted quantities as predictions. No self-definitional loops, fitted-input-as-prediction steps, or load-bearing self-citations appear in the derivation chain. Validation consists of qualitative and quantitative comparison against the conventional sinusoidal-fitting method on experimental data, but the formulas themselves are not shown to reduce to those data by construction. This is a standard first-principles derivation with external experimental checks, producing a self-contained result.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the applicability of the Fokker-Planck equation to describe X-ray intensity transport through the grating interferometer and sample. No free parameters, new physical entities, or additional ad-hoc assumptions are mentioned in the abstract.

axioms (1)
  • domain assumption X-ray intensity propagation and scattering in grating interferometry obey the Fokker-Planck equation.
    The entire retrieval algorithm is derived from this equation as stated in the abstract.

pith-pipeline@v0.9.0 · 5626 in / 1358 out tokens · 45231 ms · 2026-05-10T14:54:03.978366+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

74 extracted references · 74 canonical work pages · 1 internal anchor

  1. [1]

    Atsushi Momose et al.Jpn. J. Appl. Phys., 42(7B):L866, 2003

    work page 2003

  2. [2]

    Express, 13(16):6296–6304, 2005

    Timm Weitkamp et al.Opt. Express, 13(16):6296–6304, 2005

    work page 2005

  3. [3]

    Mater., 7:134–137, 2008

    Franz Pfeiffer et al.Nat. Mater., 7:134–137, 2008

    work page 2008

  4. [4]

    Phys., 2(4):258–261, 2006

    Franz Pfeiffer et al.Nat. Phys., 2(4):258–261, 2006

    work page 2006

  5. [5]

    Synchrotron Radiat., 16(1):43–47, 2009

    Martin Bech et al.J. Synchrotron Radiat., 16(1):43–47, 2009

    work page 2009

  6. [6]

    Rep., 7(1):14477, 2017

    Christoph Jud et al.Sci. Rep., 7(1):14477, 2017

    work page 2017

  7. [7]

    Atsushi Momose et al.Philos. Trans. R. Soc. A, 372(2010):20130023, 2014

    work page 2010

  8. [8]

    Health, 3(11):e733–e744, 2021

    Konstantin Willer et al.Lancet Digit. Health, 3(11):e733–e744, 2021

    work page 2021

  9. [9]

    Radiol., 58(11):775–781, 2023

    Theresa Urban et al.Invest. Radiol., 58(11):775–781, 2023

    work page 2023

  10. [10]

    Manuel Viermetz et al.Proc. Natl. Acad. Sci. U.S.A., 119(8):e2118799119, 2022

    work page 2022

  11. [11]

    Chem., 181:118052, 2024

    Zhili Wang.TrAC, Trends Anal. Chem., 181:118052, 2024

    work page 2024

  12. [12]

    Phys., 39(1):424–428, 2012

    Nicholas Bevins et al.Med. Phys., 39(1):424–428, 2012

    work page 2012

  13. [13]

    Phys., 38(7):4133–4140, 2011

    Thomas Weber et al.Med. Phys., 38(7):4133–4140, 2011

    work page 2011

  14. [14]

    Express, 25(6):6349–6364, 2017

    Carolina Arboleda et al.Opt. Express, 25(6):6349–6364, 2017

    work page 2017

  15. [15]

    Rep., 15(1):14223, 2025

    Alexandre Pereira et al.Sci. Rep., 15(1):14223, 2025

    work page 2025

  16. [16]

    Imaging, 4(4):58, 2018

    Volker Ludwig et al.J. Imaging, 4(4):58, 2018

    work page 2018

  17. [17]

    Phys., 52(4):2145–2154, 2025

    Karin Hellerhoff et al.Med. Phys., 52(4):2145–2154, 2025

    work page 2025

  18. [18]

    Sebastian Kaeppler et al.J. Med. Imaging, 4(3):034005, 2017

    work page 2017

  19. [19]

    Phd thesis, University of Helsinki, Helsinki, Finland, 2025

    Henrik M¨ akinen. Phd thesis, University of Helsinki, Helsinki, Finland, 2025

    work page 2025

  20. [20]

    Wolfgang Noichl et al.IEEE Trans. Med. Imaging, 43(1):28–38, 2023

    work page 2023

  21. [21]

