{"id":"7f0cba22-db91-49f9-afd6-aeb6fb48563c","arxiv_id":"2607.03079","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"Using idealised cuboidal geometries instead of MRI-reconstructed real sample shapes underestimates brain-tissue shear modulus by about 10% in shear and 48% under axial loading, mainly from compression contact errors.","lead":"Idealised cuboid models of brain tissue samples systematically under-estimate shear modulus by ~10% in shear and ~48% in axial loading versus MRI-based real geometries. This shows that contact and strain errors from geometric simplification bias inverse characterisation of very soft tissues.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The reader correctly isolates the softest residual (locked exponents, possible BC/mesh confounds) yet correctly judges it ordinary rather than load-bearing: the design isolates geometry, the compression-only first-order check already reproduces the large discrepancy, and the contact-area and strain-map evidence (Sections 3.1–3.2, Fig. 7) directly support the mechanistic attribution. No stronger internal inconsistency or uncontrolled confound appears in the full text. Consequently the CONDITIONAL verdict (sound result, residual absolute-number doubt removable by data release and a brief sensitivity check) stands; no adjustment is warranted.","tokens_in":12365,"tokens_out":477,"duration_ms":26113,"concrete_test":"Re-optimise µ1, µ2 (and, optionally, free α1, α2) independently for each of the five TC real/idealised pairs using the identical force–displacement data and SLSQP settings; if the mean (µ_ideal − µ_real)/µ_real remains within ~5 % of the reported −47.5 %, the locked-exponent choice and any minor BC/mesh differences are not material to the bias claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on a cleanly controlled comparison (identical experimental force–displacement data, same second-order Ogden form, same SLSQP objective and bounds; only the geometry supplied to the FE mesh differs). Idealised cuboids systematically overestimate top-surface contact area (15–30 % larger than MRI-derived surfaces) and therefore under-predict µ to match the measured forces; the effect is largest under compression where progressive, incomplete contact and non-symmetric strain fields appear (Fig. 7, Tables 1–2). The first-order Ogden compression-only check already recovers essentially the same ~48 % discrepancy, so the locked exponents (α1 = −8, α2 = 16) are not driving the result. Residual uncertainties in mesh quality on irregular surfaces or video-based node selection for tension/shear BCs remain, but none is large enough to reverse the sign or nullify the magnitude of the reported bias.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript quantifies how idealising nominally cuboidal brain-tissue samples as perfect cuboids biases the shear modulus recovered by inverse finite-element analysis. MRI-derived “real” geometries and dimension-matched idealised cuboids are each fitted, independently, to the same experimental force–displacement records from shear, tension and compression tests on ovine brain tissue using a second-order Ogden model (exponents fixed after a preliminary four-parameter fit). Idealised geometries systematically under-estimate the shear modulus—by ~10 % in shear and ~48 % under axial loading, the latter driven almost entirely by compression—because they over-estimate contact area and cannot reproduce the progressive, incomplete contact and non-symmetric strain fields observed experimentally. The authors therefore recommend that measured sample geometry be used in inverse characterisation of very soft tissues.","tokens_in":12583,"tokens_out":1099,"duration_ms":24203,"significance":"If the reported bias is robust, the result is of immediate practical importance: the large majority of published soft-tissue constitutive parameters have been obtained with idealised geometries and may therefore contain a systematic under-estimate of stiffness, especially under compression. The experimental design is clean (identical force data, identical constitutive form and optimiser; only geometry differs), the compression-only first-order Ogden check isolates the dominant source of the discrepancy, and the contact/strain-field explanation is directly supported by the MRI segmentations and FE strain maps. The work supplies a concrete, falsifiable quantitative bound rather than a qualitative caution, and therefore has clear value for experimental biomechanics protocols.","major_comments":[{"comment":"§2.4.2–2.4.3 and Fig. 7: mesh-independence and contact fidelity for the irregular real geometries are asserted but not demonstrated. An element size of 0.5 mm is stated to be “consistent with MRI resolution and ensures mesh-independent converged solution,” yet no refinement study (or even a single coarser/finer comparison) is reported for either geometry class. Because the central claim rests on