Pith. sign in

REVIEW 3 major objections 6 minor 24 references

Continuous scanning full-field OCT for fast volumetric imaging of multi-cellular aggregates

T0 review · 3 major / 6 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Moving the sample continuously while recording lets full-field OCT reconstruct depth from the per-pixel flicker frequency, producing 400-micron volumes in about 100 seconds with better contrast than four-phase FF-OCT.

desk verdict Genuinely simpler FF-OCT with believable organoid images, but the reconstruction threshold as written doesn't fit typical signal levels and the speed/CNR claims are overreaching. read the letter →

arxiv 2509.08101 v1 pith:OPC7VTWB submitted 2025-09-09 physics.optics

Charlie Kersuzan , Anirban Jana , Amaury Badon This is my paper
classification physics.optics
keywords continuousscanningFF-OCTfull-fieldopticalcoherencetomographytemporalFouriertransformpowerspectraldensityvolumetricimagingmulti-cellularaggregatesorganoidscontrast-to-noiseratio
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes that full-field optical coherence tomography (FF-OCT) can be made much faster and simpler by moving the sample continuously through the focus instead of stepping a piezoelectric mirror. The central claim is that a per-pixel Fourier transform of the recorded intensity turns depth into frequency: because the sample position is z=vt, the interference term oscillates at a known rate, and the power contained at that frequency is proportional to the reflectivity at that depth. The authors demonstrate on capsules containing living and fixed pluripotent-stem-cell aggregates that this 'continuous scanning FF-OCT' reconstructs volumes over 400 µm in 100–200 seconds, that a single en-face image takes about 0.5 seconds to compute, and that the contrast-to-noise ratio exceeds that of a standard four-phase acquisition by about 1.2 dB. If the method holds up, it makes label-free 3D imaging of organoids and tissue constructs practical on an ordinary microscope with no synchronization electronics.

What carries the argument

The central object is the temporal-frequency encoding of depth, carried by the identity z=vt and the interference equation. Named 'continuous scanning FF-OCT', the mechanism replaces piezoelectric phase stepping with a motorized stage and an unsynchronized camera; each pixel's time trace is Fourier-transformed, and the PSD amplitude at the motor-speed-dependent target frequency is proportional to Rsample, the sample reflectivity in the coherence plane. The target frequency is identified on the image-averaged spectrum with a 15%-of-DC threshold, so the method also absorbs slow motor-speed drift.

What would settle it

Place a sub-resolution bead in a gelatin phantom and translate it laterally at 1 µm/s inside the object volume while the stage scans axially at 2 µm/s. If the bead's motion creates a second peak or broadens the per-pixel PSD band at the target frequency such that the 15% threshold merges it, then depth assignment is no longer one-to-one and the claimed rejection of moving particles breaks.

Watch

Extended reading notes

Core claim

The discovery is that depth sectioning in FF-OCT does not require discrete phase shifts: it can be obtained from the temporal frequency content of images recorded while the sample is translated at constant speed. The detected intensity follows I=I0(Rr+Rsample+2√(RrRsample) cos Φ), where Φ evolves linearly in time when the motor moves the sample at speed v. A fast Fourier transform along the time axis of each pixel gives a power spectral density with a peak at the stage frequency, and the sum of PSD components around that peak is assigned as the en-face reflectivity of the current depth plane. In the implementation described, the target frequency is found from the image-averaged spectrum usin

Load-bearing premise

The sample's scattering structure at a given depth does not move or change during the roughly one second it takes to record a sub-stack, nor over the 100–200 second full-volume scan, so the temporal fringe frequency stays tied to stage position.