    Express, 33(1):1345–1358, 2025

    Simon Spindler et al.Opt. Express, 33(1):1345–1358, 2025

    work page 2025

  22. [22]

    Rep., 14(1):32169, 2024

    Sami Wirtensohn et al.Sci. Rep., 14(1):32169, 2024

    work page 2024

  23. [23]

    Rep., 9(1):17537, 2019

    David M Paganin and Kaye S Morgan.Sci. Rep., 9(1):17537, 2019. 16

    work page 2019

  24. [24]

    Rep., 9(1):17465, 2019

    Kaye S Morgan and David M Paganin.Sci. Rep., 9(1):17465, 2019

    work page 2019

  25. [25]

    M R Teague.J. Opt. Soc. Am., 73(11):1434–1441, 1983

    work page 1983

  26. [26]

    Masanori Mitome.Microscopy, 70(1):69–74, 2021

    work page 2021

  27. [27]

    Lasers Eng., 135:106187, 2020

    Chao Zuo et al.Opt. Lasers Eng., 135:106187, 2020

    work page 2020

  28. [28]

    Microsc., 206(1):33–40, 2002

    David M Paganin et al.J. Microsc., 206(1):33–40, 2002

    work page 2002

  29. [29]

    Thomas A Leatham et al.IEEE Trans. Med. Imaging, 42(6), 2023

    work page 2023

  30. [30]

    Express, 32(3):4588–4602, 2024

    Thomas A Leatham et al.Opt. Express, 32(3):4588–4602, 2024

    work page 2024

  31. [31]

    Jannis N Ahlers et al.Optica, 11(8):1182–1191, 2024

    work page 2024

  32. [32]

    Lett., 50(7):2171–2174, 2025

    Jannis N Ahlers et al.Opt. Lett., 50(7):2171–2174, 2025

    work page 2025

  33. [33]

    Opt., 22(12):125604, 2020

    Konstantin M Pavlov et al.J. Opt., 22(12):125604, 2020

    work page 2020

  34. [34]

    Konstantin M Pavlov et al.Phys. Rev. A, 104(5):053505, 2021

    work page 2021

  35. [35]

    Samantha J Alloo et al.J. Med. Imaging, 9:031502–031502, 2022

    work page 2022

  36. [36]

    Rep., 13(1):5424, 2023

    Samantha J Alloo et al.Sci. Rep., 13(1):5424, 2023

    work page 2023

  37. [37]

    Mario A Beltran et al.Optica, 10(4):422–429, 2023

    work page 2023

  38. [38]

    Express, 33(2):3577–3600, 2025

    Samantha J Alloo et al.Opt. Express, 33(2):3577–3600, 2025

    work page 2025

  39. [39]

    Jayvan Liu et al.Phys. Rev. A, 112(6):063530, 2025

    work page 2025

  40. [40]

    Scr., 100(7):075566, 2025

    Samantha J Alloo and Kaye S Morgan.Phys. Scr., 100(7):075566, 2025

    work page 2025

  41. [41]

    Express, 30(7):10899–10918, 2022

    Ying Ying How and Kaye S Morgan.Opt. Express, 30(7):10899–10918, 2022

    work page 2022

  42. [42]

    Marie-Christine Zdora et al.Phys. Rev. Lett., 118(20):203903, 2017

    work page 2017

  43. [43]

    Rep., 4(1):7243, 2014

    Markus Strobl.Sci. Rep., 4(1):7243, 2014

    work page 2014

  44. [44]

    Synchrotron Radiat., 27:494–502, 2020

    Zhili Wang et al.J. Synchrotron Radiat., 27:494–502, 2020

    work page 2020

  45. [45]

    Express, 31(1):635–650, 2022

    Fabio De Marco et al.Opt. Express, 31(1):635–650, 2022

    work page 2022

  46. [46]

    Werneri A Lindberg et al.Phys. Med. Biol., 70(19):195015, 2025

    work page 2025

  47. [47]

    Alloo et al.J

    Samantha J. Alloo et al.J. Synchrotron Radiat., 33(2), 2026

    work page 2026

  48. [48]

    Rep., 14(1):17807, 2024

    Michelle K Croughan et al.Sci. Rep., 14(1):17807, 2024

    work page 2024

  49. [49]

