{"id":"4f700485-45f0-4682-9bd7-05cb849d3ffe","arxiv_id":"2607.28414","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"A Kelvin–Voigt shear-only FDTD solver with solenoidal force projection recovers shear speed and attenuation to ~1% and integrates with acoustic simulators for ARF-driven tissue motion.","lead":"A reduced finite-difference solver models only the shear motion that ultrasound radiation force creates in soft tissue, skipping full elasticity. It matches analytical and phantom benchmarks to about 1% while running far cheaper than general finite-element codes, aiding elastography and neuromodulation studies.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified to the central homogeneous validation claims; reader's scope caveat is real but disclosed and non-fatal.","rationale":"Reader correctly scores this as a solid physics.med-ph methods paper: reductions are classical, numerics (2nd-order FDTD + matrix-free CG projection + separable pre-compute) are appropriate, and external anchors (analytic KV dispersion, QIBA phantom convention) support the headline errors with low circularity. The load-bearing premise for going beyond homogeneous SWE-style use is indeed dropping all compressional physics while still showing skull pipelines; that is disclosed, and the quantitative claims do not rest on it. I do not elevate that scope limit into a verdict downgrade: ACCEPT remains appropriate for what was actually measured. The variable-μ writing gap is worth tightening in revision but is secondary to the strongest_claim. Confidence tracks the reader—evaluation is of the written math and reported tables, not an independent run.","tokens_in":17360,"tokens_out":618,"duration_ms":71878,"concrete_test":"Clone the claimed GitHub repo and reproduce the 16-point 1D KV sweep (Table 2) plus the four QIBA G cases (Fig. 4c); then add one FEM comparison (e.g. LS-DYNA/Abaqus) with a soft inclusion (G_inc/G_bg=2, same Gaussian ARF) using a proper ∇·(2με)−∇p discretization. If homogeneous tables match and inclusion arrival-time/cs errors stay ≲ few percent, the central claim and heterogeneity language both hold; if inclusion errors blow up, restrict the heterogeneity claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The strongest claim is a set of homogeneous-media accuracy numbers (<0.1% cs, ~3%/<1% KV α/cφ, ~1% QIBA G-sweep) for a classical incompressible Kelvin–Voigt shear reduction with solenoidal ARF projection. Those benchmarks sit inside the regime where ∇·u=0 and f→f_S are standard and well motivated (K/μ~10^5–10^6; curl or projected sources). The reader's weakest assumption—omitting compressional content, P–S conversion, and volumetric strain—is a genuine scope limit for heterogeneous/transcranial use (§§2.4, 5, 6.1), but it is explicitly framed as a trade-off, the skull demos brain-mask the force and use homogeneous brain μ, and none of the headline error bars depend on interface elastodynamics being negligible. A secondary formulation gap (Eq. 16 and the update an=(μ/ρ)∇²u are constant-coefficient forms, while the text claims spatially varying μ,ρ,η without a variable-coefficient identity or inclusion benchmark) affects the heterogeneity bullet, not the quoted recovery figures. No internal inconsistency in the validated core was found.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript derives a reduced Kelvin–Voigt shear-wave equation for ultrasound-driven soft-tissue motion by applying near-incompressibility, small-strain linearization, Helmholtz decomposition, and solenoidal projection of the body force to the linear Navier equations, then implements it as an explicit second-order FDTD scheme with a matrix-free CG Poisson solve (and a pre-computed projection for separable ARF sources). Homogeneous validation recovers theoretical shear speed to <0.1% (linear-fit), shows trace-L2 self-convergence with p≳2, matches analytical KV attenuation and phase speed to ~3% and <1% over η∈[0,1.5] Pa·s, and recovers RSNA QIBA phantom speeds to ~1% for G=1–10 kPa. Transcranial demos through micro-CT skulls, driven by Fullwave-derived ARF, produce 1.7–5 μm displacements. Code is released.","tokens_in":17637,"tokens_out":1304,"duration_ms":44772,"significance":"If the homogeneous accuracy claims hold—as the four independent benchmarks indicate—the paper supplies a purpose-built, lightweight shear-only