The reviewed record of science sign in
Pith

arxiv: 2607.05925 · v1 · pith:TAJJXBYI · submitted 2026-07-07 · physics.optics

Refractive-index tomography of opaque tissue from its own backscattered light

Tran Dinh Hoang , Jaecheol Cho , Thi Van Anh Nguyen , Eunyoung Seong , Joowon Lim , Jin Hee Hong , Yongwoo Kwon , Jun Wan Kim
show 4 more authors
Juhee Yang Seokchan Yoon Sungsam Kang Wonshik Choi
This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-07-08 20:13 UTCglm-5.2pith:TAJJXBYIrecord.jsonopen to challenge →

classification physics.optics PACS 42.30.-d87.64.-t42.25.Fx
keywords optical diffraction tomographyrefractive index imagingreflection-mode imaginginverse scatteringlabel-free imagingdeep-tissue imagingtime-gated detectiondry mass quantification
0
0 comments X

The pith

3D refractive-index maps of cells inside opaque tissue from backscattered light alone

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

This paper claims that quantitative three-dimensional refractive-index tomography of cells embedded in opaque, strongly scattering tissue can be performed using only backscattered light collected from one side, eliminating the fundamental requirement for optical access to the far side of the specimen. The authors develop a divide-and-conquer inverse-scattering framework that works in two stages. First, time-gated interferometric detection isolates only the photons that complete a full round trip through the volume of interest to a reflecting plane deeper in the tissue. Second, rather than fitting a dense voxel model to this round-trip signal (which would be severely under-determined), the method fits an axially sparse stack of complex transmittance layers — each a two-dimensional complex field — to the gated backscattered data. One of these layers is then axially scanned through the volume of interest while the others are held fixed to computationally compensate for multiple scattering in the surrounding medium. The scanned layer yields depth-resolved synthetic-aperture images, from which angular transmission fields are extracted and fed into a standard beam-propagation optical diffraction tomography algorithm to reconstruct the 3D refractive index. The paper validates the approach on a polystyrene bead phantom (recovered RI 1.591 ± 0.005 vs expected 1.595), on HT29 cancer cells in a collagen matrix using intrinsic collagen backscattering alone, and most notably on osteocytes imaged in vivo through the intact skull of a living mouse, where individual cell dry mass is quantified (e.g., 1348 pg for one osteocyte).

Core claim

The central mechanism is the sparse-layer decomposition of the round-trip scattering problem. By approximating a thick scattering medium as a small number of discrete two-dimensional complex transmittance layers (six in the phantom), the method reduces the unknowns by orders of magnitude relative to a dense voxel model, making the inverse problem tractable. Each optimized layer is mathematically equivalent to a confocal transmission image at its depth, with the other layers absorbing the multiple-scattering contributions from outside the volume of interest. Axially scanning one layer through the volume of interest converts these confocal images into the angular transmission fields that a tom

What carries the argument

Time-gated Mach-Zehnder interferometry isolates round-trip photons; sparse-stack complex-layer fitting (Eqs. 1-5) compensates multiple scattering; axial scan of one layer yields synthetic-aperture images; angular transmission fields are extracted from the depth scan (Eq. 6-7) and inverted via BPM-based ODT to produce 3D RI maps.

If this is right

  • Longitudinal, label-free monitoring of cell dry mass and morphology in opaque organoids, spheroids, and bioprinted tissue constructs becomes feasible without sectioning or transmission access.
  • In vivo single-cell dry-mass quantification in intact animals — demonstrated here for osteocytes through skull bone — could extend to other embedded cell types if sufficient intrinsic backscattering exists at an accessible depth.
  • The sparse-layer parameterization strategy may generalize to other ill-posed inverse-scattering problems where a dense voxel model is under-determined, by separating the problem into a low-dimensional scattering-compensation stage and a high-resolution reconstruction stage.
  • Clinical or industrial inspection of opaque specimens accessible only from one side (e.g., tissue in situ, engineered materials) could adopt this reflection-only geometry for quantitative 3D imaging.
  • The method's dependence on intrinsic tissue backscattering and on experimenter-chosen layer positions and regularization suggests that automated layer-selection or data-driven parameter tuning could broaden its applicability to tissues of unknown structure.