Editorial extensions

If this is right

  • A full 400 µm volume of a fixed or living aggregate was recorded in 100–200 seconds, making long-term developmental monitoring of organoids feasible on a single setup.
  • Since acquisition speed is limited by camera frame rate rather than piezo settling, faster cameras translate directly into faster volumetric scans, provided enough frames per 2π phase shift are kept.
  • Continuous scanning reached CNR = 17.6 dB versus 16.4 dB for four-phase FF-OCT from a similar frame count, and moving particles produced fewer bright artifacts.
  • With 0.5 seconds per reconstructed en-face image, the method is compatible with live imaging at 1 fps, and the same time traces can feed Welch-method dynamic analysis for intracellular motion contrast.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The stationarity fragility is also a signal: intracellular motion will appear as spectral broadening around the target frequency, so the same dataset could yield a dynamic-OCT contrast that the static reconstruction discards.
  • The 15%-of-DC band-selection rule links depth to a clean spectral peak; a short-time Fourier transform or chirp-matched filter could tolerate motor-speed ripple and allow deliberately non-uniform stage motion, trading simplicity for speed.
  • Because each 100-frame sub-stack already contains a short axial scan, the volume is highly redundant; coherent combination across overlapping sub-stacks could push the SNR beyond the single-band summation reported.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 6 minor

Summary. The paper presents a continuous-scanning variant of full-field OCT in which the sample is translated axially at constant speed while a camera records en-face frames, and depth-resolved reflectivity is recovered by applying a per-pixel FFT and summing PSD components around a Doppler-shifted target frequency. The authors demonstrate the method on living and fixed multicellular aggregates in alginate capsules, compare its CNR to a traditional four-phase FF-OCT acquisition, and claim volumetric imaging of 400 µm in 100–200 s without piezoelectric synchronization.

Significance. The underlying reconstruction idea is self-contained and the proof-of-concept images are visually plausible. Strengths of the manuscript are the simplified experimental design, the transparent per-pixel Fourier processing with no fitted constants, and the potential for genuinely faster volumetric FF-OCT. The method could be a useful addition to the organoid and tissue-imaging toolbox if the reported acquisition and processing details are correct. However, as written, a central algorithmic step (the target-frequency detection threshold) is quantitatively inconsistent with the stated interference model, and the headline CNR advantage rests on a single un-repeated comparison. The live-sample demonstration also raises an unaddressed motion-stationarity concern. These issues prevent acceptance in the current form.

major comments (3)
  1. [§II.B, Eq. (1)] The target-frequency detection threshold is not reproducible as stated. The text says a frequency component is identified as the target if its PSD exceeds 15% of the PSD at zero frequency. For Eq. (1), with I = I0[D + A cos Φ], D = Rr + Rsample, A = 2√(Rr Rsample), the one-sided PSD ratio at f0 to the DC component is approximately (A/2D)² = Rr Rsample/(Rr + Rsample)². For a silicon reference mirror Rr ≈ 0.3 and typical biological Rsample ≈ 10⁻⁴ to 10⁻³, this ratio is ~10⁻⁴ to 10⁻³, orders of magnitude below 0.15. Thus, under the stated criterion, the target peak would never be selected for low-reflectivity samples. The successful images imply either the threshold is misreported, the PSD is computed after DC removal or normalization, or the effective sample reflectivity is much larger than typical biological values. Please specify the exact criterion used, including any DC subtraction or
  2. [§II.D, Fig. 3] The claim of a 'higher contrast-to-noise ratio than traditional four-phase FF-OCT' is not supported by the evidence. The CNR values 16.4 dB and 17.6 dB come from a single pair of images, with no repeated measurements, error bars, or statistical test. The CNR definition in Eq. (2) also depends on manually selected foreground and background regions, which can strongly influence the result. A 1.2 dB difference from one comparison is within plausible measurement uncertainty. Please provide repeated independent acquisitions, report variability, and, if possible, a simple statistical comparison. Alternatively, soften the claim to 'comparable CNR under these conditions.'
  3. [§II.B/E] The reconstruction assumes that the scattering structure at a given depth is stationary during the acquisition of each sub-stack and over the full volume. The mapping z = vt in §II.B and the summation of PSD components around a sharp target frequency are valid only if the scatterer does not move axially or laterally during the recording. This assumption is most fragile for the live-aggregate images in §II.E, where intracellular and organelle motion can broaden or shift the fringe frequency and blur the depth assignment. The manuscript does not quantify this effect or compare live versus fixed samples in terms of spectral linewidth or image sharpness. Please add an explicit discussion, or better, a measurement of the PSD peak width for live samples, and state the resulting depth-resolution limitation.
minor comments (6)
  1. [Abstract / §II.E] The abstract claims volumetric imaging 'within tens of seconds,' but the demonstrated volumes in §II.E are acquired over 200 s at 2 µm/s, and the 400 µm example in §II.B is 100 s at 4 µm/s. Please revise the wording to match the demonstrated times, e.g., 'within a few minutes.'
  2. [Fig. 4 caption] The caption for Fig. 4 does not match the text: the text identifies (A) and (D) as an empty capsule, while the caption calls (A) a fixed cellular aggregate. Please correct the caption.
  3. [References] Reference [12] and reference [21] are the same paper (Morawiec et al., Communications Biology 7, 1057, 2024). Please remove the duplicate.
  4. [Title page] The affiliation contains a typo: 'Numéerique' should be 'Numérique.'
  5. [§II.B] The phrase 'all neighboring components with values greater than 15% of the PSD at the target frequency' does not define what 'neighboring' means in frequency bins. Please state the bandwidth or number of bins included, since this affects the effective axial window and the CNR.
  6. [Eq. (1)] Please define Rr and Rsample explicitly as intensity or amplitude reflectivities. The interference term 2√(Rr Rsample) suggests intensity reflectivities, but the symbols are not defined before use.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; the reconstruction follows the standard interferometric model (Eq. 1) with independently known parameters, and the empirical CNR comparison is measured rather than derived from fitted constants.

