{"id":"7ab952d9-12ad-475e-ab6e-e89057789ffb","arxiv_id":"2507.09207","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"A method estimates thickness and stiffness of a soft layer over a stiff base from the dispersion of surface waves captured in video.","lead":"This paper shows that a video of ripples on a soft surface can reveal how thick and how stiff the layer beneath the surface is. It could make tissue stiffness measurements possible with just a camera, which may enable low-cost health monitoring.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The real-data stiffness validation is weakened by a mismatch between rheometry ground truth (10–100 Hz) and the excitation band (40–200 Hz) in a viscoelastic material, so the claimed 1.2% agreement with the rheometry range does not yet establish ground-truth accuracy.","rationale":"The reader identified model mismatch as the weakest assumption and mentioned the rheometry frequency mismatch in the rationale, but did not elevate it to the primary load-bearing concern. I agree that the two-layer uniform model is a simplification, but the more immediate threat to the central claim is that the real-data stiffness validation compares quantities measured in different frequency regimes for a frequency-dependent material. This is not an external-consensus disagreement; it is an internal comparison problem. The paper's own text flags the frequency mismatch (Sec. 5.2, Fig. 7 caption) yet still claims 'within 1.2% error of the rheometry range,' which is not a sound ground-truth accuracy statement unless the rheometry values are matched to the excitation band. The proposed concrete test—matching rheometry frequencies or incorporating a viscoelastic forward model—would settle whether the 1.2% number is meaningful. If the matched-frequency comparison fails, the abstract's 'strong agreement with ground-truth measurements' overstates the evidence. Because the reader already recommended a conditional acceptance, my finding does not change the verdict; it sharpens the condition under which acceptance should be granted: the stiffness validation must be re-run with frequency-matched ground truth or reworded to avoid claiming accuracy against a mismatched reference.","tokens_in":13054,"tokens_out":6197,"duration_ms":79177,"concrete_test":"Perform oscillatory rheometry on the same gelatin phantoms over the full excitation band (40–200 Hz) and compare the VSWE-estimated stiffness to the rheometry storage modulus at the same frequencies. If the VSWE estimate falls within the stated 1.2% tolerance of this matched-frequency ground truth, the concern is resolved. If the rheometer cannot reach 200 Hz, use a viscoelastic forward model that incorporates the measured frequency-dependent complex modulus and check whether the same (T, E) parameters are recovered from the observed dispersion relation.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of strong agreement with ground-truth measurements on real data rests on the stiffness comparison in Sec. 5.2 and Fig. 7. The paper states that rheometry measurements were taken only between 10–100 Hz while the excitation signal ranged from 40–200 Hz, and it acknowledges that the modulus of hydrogels can depend on frequency (Ref. [48]). The forward model in Sec. 3.2, however, assumes a single frequency-independent elastic modulus E. Therefore, the VSWE estimate is an effective modulus at the wave frequencies (40–200 Hz), whereas the rheometry ground truth is measured at lower frequencies (10–100 Hz). In a viscoelastic material, these are not directly comparable; a difference of more than 1.2% between moduli over this frequency span would invalidate the claimed accuracy metric. The phrase 'within 1.2% error of the rheometry range' is also ambiguous about whether the comparison is to the range of rheometry values or to individual measurements. Because the abstract's 'strong agreement with ground-truth measurements' depends on this comparison, the stiffness validation on real phantoms is not yet established. The thickness validation via calipers is more secure, but the stiffness claim is load-bearing for the paper's contribution.