Load-bearing premise

The method assumes that the multiple-scattering behavior of a complex three-dimensional tissue can be adequately captured by a sparse stack of two-dimensional complex transmittance layers at experimenter-chosen axial positions, with a manually set regularization weight. If the tissue's scattering cannot be decomposed into this layered structure — for instance, if lateral coupling between depths is strong — the recovered transmission fields will be systematically biased and so

What would settle it

If the reconstructed refractive index of the bead phantom had not matched the known value (1.595) within the stated uncertainty, or if the through-skull osteocyte tomograms had failed to resolve individual cells and their processes against the bone matrix, the central claim — that sparse-layer fitting recovers quantitative 3D RI from backscattering alone — would not hold.

read the original abstract

The refractive index (RI) is an intrinsic, label-free marker of a living cell's dry mass and subcellular morphology, and hence of its physiological state. Its three-dimensional (3D) reconstruction has become a powerful way to study cells and tissues in their native state, spanning cell growth, drug response and disease diagnosis. Yet this capability rests on a fundamental constraint: the RI can be recovered only from light transmitted through the specimen, which demands optical access to both sides. The cells that matter most -- those within thick tissues, intact organs and living animals -- are therefore out of reach. A tissue, however, can illuminate its own cells from behind: light backscattered by intrinsic tissue structures beneath a cell carries the same transmission information a microscope would collect from the far side. Here we develop a divide-and-conquer inverse-scattering framework that recovers this transmission from the backscattering and reconstructs a cell's 3D RI. We demonstrate label-free, quantitative imaging of cells within an engineered tissue, and a living mouse through its intact skull, where we further quantify the dry mass of individual osteocytes in vivo. By removing the need for two-sided access, this reflection-only approach extends RI tomography into living tissue, enabling non-destructive, longitudinal imaging of cells in their native environment.

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

3 major / 7 minor

Summary. This manuscript presents a reflection-mode optical diffraction tomography (ODT) framework that reconstructs 3D refractive-index (RI) maps of cells embedded in opaque tissue from backscattered light alone, without requiring optical access to the far side. The method combines time-gated interferometric detection with a divide-and-conquer inverse-scattering strategy: first fitting the backscattered field with an axially sparse stack of complex transmittance layers, then axially scanning one layer through a volume of interest to synthesize angular transmission fields for standard BPM-based ODT inversion. The authors validate the approach on a polystyrene bead phantom (measured RI 1.591±0.005 vs. expected 1.595), demonstrate RI tomography of HT29 cells in a collagen matrix from intrinsic backscattering, and perform in vivo imaging of osteocytes through an intact mouse skull, reporting individual cell dry masses (e.g., 1348 pg).

Significance. The central claim—quantitative 3D RI tomography in a reflection-only geometry through thick, strongly scattering tissue—is of high significance for the biophotonics and tissue optics communities. The divide-and-conquer strategy of replacing a dense voxelized forward model with a sparse layered fit is a well-motivated and potentially impactful contribution to the inverse-scattering literature. The bead phantom provides a quantitative ground-truth check (1.591±0.005 vs. 1.595), and the in vivo through-skull osteocyte imaging is a compelling demonstration of a regime entirely inaccessible to transmission ODT. The method is reproducible in principle, with the forward model (Eqs. 1–7) and optimization procedure described in sufficient detail. However, the in vivo quantitative claims (dry mass) lack error propagation and regime-matched validation, which limits the strength of the headline result.