    Kaye S Morgan et al.Appl. Phys. Lett., 100(12), 2012

    work page 2012

  50. [50]

    Phys., 37(11):6047–6054, 2010

    Eric E Bennett et al.Med. Phys., 37(11):6047–6054, 2010

    work page 2010

  51. [51]

    Lett., 36(1):55–57, 2010

    Kaye S Morgan et al.Opt. Lett., 36(1):55–57, 2010

    work page 2010

  52. [52]

    Konstantin M Pavlov et al.Phys. Rev. Appl., 13(5):054023, 2020

    work page 2020

  53. [53]

    Oxford University Press, 2006

    David M Paganin.Coherent X-ray Optics. Oxford University Press, 2006

    work page 2006

  54. [54]

    Express, 28(11):16363–16384, 2020

    Koh Hashimoto et al.Opt. Express, 28(11):16363–16384, 2020. 17

    work page 2020

  55. [55]

    Synchrotron Rad., 27(5):1395–1414, 2020

    Benedikt G¨ unther et al.J. Synchrotron Rad., 27(5):1395–1414, 2020

    work page 2020

  56. [56]

    Regine Gradl et al.IEEE Trans. Med. Imaging, 38(2):649–656, 2018

    work page 2018

  57. [57]

    A Fokker–Planck approach to grating interferometry

    Samantha J. Alloo. Resources for Alloo et al. “...A Fokker–Planck approach to grating interferometry”.https://zenodo.org/records/18395476, 2026. Accessed: 1-May-2026

    work page arXiv 2026

  58. [58]

    Express, 18(16):16890–16901, 2010

    Wataru Yashiro et al.Opt. Express, 18(16):16890–16901, 2010

    work page 2010

  59. [59]

    Express, 23(7):9233–9251, 2015

    Wataru Yashiro and Atsushi Momose.Opt. Express, 23(7):9233–9251, 2015

    work page 2015

  60. [60]

    Cardiothorac

    Florian T Gassert et al.Radiol. Cardiothorac. Imaging, 7(6):e240560, 2025

    work page 2025

  61. [61]

    Radiol., 4:1487895, 2025

    Florian T Gassert et al.Front. Radiol., 4:1487895, 2025

    work page 2025

  62. [62]

    Michael Chabior et al.J. Appl. Phys., 110(5):053105, 2011

    work page 2011

  63. [63]

    Phys., 44(6):2453–2465, 2017

    Xu Ji et al.Med. Phys., 44(6):2453–2465, 2017

    work page 2017

  64. [64]

    Status Solidi A, 204(8):2746–2752, 2007

    Peter Modregger et al.Phys. Status Solidi A, 204(8):2746–2752, 2007

    work page 2007

  65. [65]

    Rep., 10(1):7890, 2020

    Timur Gureyev et al.Sci. Rep., 10(1):7890, 2020

    work page 2020

  66. [66]

    Martin Bech et al.Phys. Med. Biol., 55(18):5529–5539, 2010

    work page 2010

  67. [67]

    Opt., 50(22):4310–4319, 2011

    Susanna K Lynch et al.Appl. Opt., 50(22):4310–4319, 2011

    work page 2011

  68. [68]

    Lett., 112(6):68002, 2016

    F Prade et al.Europhys. Lett., 112(6):68002, 2016

    work page 2016

  69. [69]

    David M Paganin et al.Phys. Rev. A, 108(1):013517, 2023

    work page 2023

  70. [70]

    Hunter C Meyer et al.arXiv preprint:2509.16503, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  71. [71]

    Jakob Haeusele et al.IEEE Trans. Med. Imaging, 42(10):2876–2885, 2023

    work page 2023

  72. [72]

    Tech., 15(8):937–945, 1980

    Eckhart F¨ orster et al.Krist. Tech., 15(8):937–945, 1980

    work page 1980

  73. [73]

    Phys., 28(8):1610–1619, 2001

    A Olivo et al.Med. Phys., 28(8):1610–1619, 2001

    work page 2001

  74. [74]

    Opt., 22(11):115607, 2020

    David M Paganin et al.J. Opt., 22(11):115607, 2020. 18

    work page 2020