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces Visual Surface Wave Elastography (VSWE), a video-based method for estimating the thickness and stiffness of a soft tissue layer over a stiff base. The pipeline extracts sub-pixel surface motion with phase-based processing, forms an observed dispersion relation by two-dimensional FFT of the motion field, and fits (T, E) by maximizing SSIM between the observed dispersion image and dispersion curves computed from an FEM eigenproblem. The method is validated on COMSOL plane-strain simulations, on real gelatin phantoms with caliper and rheometry ground truth, and on a full three-dimensional anatomical leg simulation. The paper also includes ablation studies of the objective function and mesh resolution, plus a characteristic-number analysis for transferring the method to other parameter regimes.","tokens_in":13305,"tokens_out":4748,"duration_ms":57318,"significance":"If the real-data stiffness validation is placed on a firmer footing, this is a valuable contribution. The central inversion is genuinely physics-based and not circular: the dispersion relation is measured from video independently of the fitted parameters, and the forward model is derived from elasticity rather than learned from the data. The thickness validation on real phantoms is convincing, with VSWE estimates tracking caliper-based confidence intervals across three thicknesses and a temperature scan. The three-dimensional leg simulation is a useful model-mismatch test, and the characteristic-number discussion provides practical guidance for scaling the method. The principal weakness is that the headline 'strong agreement with ground-truth measurements' rests on a stiffness comparison whose frequency bands do not match.","major_comments":[{"comment":"The claimed 1.2% agreement between VSWE stiffness estimates and rheometry ground truth is not yet established. The rheometry measurements were made at 10-100 Hz while the excitation signal spans 40-200 Hz, and the forward model assumes a single frequency-independent modulus E. In a viscoelastic hydrogel the modulus can depend on frequency, as the paper itself notes via Ref. [48], so the VSWE estimate is an effective modulus over the excitation band and can differ from the rheometry value for reasons unrelated to inversion accuracy. In addition, the phrase 'within 1.2% error of the rheometry range' is ambiguous: it is not clear whether the comparison is made against the range of rheometry values or against individual rheometry measurements. Please state the exact comparison protocol, report rheometry data at overlapping frequencies if available (e.g., 40-100 Hz), and quantify the resulting VSWE accuracy. Until then, the abstract's 'strong agreement with ground-truth measurements' overstates what is demonstrated for stiffness.","section":"Sec. 5.3, Fig. 8"},{"comment":"The simulated 3D leg experiment is presented only qualitatively for stiffness ('we also recovered the constant stiffness throughout the calf'), and the thickness comparison is described as 'roughly agrees' without numerical error. Since this experiment is the main test of robustness to model mismatch (curved surface, nonuniform thickness, full 3D physics), please report quantitative errors for both inferred T and E against the known simulation ground truth, for each observation window. Without such numbers, the strength of the model-mismatch validation cannot be assessed.","section":"Sec. 5.3, Fig. 8"}],"minor_comments":[{"comment":"The Gaussian kernel width σ in Eq. (7) is a free parameter that shapes the SSIM objective, but its value and sensitivity are never reported. Please state the value used and show that the inferred parameters are stable over a reasonable range of σ.","section":"Sec. 4.2 and Appendix B"},{"comment":"The sensitivity experiment is under-specified: '5 distinct thicknesses and 4 distinct moduli' is followed by '9 clusters with 9 simulated samples each,' but the organization of clusters and the assignment of ±5% and ±10% perturbations are not defined. Please clarify the design and report a quantitative error metric (e.g., mean absolute percentage error) in addition to Fig. 5.","section":"Sec. 5.1"},{"comment":"The row-wise averaging mixes horizontal and vertical displacement magnitudes; please clarify whether the two components are normalized before averaging and whether image-space pixel units affect the observed dispersion relation.","section":"Sec. 4.1.2, Eq. (4)"},{"comment":"The appendix ends with an incomplete sentence ('The compu'), which appears to be a truncated passage; please complete or remove it. There are also typos elsewhere, such as 'characterizaton' in Sec. 2.2 and 'VSWT' at the start of Appendix A, which should be corrected.","section":"Appendix A"}],"recommendation":"major_revision","confidential_remarks":"The paper fits the journal's scope reasonably: it is video-based inference with a strong physics component, but the core contribution is computational elastography rather than a purely computer-vision method. The self-citation to Visual Vibration Tomography is appropriate and not used to prop up the method. The main concern is that the real-data stiffness claim depends on comparing moduli at different frequency ranges; if overlapping-frequency rheometry is not available, the authors should soften the claim and clearly label the VSWE result as an effective high-frequency modulus."