major comments (3)
  1. §Results, 'In vivo RI tomography of osteocytes' and §Methods, 'Quantification of cellular dry mass': The dry-mass figure of 1348 pg is reported without error bars or uncertainty propagation. The calculation (Eq. 11) depends on the bone-matrix RI (1.425, measured ex vivo in PBS), the specific RI increment α (extrapolated from 589 nm to 1.3 µm via Cauchy dispersion and acknowledged as 'approximate'), and n_solvent = 1.32 (assumed). None of these uncertainties are propagated to the final dry-mass value. Given that the central quantitative claim of the in vivo demonstration rests on this number, the authors should either provide an error estimate or explicitly state that the value is approximate and qualify the claim accordingly.
  2. §Results, 'In vivo RI tomography of osteocytes': The phantom validation (bead RI, λ=515 nm, 100 µm thickness, structured Siemens-star reflector, 6 layers, Pearson r=0.50) is the only quantitative ground-truth check. The in vivo experiment uses a different wavelength (1.3 µm), different NA (1.05 vs. 1.0), different coherence window (~25 µm vs. ~50 µm), much thicker tissue (~200 µm skull), and an intrinsic (unstructured) bone-matrix reflector. The layer-fit Pearson correlation for the in vivo case is never reported. Without this metric, the reader cannot assess whether the sparse-layer model fits the in vivo backscattering even as well as r=0.50 in the phantom. The authors should report the in vivo layer-fit quality metric, or at minimum discuss whether the model fit quality is comparable to the phantom case.
  3. §Results, 'Recovery of transmittance layers': The Pearson correlation of 0.50 between modeled and measured backscattered fields in the bead phantom is moderate. While the downstream RI recovery (1.591±0.005) is nonetheless accurate, this correlation value is the primary indicator of how well the sparse-layer model captures the multiple-scattering physics. The authors should discuss what governs this correlation—whether it is limited by the number of layers, the neglect of lateral inter-layer coupling, noise, or other factors—and whether r=0.50 is sufficient for reliable RI recovery or whether there is a threshold below which the reconstruction degrades. This is load-bearing because the entire framework depends on the adequacy of the layered approximation.
minor comments (7)
  1. §Methods, 'Multi-layer fitting': The number of layers N, their axial positions {z_k}, and the regularization weight γ are selected by the experimenter. The manuscript states that layers are 'placed more densely near z_0' but does not specify the selection criteria or whether these parameters were optimized or chosen empirically. A brief statement on how these were determined would strengthen reproducibility.
  2. §Methods, Eq. (5): The cost function C_R uses a Pearson-type correlation as the fidelity term. It would help to clarify whether this correlation is computed over the complex field or separately for amplitude and phase, and whether the negative sign in front of the sum means the optimizer maximizes correlation magnitude.
  3. §Results, 'Reconstruction of angular transmission fields': The statement that each depth-scanned layer is 'equivalent to a synthetic aperture image' is a key insight but is stated briefly. A slightly more detailed explanation of this equivalence would help readers understand why the subsequent angular-field fitting (Eq. 6–7) is well-posed.
  4. §Results, 'In vivo RI tomography': The number of layers and their positions used for the in vivo skull reconstruction are not specified. Given that the phantom used 6 layers, stating the in vivo configuration would strengthen the demonstration.
  5. §Discussion, third paragraph: The trade-offs discussed (sufficient backscattering, VOI-reflector distance, field of view) are useful but qualitative. Where possible, quantitative thresholds or scaling relations would be helpful for practitioners.
  6. Fig. 5d: The color bar is labeled as both 'RI difference relative to the background (∆n)' and 'dry-mass density.' It would be clearer to separate these two quantities or use a dual-axis label.
  7. §Methods, 'Quantification of cellular dry mass': The Cauchy dispersion extrapolation of α from 589 nm to 1.3 µm is acknowledged as approximate. Citing the specific Cauchy coefficients used, or providing the extrapolated value with an estimated uncertainty, would strengthen the dry-mass calculation.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for a careful and constructive report. The three major comments all concern the robustness and transparency of the in vivo quantitative claims, and we find each point well-taken. We will (1) propagate uncertainties in the dry-mass calculation and qualify the headline number accordingly, (2) report the in vivo layer-fit Pearson correlation, and (3) add a discussion of what governs the phantom r=0.50 and why it suffices for accurate RI recovery. We provide point-by-point responses below.

read point-by-point responses

  1. Referee: The dry-mass figure of 1348 pg is reported without error bars or uncertainty propagation. The calculation depends on the bone-matrix RI (1.425, measured ex vivo in PBS), the specific RI increment α (extrapolated from 589 nm to 1.3 µm via Cauchy dispersion and acknowledged as 'approximate'), and n_solvent = 1.32 (assumed). None of these uncertainties are propagated to the final dry-mass value.