differences in progressive contact and localised strain, a short mesh-convergence check on at least one real-geometry compression case is needed to confirm that the ~48 % modulus gap is not an artefact of under-resolved contact on the irregular surface.","section":null},{"comment":"§2.4.3 and §3.2: boundary-condition implementation is not identical across loading modes. Tension and shear apply prescribed displacements to video-selected top-surface nodes, while compression uses rigid-plate contact. The authors correctly note that nodal prescription would be inappropriate for real geometries at large compression, but the mixed BC strategy leaves open a residual confound when comparing the shear (~10 %) and axial (~48 %) discrepancies. A brief sensitivity test—e.g., applying the same rigid-plate contact formulation to one tension or shear specimen—would strengthen the claim that geometry, rather than BC formulation, is the dominant driver.","section":null}],"minor_comments":[{"comment":"§2.2: two different MRI scanners and resolutions (0.25×0.25×0.5 mm vs 0.6 mm isotropic) were used for axial versus shear samples. A short statement on whether the coarser shear resolution systematically under- or over-estimates top-surface area would help the reader judge comparability of the two cohorts.","section":null},{"comment":"Table 1 and Table 2: the difference formula is written as (μIdeal − μReal)/μReal × 100, which correctly yields negative values; the abstract and text, however, speak of “lower” moduli without always quoting the signed percentage. Adding the signed mean ± SD in the abstract would remove any ambiguity.","section":null},{"comment":"Fig. 5 and Fig. 6: force–displacement curves are shown with every fiftieth experimental point; residual plots or an R²/RMSE table would make the quality of the fits more transparent, especially given that the objective is absolute area between curves.","section":null},{"comment":"§2.4.4, Eq. (1): the second-order Ogden form is standard, but the decision to freeze α1 = −8, α2 = 16 after a four-parameter pilot is only briefly justified. A one-sentence note that re-optimising the exponents on a subset of samples left the modulus gap essentially unchanged would close the residual concern about model-form sensitivity.","section":null},{"comment":"References [18,19] are listed as “in press” / “2025–2026”; if they are the authors’ own related work, a brief parenthetical clarification would help the reader place the present contribution.","section":null}],"recommendation":"minor_revision","confidential_remarks":"The paper is a solid, carefully controlled methods contribution that belongs in a biomechanics or computational-mechanics journal. The two major points are addressable with modest additional analysis already within the authors’ existing FE framework; I do not see a need for new experiments. Fit to cs.CE is reasonable given the inverse-FE focus, though a pure experimental-mechanics venue might also be appropriate."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This paper does something useful and overdue: it puts a number on the error introduced by the almost universal practice of treating soft-tissue samples as perfect cuboids or cylinders in inverse FE characterisation. Same force–displacement data, same Ogden form, same optimiser; only the mesh geometry changes. Idealised shapes give moduli ~10% lower in shear and ~48% lower under axial loading, almost entirely from compression. The reason is transparent: real top surfaces are imperfect, contact develops progressively and non-uniformly, and the idealised models over-estimate contact area from the first instant, so the optimiser has to soften the material to match the measured forces. Fig. 7 and the first-order compression-only check (Table 2) make that mechanism hard to dismiss.\n\nWhat is new is the controlled magnitude across three loading modes with MRI-reconstructed meshes. Prior papers noted that samples are irregular; almost none quantified the parameter bias that follows. The experimental design is clean, the force fits are good for both geometries, and the citation pattern is appropriate (they know the literature on both brain testing and geometric imperfections in compression). Small n (five per mode) and locked exponents after a preliminary fit are ordinary limitations for this field, not load-bearing flaws; the stress-test note is right that residual mesh or BC details are unlikely to reverse the sign or erase the size of the effect.\n\nSoft spots worth noting: no public meshes or force data, so absolute numbers cannot be re-run; the idealised cuboids are still sample-specific averages rather than pure nominal dimensions, which slightly under-states the real-world error; and the recommendation to always use measured geometry is stronger than the data strictly require for shear. None of that undercuts the central result.