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The genuinely new thing here is the specific pipeline: dense phase-based motion extraction from video, FFT-based dispersion estimation, and a physics-based FEM forward model to invert for layer thickness and stiffness via SSIM. Seismology and visual vibrometry provided the pieces, but combining them for soft tissue characterization from ordinary video is a real step. The strongest evidence is the real gelatin thickness recovery: estimates sit inside the caliper confidence interval across three thicknesses and a temperature ramp, and the sensitivity study shows the method can detect 5-10% parameter changes. The paper is also notably candid. It explicitly states that rheometry was only measured at 10-100 Hz while the excitation was 40-200 Hz, acknowledges possible viscoelastic frequency dependence, and includes ablations of objective function, mesh resolution, and observation window. That level of self-reporting deserves credit.\n\nThe soft spots are real but not fatal. The stiffness validation is the biggest one: the phrase “within 1.2% error of the rheometry range” appears in Fig. 7's caption right after the caveat about the frequency mismatch. For a viscoelastic hydrogel, a modulus at 40-200 Hz does not have to match the rheometry value at 10-100 Hz, so the claimed accuracy is suggestive but not established as ground truth. The paper would be stronger if it reported frequency-dependent rheometry or at least discussed the expected magnitude of dispersion. There are also no error bars on repeated video measurements, the leg validation is simulation-only, and the code release is promised but not yet available. These are all correctable in revision.\n\nThe central claim that camera-observed surface waves can recover layer thickness holds up well; the stiffness claim needs tempering. There is no circularity in the fitting itself—the forward model is independent of the observed data. The at-home monitoring language in the abstract overreaches given the high-speed camera and shaker setup, but that does not undermine the method's contribution to video-based material characterization.\n\nI would send this to serious reviewers. It deserves a fair shot, with requests for code, error bars, and a clearer definition of the 1.2% metric. I'd also bring it to a reading group, since it bridges vision and solid mechanics in a way that often sparks useful discussion.","headline":"Solid physics-based pipeline for video surface-wave elastography; thickness inference is convincing, stiffness validation is honest but weaker than the abstract implies.","tokens_in":13842,"tokens_out":1837,"would_cite":true,"duration_ms":24564,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Visual Surface Wave Elastography shows that a video of surface waves carries enough information to recover the thickness and stiffness of a soft tissue layer over bone.","keywords":["surface wave elastography","dispersion relation","video-based material characterization","tissue stiffness estimation","tissue thickness estimation","phase-based motion processing","Bloch-Floquet boundary conditions","structural similarity index"],"falsifier":"Build two gelatin phantoms with the same measured thickness and stiffness but different densities, for example by adding glycerol to one, run VSWE with the paper's fixed $\\rho = 1\\,\\mathrm{g\\,cm^{-3}}$ assumption, and check whether the recovered thickness and stiffness stay the same; if they shift with density, the inversion's prior on density is doing real work and real-tissue density variations will bias results.","tokens_in":1607,"feed_emoji":"🎥","tokens_out":4974,"duration_ms":136291,"temperature":0.7,"pith_summary":"The paper proposes that a plain video camera can measure what lies beneath a surface: specifically, the thickness and stiffness of a soft tissue layer such as skin, fat, and muscle over bone. It claims this is possible because surface waves are dispersive, meaning different wavelengths travel at different speeds, and the entire frequency-wavenumber relationship, called the dispersion relation, is determined by the layer's thickness and stiffness. The method extracts this dispersion