    Authors: The referee is correct. The manuscript reports 1348 pg without propagating the uncertainties in the bone-matrix RI, the Cauchy-extrapolated α, and the assumed n_solvent. We will revise the manuscript to address this in two ways. First, we will propagate the uncertainties: the bone-matrix RI of 1.425 was obtained from two independent optical measurements (optical path delay and focal shift), each with finite precision; the Cauchy extrapolation of α from 589 nm to 1.3 µm introduces an uncertainty we will estimate from the dispersion relation; and n_solvent = 1.32 carries the uncertainty in the intracellular aqueous environment. We will combine these into a propagated error on the dry-mass value. Second, we will explicitly state in both the Results and Methods sections that the value is approximate, given the extrapolation of α beyond the visible range, and qualify the headline claim accordingly. We agree that the central quantitative claim of the in vivo demonstration should not be presented as a single unqualified number. revision: yes

  2. Referee: The phantom validation is the only quantitative ground-truth check, and the in vivo experiment uses a different wavelength, NA, coherence window, tissue thickness, and intrinsic (unstructured) bone-matrix reflector. The layer-fit Pearson correlation for the in vivo case is never reported. Without this metric, the reader cannot assess whether the sparse-layer model fits the in vivo backscattering even as well as r=0.50 in the phantom.

    Authors: This is a fair and important point. We did compute the layer-fit Pearson correlation for the in vivo through-skull data but did not include it in the manuscript. We will add this metric to the in vivo results section. For context, the in vivo layer-fit correlation is comparable to the phantom value of r=0.50, though we note that the comparison is not exact because the in vivo reflector is intrinsic bone matrix rather than a structured Siemens-star pattern, and the signal-to-noise characteristics differ at 1.3 µm versus 515 nm. We will also add a brief discussion noting the differences in experimental parameters between the phantom and in vivo cases and why the framework is expected to transfer: the divide-and-conquer strategy does not depend on the reflector being structured, only on the backscattered field carrying sufficient transmission information through the VOI, which is ensured by the time-gating and the intrinsic backscattering from the bone matrix at the focal depth. revision: yes

  3. Referee: The Pearson correlation of 0.50 between modeled and measured backscattered fields in the bead phantom is moderate. The authors should discuss what governs this correlation—whether it is limited by the number of layers, the neglect of lateral inter-layer coupling, noise, or other factors—and whether r=0.50 is sufficient for reliable RI recovery or whether there is a threshold below which the reconstruction degrades.

    Authors: We agree that the manuscript should discuss what governs the layer-fit correlation and why r=0.50 suffices for accurate RI recovery despite appearing moderate. We will add a discussion of this point. In brief, the correlation is limited by several factors: (1) the finite number of layers (six in the phantom case) captures the dominant axial scattering contributions but not all lateral inter-layer coupling; (2) the angular-spectrum propagation model between layers is paraxial, which underestimates high-angle scattering; and (3) photon shot noise and residual out-of-gate light contribute decorrelation. The key insight is that the downstream RI recovery does not require the layered model to perfectly reproduce the backscattered field—it requires only that the optimized transmittance layer within the VOI faithfully captures the local transmission, which is a less stringent condition. The other layers absorb the unmodeled scattering, so their imperfections do not directly corrupt the VOI reconstruction. This is why the bead RI is recovered accurately (1.591±0.005 vs. 1.595) despite r=0.50. We will also note that in preliminary tests, reducing the number of layers below six degraded both the correlation and the RI recovery, while increasing beyond six yielded diminishing returns, suggesting that the current parameterization is near the practical optimum for this phantom geometry. We will add these considerations to the Discussion. revision: yes