full rationale

The paper's central claim is that continuous axial scanning of the sample produces a temporal intensity modulation whose Fourier component at the Doppler frequency encodes depth-resolved reflectivity. This follows directly from Eq. (1), I = I0(Rr + Rsample + 2 sqrt(Rr Rsample) cos(Phi)), with the target frequency determined by the known stage speed and source wavelength, and the z = vt mapping stated in Section II.B. No parameter is fitted to the data to force agreement: the 15% threshold and 0.5% saturation levels are processing choices, not fitted constants, and the CNR comparison in Section II.D is an experimental measurement, not a prediction derived from the model. The self-citations present (Refs. [15], [19], [23], [24]) are background or context citations and are not load-bearing for the reconstruction algorithm or the CNR comparison. The skeptic's concern about the 15%-of-DC threshold being inconsistent with the expected signal level for low-reflectivity biological samples is a correctness and robustness issue, not a circularity issue: even if the threshold were misreported or the algorithm were tuned differently, the derivation would still not reduce to its inputs. Similarly, the sample-motion stationarity assumption noted by the reader is a physical limitation of the approach, not a circular step. Overall, the paper's derivation chain is self-contained against the standard optical interference model and external physical benchmarks, so circularity is minimal.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical entities. The free parameters are image-processing thresholds, not physical constants fitted to produce the claim. The core physical model is standard interferometry.

free parameters (4)
  • frequency identification threshold = 15% of DC PSD
    Chosen by hand for peak detection; affects which frequencies are summed as signal.
  • neighboring component threshold = 15% of target peak PSD
    Chosen by hand to include motor-speed-fluctuation sidebands.
  • saturation percentile = 0.5% top and bottom
    Chosen by hand for normalization and contrast enhancement.
  • sub-stack size = 100 frames (2 µm)
    Chosen as ~4x smaller than coherence length to preserve axial resolution.
assumptions (4)
  • standard math Interference intensity follows I = I0(Rr + Rsample + 2 sqrt(Rr Rsample) cos Φ) (Eq. 1)
    Standard low-coherence interferometry model, used to justify FFT demodulation.
  • domain assumption Motor stage speed is constant so z = vt
    Needed to map time to depth; Fig. 2A shows linear frequency-speed relation supporting this, but any jitter broadens the peak.
  • domain assumption Sample scattering structure is stationary during each sub-stack acquisition
    Required for clean sinusoidal time signal; live samples with intracellular motion could violate this.
  • domain assumption The temporal FFT of the intensity isolates Rsample at the target frequency
    Requires the coherence gate envelope to be approximately constant over the sub-stack window (2 µm << 11 µm coherence length).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Continuous scanning full-field OCT for fast volumetric imaging of multi-cellular aggregates." pith.science (2026). https://pith.science/paper/OPC7VTWB