FDTD tool that is substantially cheaper than general-purpose FEM for SWE and elastomechanical neuromodulation workflows, while integrating cleanly with acoustic simulators. Strengths include reproducible validation scripts, an external QIBA cross-check previously exercised with LS-DYNA/Abaqus, a linearity-based pre-computed projection that is not circular, and open code. The reduction is classical but carefully specialized and documented for the ultrasound community; that is a genuine practical contribution even if the continuum reduction itself is not new.","major_comments":[{"comment":"§2.4 Eq. (16) and §3.1 Eqs. (17)–(18) are written in constant-coefficient form, ρ u_tt = μ ∇²u + η ∇² u_t + f_S (and a_n = (μ/ρ)∇²_h u^n + …). The abstract and §1/§7 claim the framework “accommodates spatial heterogeneity in shear modulus, density, and viscosity,” but the correct variable-coefficient reduction of ∇·σ for isotropic Kelvin–Voigt media replaces μ∇²u by terms of the form ∇·(μ ∇u) (plus analogous viscous terms), and ρ(x) cannot simply sit outside as a global factor. No variable-coefficient identity, discrete stencil, or heterogeneous inclusion benchmark is given. Either derive and implement the variable-μ,ρ,η update and add at least one heterogeneous test, or qualify the heterogeneity claim to piecewise-homogeneous / constant-coefficient media. This does not undermine the homogeneous recovery figures, but it is load-bearing for the stated scope.","section":"§2.4 Eq. (16); §3.1"},{"comment":"The abstract and §3.1 advertise “second-order spatial and temporal accuracy,” yet §3.1 correctly notes that the viscous term is only first-order in time (backward difference for u̇). For the η-sweep that is a headline result (§4.3, Table 2), the formal order is therefore mixed. Please state the temporal order of the full scheme consistently in the abstract and methods, and, if practical, report a brief viscous timestep-refinement check so that the ~3% attenuation error is not partly first-order truncation.","section":"Abstract; §3.1; §4.3"}],"minor_comments":[{"comment":"§4.1 reports both a mean peak-picked speed (1.976 m/s, ~1.2% low) and a distance–time linear-fit speed (1.998 m/s, <0.1%). The abstract quotes only the latter. A brief clause noting that the headline <0.1% is the linear-fit (not peak-pick) estimate would avoid over-reading.","section":"§4.1; Abstract"},{"comment":"Fixed Dirichlet outer boundaries (§3.5, §6.1) are acknowledged; validation is restricted to early/intermediate times. For the QIBA runs (lateral offsets 4–14 mm on a 60×50×50 mm domain) a one-sentence confirmation that reported arrivals are free of wall reflections would help readers reproduce the protocol.","section":"§3.5; §4.4"},{"comment":"Transcranial sections (§5) brain-mask the force and use homogeneous brain μ; they illustrate the multiphysics pipeline rather than validate interface elastodynamics. Labeling them explicitly as “demonstrations” (as the heading does) in the abstract’s final sentence would align claim strength with content.","section":"Abstract; §5"},{"comment":"Minor notation: G and μ are both used for shear modulus (QIBA vs. derivation); a single symbol or an explicit G≡μ note would help. Also “Lam´ e” encoding glitches appear in §2.2.","section":"§2.2; §4.4"},{"comment":"Table 1 global fit p=2.9 is steeper than formal order; the text already notes self-convergence bias near the reference—consider also quoting the coarser pairwise slopes (≈2.2–2.9) in the abstract’s “p≳2” phrase for transparency.","section":"§4.2; Table 1"}],"recommendation":"minor_revision","confidential_remarks":"The homogeneous core is solid and the QIBA cross-check is the right external anchor; I would not send this back for heavy re-work. The variable-coefficient gap is the only scope overclaim that needs a clean fix or walk-back before acceptance. Fit for a methods-oriented medical-physics / ultrasonics journal is good. Code release is a plus and should be preserved through revision."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"Punchline: this is a practical reduced Kelvin–Voigt shear FDTD with matrix-free CG solenoidal projection of ARF body forces and a cheap pre-compute for separable sources. The homogeneous numbers are clean and externally anchored. It will not rewrite elastodynamics, but it is a usable multiphysics piece for SWE and neuromodulation pipelines.