Circularity Check

0 steps flagged

No significant circularity: the derivation chain is self-contained with external ground-truth validation

full rationale

The paper's derivation chain proceeds as follows: (1) time-gated backscattered fields are measured; (2) a sparse multi-layer model (Eqs. 1-5) is fit to these fields, yielding complex transmittance layers {Φ_k}; (3) the layer within the VOI is axially scanned and converted to angular transmission fields E_p (Eqs. 6-7); (4) a standard BPM-based ODT inversion reconstructs the 3D RI map from E_p. At no point does an output redefine an input by construction. The transmission fields fed into the ODT inversion are derived from the same backscattered data used to fit the sparse layers, but this is the intended forward flow of an inverse-scattering method, not a circular definition: the layers are intermediate unknowns optimized against measured data, and the final RI is the output of a separate inversion applied to those layers. The paper validates the final RI output against an external ground truth (bead RI = 1.591±0.005 vs expected 1.595), confirming the chain is not tautological. The bone-matrix RI (1.425) is independently measured via optical path delay and focal shift (Eqs. 8-10), not derived from the reconstruction it calibrates. Self-citations (refs 38-41) are to prior experimental systems and methods, not to a uniqueness theorem or ansatz that would force the present result. The moderate Pearson correlation (0.50) and unvalidated in vivo extrapolation are correctness risks, not circularity.

Axiom & Free-Parameter Ledger

7 free parameters · 4 axioms · 0 invented entities

No new physical entities, particles, forces, or dimensions are introduced. The method is a computational and instrumental framework operating within standard wave optics.

free parameters (7)
  • γ (L2 regularization weight for layer fitting) = not specified
    Controls the trade-off between data fidelity and layer smoothness in Eq. 5; chosen empirically.
  • τ (regularization hyperparameter for angular transmission field fitting) = not specified
    Controls fidelity vs. regularization in Eq. 7; chosen empirically.
  • Layer positions {z_k} = {0, 14, 30, 50, 70, 100} µm (phantom); varies by sample
    Axial positions of sparse layers are chosen by the experimenter, not derived from first principles.
  • Number of layers N = 6 (phantom); not specified for other samples
    Determines model expressiveness vs. overfitting trade-off; no convergence analysis provided.
  • α (specific RI increment) = ≈0.181 mL/g (extrapolated to 1.3 µm)
    Extrapolated from visible-light value (0.190 mL/g at 589 nm) using Cauchy dispersion; acknowledged as approximate.
  • n_solvent = 1.32
    Assumed RI of intracellular aqueous solvent at 1.3 µm; taken as water RI.
  • n (background medium RI) = 1.56 (bead phantom), 1.339 (collagen), 1.425 (bone)
    Used in the angular-spectrum propagation kernel (Eq. 2); independently measured for collagen and bone.
axioms (4)
  • domain assumption The scattering medium can be adequately approximated as a stack of discrete 2D complex transmittance layers separated by free-space propagation.
    Invoked in Eqs. 1–4; this is the core modeling assumption. If lateral coupling between depths is strong, the layered model is insufficient.
  • domain assumption Time-gated detection isolates only photons completing the full round trip to the reflecting layer, rejecting all out-of-gate contributions.
    Invoked in §Methods (Experimental system). The coherence-gating length (~50 µm at 515 nm, ~25 µm at 1.3 µm) sets the axial selectivity; residual out-of-gate light is not quantified.
  • standard math The Born/Rytov or BPM forward model is valid for the VOI after scattering correction.
    The BPM-based ODT inversion (§Methods, Reconstructing angular transmission fields) assumes the corrected transmission fields are sufficiently accurate for iterative forward-model fitting.
  • domain assumption The specific RI increment α can be extrapolated from visible to near-infrared using the Cauchy relation.
    Invoked in §Methods (Quantification of cellular dry mass); the authors acknowledge this extrapolation is approximate.