@misc{pith2026250908101,
  author       = {Pith},
  title        = {Pith review of: Continuous scanning full-field OCT for fast volumetric imaging of multi-cellular aggregates},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OPC7VTWB}},
  note         = {Machine review of arXiv:2509.08101}
}
read the original abstract

Full-field optical coherence tomography (FF-OCT) offers label-free, high-resolution imaging of biological samples but remains limited by slow acquisition due to piezoelectric mirror modulation. We present a continuous-scanning FF-OCT method that eliminates piezoelectric displacement and synchronization by continuously translating the sample with a motorized stage while recording images on the fly. Depth-resolved information is retrieved via Fourier analysis of the temporal signal at each pixel. This approach enables volumetric imaging over several hundred micrometers within tens of seconds and provides a higher contrast-to-noise ratio than traditional four-phase FF-OCT. Continuous-scanning FF-OCT thus represents a simpler and faster alternative for 3D bio-imaging of living tissues and organoids.

Figures

Figures reproduced from arXiv: 2509.08101 by the authors.

Figure 1
Figure 1. Continuous scanning FF-OCT principle. (A) Experimental setup. The sample placed in the object arm is moved [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Effect of stage speed on the performance of the continuous scanning FF-OCT approach. (A) Average power [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Comparison between the four-phase acquisition method and continuous scanning mode in FF-OCT of a living [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Volumetric imaging with continuous scanning FF-OCT. (A) Continuous FF-OCT image of a fixed cellular aggre [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

24 extracted references · 24 canonical work pages

  1. [21]

    Baczewska, K

    M. Baczewska, K. Eder, S. Ketelhut, B. Kemper, and M. Kujawińska, Cytometry Part A99, 388 (2021)

    work page 2021

  2. [22]

    Morawiec, A

    S. Morawiec, A. Ajduk, P. Stremplewski, B. F. Kennedy, and M. Szkulmowski, Communications Bi- ology7, 1057 (2024)

    work page 2024

  3. [1]

    Figure 1(E) shows the resulting image of a living multi-cellular aggregate obtained with this method for a motor speed of 2µm/s and with 200 frames acquired over2seconds

    Last, to compensate for frame to frame insta- bility due to variations in stage speed, the top 0.5% of pixel intensities are saturated, while the bottom 0.5% are set to zero. Figure 1(E) shows the resulting image of a living multi-cellular aggregate obtained with this method for a motor speed of 2µm/s and with 200 frames acquired over2seconds. Thestructur...

    work page

  4. [2]

    D. B. Murphy and M. W. Davidson,Fundamentals of light microscopy and electronic imaging(John Wiley & Sons, 2012)

    work page 2012

  5. [3]

    Huang, E

    D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W.G.Stinson, W.Chang, M.R.Hee, T.Flotte, K.Gre- gory, C. A. Puliafito,et al., science254, 1178 (1991)

    work page 1991

  6. [4]

    E. H. Stelzer, F. Strobl, B.-J. Chang, F. Preusser, S. Preibisch, K. McDole, and R. Fiolka, Nature Re- views Methods Primers1, 73 (2021)

    work page 2021

  7. [5]

    Wilsonet al.,Confocal microscopy, Vol

    T. Wilsonet al.,Confocal microscopy, Vol. 426 (Aca- demic press London, 1990)

    work page 1990

  8. [6]