\n\nWhat is actually new is the packaging, not the continuum reduction. Incompressible/shear-only FDTD and Green’s ARF solvers already exist; Orescanin et al. did staggered velocity–stress KV. Pinton’s distinctives are a displacement leapfrog update, explicit CG Helmholtz projection so arbitrary radiation-force sources do not need a needle geometry, and the linearity trick that turns a fixed spatial ARF pattern into one Poisson solve plus scalar multiplies. That is real engineering value when you already run Fullwave-style acoustics and want full-field shear cheaply.\n\nThe validation is the strong part. Non-viscous linear-fit cs within <0.1%, trace-L2 self-convergence with p≳2 on a fixed physical source, a 16-point 1D harmonic KV sweep (~3% attenuation, <1% phase speed), and end-to-end QIBA elastic phantoms ~1% over G=1–10 kPa. Those checks are independent of the solver’s own fits and match closed-form KV plus the RSNA phantom spec previously run in LS-DYNA/Abaqus. Fixed walls and early-time analysis are handled honestly. Code is claimed on GitHub.\n\nSoft spots, in proportion. Dropping all compressional content (∇·u=0, f→f_S) is load-bearing for heterogeneous and skull–brain claims; near-source pressure, P–S conversion, and volumetric-strain biomarkers are out of scope. The paper says so in §§2.4 and 6.1, brain-masks the force, and uses homogeneous brain μ in the demos, so the headline error bars do not secretly depend on interface elastodynamics. Still, do not over-read the transcranial figures as full acousto-elastic fidelity. Second, the text advertises spatially varying μ,ρ,η while Eq. 16 and the update are written in constant-coefficient form; there is no variable-coefficient identity or inclusion benchmark. That weakens the heterogeneity bullet, not the homogeneous recovery figures. Kelvin–Voigt is a single-parameter damper; fine for the stated use, thin for broadband rheology. PML is future work.\n\nWho it is for: people who couple acoustic simulators to shear motion and want something lighter than commercial FEM and more full-field than Green’s functions. Math and citation pattern look solid; self-cites are mostly prior Fullwave/skull work, not circular validation.\n\nI would send it to peer review. Engage if you care about ARF–shear workflows; skim if you only want new continuum theory.","headline":"Solid, well-validated shear-only FDTD tool for ARF workflows; novelty is engineering assembly more than new physics, and the main soft spots are disclosed scope limits plus a thin heterogeneity claim.","tokens_in":18288,"tokens_out":707,"would_cite":true,"duration_ms":17214,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"A reduced Kelvin–Voigt shear equation, solved with explicit FDTD and solenoidal force projection, recovers ultrasound-driven tissue motion accurately at a fraction of full elastodynamic cost.","keywords":["shear wave elastography","FDTD","acoustic radiation force","viscoelasticity","Kelvin–Voigt","solenoidal projection","ultrasonic neuromodulation","transcranial ultrasound"],"falsifier":"Compare the reduced FDTD displacement field against a full viscoelastic FEM solution on the same heterogeneous skull–brain geometry with the same radiation-force source; large near-interface or near-focus discrepancies in shear displacement or arrival time would falsify adequacy of the solenoidal reduction.","tokens_in":18177,"feed_emoji":"🌊","tokens_out":985,"duration_ms":22307,"temperature":0.7,"pith_summary":"Ultrasound pushes on soft tissue create tiny shear waves used in elastography and studied as possible drivers of neuromodulation. Full elastic finite-element codes can model that motion but spend effort on compressional physics that barely matters when tissue is nearly incompressible. This paper strips the Navier equations down to a shear-only Kelvin–Voigt wave equation by enforcing incompressibility and projecting the radiation force onto its divergence-free part, then advances that equation with a second-order finite-difference time-domain scheme. A one-time Poisson solve precomputes the force projection when the acoustic push pattern is fixed, cutting that cost by one to two orders of magnitude. Homogeneous benchmarks recover shear speed to under 0.1 percent, match Kelvin–Voigt attenuation and phase speed across a viscosity sweep, and hit standardized QIBA phantom