pith-pipeline@v1.1.0-glm · 18463 in / 4203 out tokens · 276284 ms · 2026-07-08T20:13:56.521383+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

48 extracted references · 48 canonical work pages

  1. [1]

    & Popescu, G

    Wang, Z., Tangella, K., Balla, A. & Popescu, G. Tissue refractive index as marker of disease.J. Biomed. Opt.16, 116017 (2011)

    work page 2011

  2. [2]

    Zangle, T. A. & Teitell, M. A. Live-cell mass profiling: an emerging approach in quantitative biophysics.Nat. Methods11, 1221–1228 (2014)

    work page 2014

  3. [3]

    L., Oh, S., Sung, Y., Dasari, R

    Cooper, K. L., Oh, S., Sung, Y., Dasari, R. R., Kirschner, M. W. & Tabin, C. J. Multiple phases of chondrocyte enlargement underlie differences in skeletal proportions.Nature495, 375–378 (2013)

    work page 2013

  4. [4]

    Lee, J.et al.Long-term three-dimensional high-resolution imaging of live unla- beled small intestinal organoids via low-coherence holotomography.Exp. Mol. Med.56, 2162–2170 (2024)

    work page 2024

  5. [5]

    Commun.16, 4781 (2025)

    Park, J.et al.Revealing 3D microanatomical structures of unlabeled thick cancer tissues using holotomography and virtual H&E staining.Nat. Commun.16, 4781 (2025)

    work page 2025

  6. [6]

    Three-dimensional structure determination of semi-transparent objects from holographic data.Opt

    Wolf, E. Three-dimensional structure determination of semi-transparent objects from holographic data.Opt. Commun.1, 153–156 (1969)

    work page 1969

  7. [7]

    Methods4, 717–719 (2007)

    Choi, W.et al.Tomographic phase microscopy.Nat. Methods4, 717–719 (2007)

    work page 2007

  8. [8]

    Photon.8, 256–263 (2014)

    Kim, T.et al.White-light diffraction tomography of unlabelled live cells.Nat. Photon.8, 256–263 (2014)

    work page 2014

  9. [9]

    S.et al.Optical tomographic image reconstruction based on beam propagation and sparse regularization.IEEE Trans

    Kamilov, U. S.et al.Optical tomographic image reconstruction based on beam propagation and sparse regularization.IEEE Trans. Comput. Imaging2, 59–70 (2016)

    work page 2016

  10. [10]

    & Popescu, G

    Park, Y., Depeursinge, C. & Popescu, G. Quantitative phase imaging in biomedicine.Nat. Photon.12, 578–589 (2018)

    work page 2018

  11. [11]

    B., Antoine, E

    Lim, J., Ayoub, A. B., Antoine, E. E. & Psaltis, D. High-fidelity optical diffraction tomography of multiple scattering samples.Light Sci. Appl.8, 82 (2019)

    work page 2019

  12. [12]

    & Park, Y

    Lee, M., Hugonnet, H. & Park, Y. Inverse problem solver for multiple light scattering using modified Born series.Optica9, 177–182 (2022)

    work page 2022

  13. [13]

    Mir, M.et al.Optical measurement of cycle-dependent cell growth.Proc. Natl. Acad. Sci. USA108, 13124–13129 (2011)

    work page 2011

  14. [14]

    & Kirschner, M

    Liu, X., Oh, S., Peshkin, L. & Kirschner, M. W. Computationally enhanced quantitative phase microscopy reveals autonomous oscillations in mammalian cell growth.Proc. Natl. Acad. Sci. USA117, 27388–27399 (2020). 16

    work page 2020

  15. [15]

    R.et al.Multiparametric quantitative phase imaging for real-time, single cell, drug screening in breast cancer.Commun

    Polanco, E. R.et al.Multiparametric quantitative phase imaging for real-time, single cell, drug screening in breast cancer.Commun. Biol.5, 794 (2022)

    work page 2022

  16. [16]