    Levoy, R

    M. Levoy, R. Ng, A. Adams, M. Footer, and M. Horowitz, inAcm siggraph 2006 papers(2006) pp. 924–934

    work page 2006

Show all 24 references
  1. [7]

    Drexler and J

    W. Drexler and J. G. Fujimoto,Optical coherence to- mography: technology and applications(Springer Sci- ence & Business Media, 2008)

    2008

  2. [8]

    Beaurepaire, A

    E. Beaurepaire, A. C. Boccara, M. Lebec, L. Blanchot, and H. Saint-Jalmes, Optics letters23, 244 (1998)

    1998

  3. [9]

    Dubois, L

    A. Dubois, L. Vabre, A.-C. Boccara, and E. Beaure- paire, Applied optics41, 805 (2002)

    2002

  4. [10]

    Apelian, F

    C. Apelian, F. Harms, O. Thouvenin, and A. C. Boc- cara, Biomedical Optics Express7, 1511 (2016), pub- lisher: Optica Publishing Group

    2016

  5. [11]

    A.LatriveandA.C.Boccara,Biomedicalopticsexpress 2, 2897 (2011)

    2011

  6. [12]

    Scholler, V

    J. Scholler, V. Mazlin, O. Thouvenin, K. Groux, P. Xiao, J.-A. Sahel, M. Fink, C. Boccara, and K. Grieve, Biomedical Optics Express10, 731 (2019)

    2019

  7. [14]

    Scholler, K

    J. Scholler, K. Groux, O. Goureau, J.-A. Sahel, M. Fink, S. Reichman, C. Boccara, and K. Grieve, Light: Science & Applications9, 140 (2020)

    2020

  8. [15]

    Monfort, K

    T. Monfort, K. Grieve, and O. Thouvenin, Optics Let- ters50, 2239 (2025)

    2025

  9. [16]

    Recher, P

    G. Recher, P. Nassoy, and A. Badon, Biomedical optics express11, 2578 (2020)

    2020

  10. [17]

    Auksorius, D

    E. Auksorius, D. Borycki, P. Wegrzyn, B. L. Siko- rski, K. Lizewski, I. Zickiene, M. Rapolu, K. Adomavi- cius, S. Tomczewski, and M. Wojtkowski, IScience25 (2022)

    2022

  11. [18]

    Alessandri, B

    K. Alessandri, B. R. Sarangi, V. V. Gurchenkov, B. Sinha, T. R. Kießling, L. Fetler, F. Rico, S. Scheur- ing, C. Lamaze, A. Simon, S. Geraldo, D. Vignjević, H.Doméjean, L.Rolland, A.Funfak, J.Bibette, N.Bre- mond, and P. Nassoy, Proceedings of the National Academy of Sciences1...

    2013

  12. [19]

    P. J. Cohen, E. Luquet, J. Pletenka, A. Leonard, E. Warter, B. Gurchenkov, J. Carrere, C. Rieu, J. Hardouin, F. Moncaubeig,et al., Biomaterials295, 122033 (2023)

    2023

  13. [20]

    Suire, A

    L. Suire, A. Jana, P. Nassoy, and A. Badon, Advanced Science , e05501 (2025)

    2025

  14. [23]

    Z. Yin, B. He, Y. Ying, S. Zhang, P. Yang, Z. Chen, Z. Hu, Y. Shi, R. Xue, C. Wang, S. Wang, G. Wang, and P. Xue, npj Imaging3, 12 (2025)

    2025

  15. [24]

    Badon, L

    A. Badon, L. Andrique, A. Mombereau, L. Rivet, A. Boyreau, P. Nassoy, and G. Recher, Royal Society Open Science9, 211444 (2022)

    2022

  16. [25]

    A. Jana, N. Mekhileri, A. Boyreau, A. Bazin, N. Pujol, K. Alessandri, G. Recher, P. Nassoy, and A. Badon, PloS one20, e0309035 (2025)

    2025

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.