relation from video with sub-pixel motion processing and a 2D FFT, then solves an inverse problem by generating physics-based dispersion relations for candidate thickness-stiffness pairs and choosing the one that best matches observation. Validation on simulated and real gelatin phantoms recovers thicknesses within caliper measurements and stiffness within 1.2% of rheometry values, and a simulated 3D human leg shows recovery of spatially varying thickness. If true, this is a step toward at-home health monitoring and camera-based material characterization without specialized ultrasound or MRI equipment.","feed_headline":"Video of surface waves estimates tissue thickness and stiffness","feed_subtitle":"Camera-only motion analysis matches observed wave patterns to physics to recover layer depth and stiffness.","key_machinery":"The central object is the dispersion relation, the set of frequency-wavenumber pairs $(\\omega,\\gamma)$ describing all wave modes a medium supports; for the assumed two-layer model the paper shows this set is uniquely fixed by thickness $T$ and stiffness $E$. The relation is computed by solving the harmonic elastic wave equation with Bloch-Floquet, or phase-shifted, periodic boundary conditions, and the observed version is obtained from video by a 2D FFT of phase-based motion estimates. The inversion is carried out by maximizing the structural similarity index, or SSIM, between the observed dispersion image and the image of hypothesized dispersion curves, solved by grid search. Dimensionless characteristic numbers such as $\\pi_1 = \\gamma L$ and $\\pi_6 = \\gamma T$ are introduced in the discussion to describe when the observation window, sampling rates, and finite-element mesh are adequate.","core_discovery":"The paper establishes that wave motion visible at the surface of a soft-over-hard layered medium carries enough information to recover the unknown thickness $T$ and elastic modulus $E$ of the soft layer from video alone. Under standard biomechanical assumptions, a uniform isotropic linear-elastic layer of known density and Poisson's ratio resting on a rigid foundation, with waves traveling horizontally, the dispersion relation $D(T,E)$ is fully determined by $T$ and $E$. The paper shows that a dispersion relation extracted from video via phase-based motion processing and a 2D FFT can be matched against numerically computed dispersion relations, and that maximizing the structural similarity index between observed and hypothesized dispersion images yields accurate parameter estimates. The evidence includes sensitivity to 5% parameter changes in plane-strain simulations, agreement with caliper and rheometry ground truth on real gelatin phantoms, and agreement with known thickness distributions in a full 3D simulation of a human calf.","pith_inferences":["The fixed-density assumption means the method is effectively estimating stiffness relative to an assumed density of $\\rho = 1\\,\\mathrm{g\\,cm^{-3}}$; a phantom test that holds thickness and stiffness fixed while varying density would reveal how much accuracy depends on that prior.","The same dispersion-matching pipeline could be adapted to multi-layer, viscoelastic, or anisotropic tissue models by replacing the forward solver, but the paper's evidence supports only the single uniform layer case.","The characteristic-number analysis implies a simple smartphone-feasibility check before building hardware: compare the camera's frames-per-second and pixels-per-meter against the required dimensionless groups of the target tissue, since matching those groups, not raw resolution alone, is what preserves performance.","Because the method needs only surface wave video and no contact sensors, a fixed camera could in principle track muscle stiffness changes during gestures, a direct extension toward human-computer interaction."],"forward_implications":["A camera alone, without ultrasound, MRI, or contact sensors, can recover coarse estimates of tissue thickness and stiffness in a soft-over-hard layered medium.","In plane-strain simulations, the method distinguishes parameter changes as small as 5%, so it can track small variations that may be clinically relevant.","On real gelatin phantoms, estimated thicknesses fall within the caliper confidence interval and stiffness within 1.2% of the rheometry range across a range of temperatures.","On a simulated 3D human leg, sweeping observation windows recover a spatially varying thickness distribution while also recovering the constant stiffness throughout the calf.","The characteristic numbers give a scaling recipe: preserving dimensionless groups such as spatial and temporal sampling relative to wavelength and frequency maintains performance even when the object or video parameters are very different."],"supporting_citations":[{"why":"Supplies the phase-based motion processing used to extract sub-pixel surface displacements from the video.","marker":"[43]"},{"why":"Defines the phase-shifted periodic boundary conditions with which the theoretical dispersion relation is computed.","marker":"[9, 20]"},{"why":"Defines the structural similarity metric used as the optimization objective between observed and hypothesized dispersion images.","marker":"[45]"},{"why":"Used to generate the ground-truth plane-strain and 3D leg wave simulations on which the method is validated.","marker":"[1]"},{"why":"Provides the 3D lower-extremity geometry that makes the human leg simulation anatomically realistic.","marker":"[3]"},{"why":"Supplies the assumed Poisson's ratio of soft tissue used in the physics model.","marker":"[27]"},{"why":"Supplies the assumed tissue density of 1 g per cubic centimeter used in the physics model.","marker":"[36]"},{"why":"Documents the temperature and frequency dependence of hydrogel modulus, used to interpret the stiffness differences between rheometry and VSWE.","marker":"[48]"}],"fun_headline_variants":["Video surface waves unveil layer thickness and stiffness","Camera-only wave video recovers tissue depth and elasticity","Surface wave dispersion from video gauges stiffness and depth","Wave video plus physics yields subsurface stiffness and thickness"],"cache_read_input_tokens":16000,"weakest_assumption_plain":"The inversion assumes the medium is one uniform elastic layer of known density and Poisson's ratio on a rigid base, with waves traveling in a single horizontal direction; real anatomy is curved, graded or anisotropic, and nonuniform in thickness, so any deviation biases the recovered $(T,E)$.","fun_headline_variants_meta":{"raw":{"variants":["Video surface waves unveil layer thickness and stiffness","Camera-only wave video recovers tissue depth and elasticity","Surface wave dispersion from video gauges stiffness and depth","Wave video plus physics yields subsurface stiffness and thickness"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000232,"raw_usage":{"total_tokens":1429,"prompt_tokens":825,"completion_tokens":604,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":441,"completion_tokens_details":{"reasoning_tokens":543}},"tokens_in":441,"tokens_out":604,"duration_ms":7166,"temperature":1.0,"reasoning_tokens":543,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T18:01:26.754465+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build two gelatin phantoms with the same measured thickness and stiffness but different densities, for example by adding glycerol to one, run VSWE with the paper's fixed $\\rho = 1\\,\\mathrm{g\\,cm^{-3}}$ assumption, and check whether the recovered thickness and stiffness stay the same; if they shift with density, the inversion's prior on density is doing real work and real-tissue density variations will bias results.","supporting_citations":[{"cited_title":"Phase-based video motion processing","cited_arxiv_id":null,"evidence_quote":"Supplies the phase-based motion processing used to extract sub-pixel surface displacements from the video."},{"cited_title":"Bovik, H.R","cited_arxiv_id":null,"evidence_quote":"Defines the structural similarity metric used as the optimization objective between observed and hypothesized dispersion images."},{"cited_title":"Comsol multiphysics® v5.5, 2019","cited_arxiv_id":null,"evidence_quote":"Used to generate the ground-truth plane-strain and 3D leg wave simulations on which the method is validated."},{"cited_title":"Three dimensional lower extremity musculoskeletal geometry of the visible human female and male","cited_arxiv_id":null,"evidence_quote":"Provides the 3D lower-extremity geometry that makes the human leg simulation anatomically realistic."},{"cited_title":"Quantifying cell- generated forces: Poisson’s ratio matters","cited_arxiv_id":null,"evidence_quote":"Supplies the assumed Poisson's ratio of soft tissue used in the physics model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the assumed tissue density of 1 g per cubic centimeter used in the physics model."},{"cited_title":"Measurement of the elasticity modulus of soft tissues","cited_arxiv_id":null,"evidence_quote":"Documents the temperature and frequency dependence of hydrogel modulus, used to interpret the stiffness differences between rheometry and VSWE."}],"review_version":1}