\n\nThis is for anyone who publishes or uses inverse-FE constitutive parameters for brain or other very soft tissues. It deserves a serious referee. I would cite the bias numbers and bring it to reading group.","headline":"Clean controlled quantification of a real methodological bias: idealised cuboids systematically under-estimate brain shear modulus (~10% shear, ~48% axial, driven by compression contact).","tokens_in":13183,"tokens_out":496,"would_cite":true,"duration_ms":4888,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Idealised sample shapes systematically understate brain-tissue stiffness, by about 10% in shear and nearly 50% in compression.","keywords":["very soft tissue","brain tissue","sample geometry","inverse finite-element analysis","shear modulus","Ogden model","contact mechanics"],"falsifier":"Repeat the identical MRI-to-idealised comparison on a new set of specimens while forcing full platen contact (e.g., by pre-compressing until optical contact is complete) and check whether the ~50% compressive modulus gap disappears.","tokens_in":13248,"feed_emoji":"🧠","tokens_out":615,"duration_ms":5342,"temperature":0.7,"pith_summary":"Most mechanical tests of very soft tissues such as brain still treat each specimen as a perfect cube or cylinder when converting measured forces into material parameters. Real samples, however, are imperfect: their surfaces are uneven, edges are rounded, and contact with the loading platen develops only gradually. This paper shows, by imaging every ovine-brain specimen with MRI and then running identical finite-element inverse analyses on both the true geometry and its idealised cuboidal counterpart, that the idealisation systematically underestimates the shear modulus. The bias is modest in simple shear (~10%) but reaches nearly 50% under compression, because the perfect-cube models assume full, uniform contact from the first instant of loading and therefore over-predict force for any given stiffness. The authors conclude that constitutive parameters obtained from idealised geometries carry a systematic downward bias and that the actual measured geometry should be used whenever material properties of very soft tissues are identified by inverse analysis.","feed_headline":"Ideal cubes understate soft-tissue stiffness by up to 50%","feed_subtitle":"MRI-true geometries of brain samples show idealised shapes systematically bias inverse analysis, especially in compression.","key_machinery":"Controlled real-versus-idealised finite-element comparison: each specimen is meshed twice—once from its MRI segmentation and once as a volume-matched cuboid—then subjected to identical inverse optimisation of a second-order Ogden model against the same experimental force-displacement curves.","core_discovery":"When the same force-displacement data from brain-tissue specimens are inverted with MRI-reconstructed geometries versus idealised cuboids, the idealised models return shear moduli that are on average 10% lower in shear and 48% lower under axial loading (driven almost entirely by compression). The discrepancy arises because idealised models cannot capture progressive, partial contact or the resulting non-uniform strain fields.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Ideal cuboids understate soft-tissue moduli by up to 48%","MRI-true sample shapes raise brain shear modulus estimates","Idealised geometries bias soft-tissue inverse analysis lower","Cuboid models miss compressive contact, cut moduli ~48%","Real geometries show ideal cubes understate stiffness markedly"],"cache_read_input_tokens":128,"weakest_assumption_plain":"The modulus gap is produced mainly by geometric contact and strain differences, not by residual mismatches in mesh density, boundary-condition details, or the fixed Ogden exponents chosen after a preliminary fit.","fun_headline_variants_meta":{"raw":{"variants":["Ideal cuboids understate soft-tissue moduli by up to 48%","MRI-true sample shapes raise brain shear modulus estimates","Idealised geometries bias soft-tissue inverse analysis lower","Cuboid models miss compressive contact, cut moduli ~48%","Real geometries show ideal cubes understate stiffness markedly"]},"model":"grok-4.5","effort":"low","cost_usd":0.004398,"raw_usage":{"total_tokens":1347,"prompt_tokens":825,"num_sources_used":0,"completion_tokens":82,"cost_in_usd_ticks":43980000,"prompt_tokens_details":{"text_tokens":825,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":440,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":825,"tokens_out":82,"duration_ms":3864,"temperature":1.0,"reasoning_tokens":440,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-12T05:02:26.732871+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Repeat the identical MRI-to-idealised comparison on a new set of specimens while forcing full platen contact (e.g., by pre-compressing until optical contact is complete) and check whether the ~50% compressive modulus gap disappears.","supporting_citations":[],"review_version":1}