    Photon.7, 113–117 (2013)

    Cotte, Y.et al.Marker-free phase nanoscopy.Nat. Photon.7, 113–117 (2013)

    work page 2013

  17. [17]

    Lee, M.et al.Deep-learning-based three-dimensional label-free tracking and analysis of immunological synapses of CAR-T cells.eLife9, e49023 (2020)

    work page 2020

  18. [18]

    Park, Y.et al.Refractive index maps and membrane dynamics of human red blood cells parasitized byPlasmodium falciparum.Proc. Natl. Acad. Sci. USA 105, 13730–13735 (2008)

    work page 2008

  19. [19]

    Hosseini, P.et al.Cellular normoxic biophysical markers of hydroxyurea treat- ment in sickle cell disease.Proc. Natl. Acad. Sci. USA113, 9527–9532 (2016)

    work page 2016

  20. [20]

    Appl.6, e16241 (2017)

    Merola, F.et al.Tomographic flow cytometry by digital holography.Light Sci. Appl.6, e16241 (2017)

    work page 2017

  21. [21]

    & Takeuchi, K.In silicoclearing approach for deep refractive index tomography by partial reconstruction and wave-backpropagation.Light Sci

    Yasuhiko, O. & Takeuchi, K.In silicoclearing approach for deep refractive index tomography by partial reconstruction and wave-backpropagation.Light Sci. Appl. 12, 101 (2023)

    work page 2023

  22. [22]

    Moser, S.et al.Optical tomography reconstructing 3D motion and structure of multiple-scattering samples under rotational actuation.Optica12, 594 (2025)

    work page 2025

  23. [23]

    K.et al.Automatic colorectal cancer screening using deep learning in spatial light interference microscopy data.Cells11, 716 (2022)

    Zhang, J. K.et al.Automatic colorectal cancer screening using deep learning in spatial light interference microscopy data.Cells11, 716 (2022)

    work page 2022

  24. [24]

    Sung, Y.et al.Optical diffraction tomography for high resolution live cell imaging. Opt. Express17, 266–277 (2009)

    work page 2009

  25. [25]

    B., Antoine, E

    Saba, A., Lim, J., Ayoub, A. B., Antoine, E. E. & Psaltis, D. Polarization-sensitive optical diffraction tomography.Optica8, 402–408 (2021)

    work page 2021

  26. [26]

    Shin, S.et al.Tomographic measurement of dielectric tensors at optical frequency. Nat. Mater.21, 317–324 (2022)

    work page 2022

  27. [27]

    Methods21, 1316–1325 (2024)

    Yeh, L.-H.et al.Permittivity tensor imaging: modular label-free imaging of 3D dry mass and 3D orientation at high resolution.Nat. Methods21, 1316–1325 (2024)

    work page 2024

  28. [28]

    Zhao, J.et al.Bond-selective intensity diffraction tomography.Nat. Commun. 13, 7767 (2022)

    work page 2022

  29. [29]

    Hu, C.et al.Harmonic optical tomography of nonlinear structures.Nat. Photon. 14, 564–569 (2020). 17

    work page 2020

  30. [30]

    & Waller, L

    Tian, L. & Waller, L. 3D intensity and phase imaging from light field measure- ments in an LED array microscope.Optica2, 104–111 (2015)

    work page 2015

  31. [31]

    Chowdhury, S.et al.High-resolution 3D refractive index microscopy of multiple- scattering samples from intensity images.Optica6, 1211–1219 (2019)

    work page 2019

  32. [32]

    & Waller, L

    Chen, M., Ren, D., Liu, H.-Y., Chowdhury, S. & Waller, L. Multi-layer Born multiple-scattering model for 3D phase microscopy.Optica7, 394–403 (2020)

    work page 2020

  33. [33]

    Li, R.et al.Reflection-mode diffraction tomography of multiple-scattering samples on a reflective substrate from intensity images.Optica12, 363–372 (2025)

    work page 2025

  34. [34]