speeds to about 1 percent; skull-coupled demos produce micrometer-scale brain displacements consistent with clinical ARFI. The point is a purpose-built, acoustic-simulator-friendly tool for the shear motion that actually matters in these applications.","feed_headline":"Shear-only FDTD matches ultrasound tissue motion to ~1%","feed_subtitle":"A reduced Kelvin–Voigt solver hits QIBA phantom speeds without full elastodynamic FEM cost","key_machinery":"The reduced Kelvin–Voigt shear equation ρ ∂²u/∂t² = μ ∇²u + η ∇² ∂u/∂t + f_S, with f_S the solenoidal projection of the radiation body force obtained from a matrix-free conjugate-gradient Poisson solve (precomputed once when the force is spatially separable).","core_discovery":"Under near-incompressibility, small strain, Helmholtz decomposition, and solenoidal projection of the body force, the full linear viscoelastic Navier system reduces to a forced Kelvin–Voigt shear wave equation whose explicit second-order FDTD solution recovers theoretical shear wavespeed to under 0.1 percent, analytical attenuation and phase speed to roughly 3 percent and under 1 percent over η from 0 to 1.5 Pa·s, and RSNA QIBA phantom shear speeds to about 1 percent across G = 1–10 kPa, while remaining far cheaper than general-purpose FEM for shear-dominant ultrasound problems.","pith_inferences":["Fixed-wall boundaries will limit long-time or large-domain use until absorbing layers are added; early-time sensor windows are a temporary workaround, not a full fix.","The same reduction could stress-test whether flexoelectric or strain-gradient neuromodulation mechanisms remain plausible once skull-filtered ARF amplitudes are used consistently.","If P–S conversion at bone–soft-tissue interfaces proves non-negligible in vivo, hybrid near-field full-elastic / far-field shear coupling may be the natural next method rather than pure shear everywhere."],"forward_implications":["SWE and ARFI pipelines can couple acoustic simulators to this shear solver instead of full elastodynamic FEM for routine forward modeling.","Precomputed force projection makes full-field heterogeneous shear runs cheap enough for iterative inverse problems and machine-learning training sets.","Transcranial neuromodulation hypotheses can be checked quantitatively against predicted micrometer-scale brain displacements from realistic skull-filtered pushes.","Spatial maps of shear modulus, density, and viscosity can be inserted directly without changing the update structure."],"fun_headline_variants":["Reduced Kelvin–Voigt FDTD recovers QIBA shear speeds to ~1%","Shear-only FDTD hits theoretical wavespeed within 0.1%","Solenoidal projection cuts viscoelastic shear solve cost sharply","Explicit FDTD matches Kelvin–Voigt attenuation to ~3% over η sweep","Near-incompressible shear solver rivals FEM for ultrasound ARFI"],"cache_read_input_tokens":128,"weakest_assumption_plain":"That radiation-force tissue motion can be treated as pure shear—dropping compressional waves, mode conversion at interfaces, and any need for volumetric strain—without spoiling the displacements that matter.","fun_headline_variants_meta":{"raw":{"variants":["Reduced Kelvin–Voigt FDTD recovers QIBA shear speeds to ~1%","Shear-only FDTD hits theoretical wavespeed within 0.1%","Solenoidal projection cuts viscoelastic shear solve cost sharply","Explicit FDTD matches Kelvin–Voigt attenuation to ~3% over η sweep","Near-incompressible shear solver rivals FEM for ultrasound ARFI"]},"model":"grok-4.5","effort":"low","cost_usd":0.004982,"raw_usage":{"total_tokens":1510,"prompt_tokens":965,"num_sources_used":0,"completion_tokens":97,"cost_in_usd_ticks":49824000,"prompt_tokens_details":{"text_tokens":965,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":448,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":965,"tokens_out":97,"duration_ms":8289,"temperature":1.0,"reasoning_tokens":448,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T08:06:30.813861+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Compare the reduced FDTD displacement field against a full viscoelastic FEM solution on the same heterogeneous skull–brain geometry with the same radiation-force source; large near-interface or near-focus discrepancies in shear displacement or arrival time would falsify adequacy of the solenoidal reduction.","supporting_citations":[],"review_version":1}