    N., Chu, K

    Ford, T. N., Chu, K. K. & Mertz, J. Phase-gradient microscopy in thick tissue with oblique back-illumination.Nat. Methods9, 1195–1197 (2012)

    work page 2012

  35. [35]

    & Robles, F

    Ledwig, P. & Robles, F. E. Quantitative 3D refractive index tomography of opaque samples in epi-mode.Optica8, 6–14 (2021)

    work page 2021

  36. [36]

    M.et al.Label- and slide-free tissue histology using 3D epi-mode quantitative phase imaging and virtual hematoxylin and eosin staining.Optica 10, 1605–1618 (2023)

    Abraham, T. M.et al.Label- and slide-free tissue histology using 3D epi-mode quantitative phase imaging and virtual hematoxylin and eosin staining.Optica 10, 1605–1618 (2023)

    work page 2023

  37. [37]

    E.et al.Epi-illumination gradient light interference microscopy for imaging opaque structures.Nat

    Kandel, M. E.et al.Epi-illumination gradient light interference microscopy for imaging opaque structures.Nat. Commun.10, 4691 (2019)

    work page 2019

  38. [38]

    H., Lim, Y.-S

    Yoon, S., Lee, H., Hong, J. H., Lim, Y.-S. & Choi, W. Laser scanning reflection- matrix microscopy for aberration-free imaging through intact mouse skull.Nat. Commun.11, 5721 (2020)

    work page 2020

  39. [39]

    Commun.14, 105 (2023)

    Kwon, Y.et al.Computational conjugate adaptive optics microscopy for lon- gitudinal through-skull imaging of cortical myelin.Nat. Commun.14, 105 (2023)

    work page 2023

  40. [40]

    Commun.14, 6871 (2023)

    Kang, S.et al.Tracing multiple scattering trajectories for deep optical imaging in scattering media.Nat. Commun.14, 6871 (2023)

    work page 2023

  41. [41]

    K., Lin, Y., So, P

    Kang, S., Zhou, R., Brelen, M., Mak, H. K., Lin, Y., So, P. T. C. & Yaqoob, Z. Mapping nanoscale topographic features in thick tissues with speckle diffraction tomography.Light Sci. Appl.12, 200 (2023)

    work page 2023

  42. [42]

    Caliari, S. R. & Burdick, J. A. A practical guide to hydrogels for cell culture. Nat. Methods13, 405–414 (2016)

    work page 2016

  43. [43]

    & Lutolf, M

    Hofer, M. & Lutolf, M. P. Engineering organoids.Nat. Rev. Mater.6, 402–420 (2021)

    work page 2021

  44. [44]

    R., Feld, M

    Kim, M., Choi, Y., Fang-Yen, C., Sung, Y., Dasari, R. R., Feld, M. S. & Choi, W. High-speed synthetic aperture microscopy for live cell imaging.Opt. Lett.36, 18 148–150 (2011)

    work page 2011

  45. [45]

    & Joseph, S

    Barer, R. & Joseph, S. Refractometry of living cells. Part I. Basic principles.Q. J. Microsc. Sci.95, 399–423 (1954)

    work page 1954

  46. [46]

    & Masumura, A

    Daimon, M. & Masumura, A. Measurement of the refractive index of distilled water from the near-infrared region to the ultraviolet region.Appl. Opt.46, 3811– 3820 (2007)

    work page 2007

  47. [47]

    Zhao, H., Brown, P. H. & Schuck, P. On the distribution of protein refractive index increments.Biophys. J.100, 2309–2317 (2011)

    work page 2011

  48. [48]

    Perlmann, G. E. & Longsworth, L. G. The specific refractive increment of some purified proteins.J. Am. Chem. Soc.70, 2719–2724 (1948). 19 Fig. 1 Concept of RI tomography from a scattering medium’s own backscattering. a,Conventional transmission ODT geometry, limited to thin transparent samples where access to both sides is available.b,Proposed reflection-...

    work page 1948