{"id":"53771b6f-9887-4c05-b258-feb1f6339d8b","arxiv_id":"2505.17647","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A 10-patient 4D-CTA image and geometry dataset with FE-based synthetic ground truth is released to support benchmarking of AAA wall kinematics analysis.","lead":"This paper releases a new public dataset of time-resolved 3D CT images of abdominal aortic aneurysms from ten patients, together with extracted geometries and synthetic ground truth for benchmarking. It matters because it may be the first openly available 4D-CTA resource for studying aneurysm wall motion, which could help improve non-invasive rupture-risk assessment.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The synthetic ground truth may rest on a sign error in Eq. (1): the printed strain-energy function is non-convex and does not match the cited Raghavan–Vorp model, so the FE displacement field is either nonphysical or misdocumented.","rationale":"The reader's weakest_assumption stressed the FE simplifications (uniform pressure, rigid end constraints, homogeneous wall, literature material parameters). I agree those limit how far the synthetic ground truth can be trusted as a surrogate for real in-vivo motion, but the sign error in Eq. (1) is more immediately load-bearing: it creates an internal inconsistency in the description of the ground truth generation. If the printed equation was actually used, the material model is non-convex and the synthetic deformation field is not even a plausible AAA wall response; if it was not used, the paper misdocuments the construction of the key verification artifact. Either way the dataset's central verification claim is compromised until the discrepancy is resolved. The reader did notice the sign error in the rationale but did not elevate it to the weakest assumption, so my agreement is partial. The Zenodo DOI, open file formats, and inclusion of the actual FE input files are positives that make this concern testable rather than speculative. The 'first publicly available 4D-CTA resource' claim is unverified but not load-bearing for the dataset's usability, so I do not rest the critique on it. Given that the concern can be settled by inspecting the provided FE_model.inp and re-running one simulation, the appropriate outcome remains conditional acceptance pending this check.","tokens_in":8208,"tokens_out":4125,"duration_ms":33519,"concrete_test":"Inspect FE_model.inp in the \"Ground Truth\" folder for the material coefficients and constitutive form actually used in Abaqus. Independently re-run the Patient 1 simulation with the corrected Raghavan–Vorp form W = alpha*(I_B - 3) + beta*(I_B - 3)^2 and compare the resulting nodal displacement field against FE_nodes_displacement.vtu and the synthetic CTA_deformed.nrrd. If the displacement field changes materially, the dataset ground truth was produced with the nonphysical sign; if it matches exactly, Eq. (1) is a typo that must be corrected in the paper. Also verify that the warping transform is lossless by inverse-warping CTA_deformed.nrrd back to CTA_undeformed.nrrd and measuring residual image difference.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The dataset's verification value depends on the FE displacement field used to create the synthetic systolic image. In the \"Ground Truth\" methods section, the wall material is given as W = alpha*(I_B - 3) - beta*(I_B - 3)^2 with alpha = 0.174 MPa and beta = 1.881 MPa. With a minus sign, dW/dI_B = alpha - 2*beta*(I_B - 3), which becomes negative once I_B - 3 exceeds alpha/(2*beta) ~ 0.046. For the wall stretches expected under 13 kPa loading, I_B can readily exceed 3.05, so the strain energy is non-convex and the model is not a valid stable hyperelastic material. This differs from the cited Raghavan–Vorp form W = alpha*(I_B - 3) + beta*(I_B - 3)^2, where both coefficients are positive. Thus either the FE model was run with the printed (nonphysical) form, in which case the synthetic ground truth is not a physiologically plausible deformation and should not be used as a benchmark, or the FE model was run with the correct positive sign, in which case Eq. (1) misdescribes the dataset and the reproducibility claim for the ground truth is broken. The authors' FDA/ASME plausibility argument cannot repair this because the concern is internal inconsistency, not mere model simplification.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This data descriptor presents a publicly available dataset (Zenodo DOI 10.5281/zenodo.15477710) of ECG-gated 4D-CTA image sequences and patient-specific AAA wall geometries for ten patients, together with a synthetic ground-truth package derived from Patient 1's diastolic image: a patient-specific finite element model, nodal displacement field, image-warping transform, and a synthetic systolic 3D-CTA image. The stated purpose is to enable reproduction of the companion kinematics study [1] and to provide a benchmark for non-invasive estimation of AAA wall displacement and strain.","tokens_in":8509,"tokens_out":5142,"duration_ms":41441,"significance":"If the ground truth is physically sound, the dataset fills a clear gap in public resources for AAA kinematic analysis; the open Zenodo deposition, standard NRRD/STL/inp/VTK formats, and explicit file inventory are genuine strengths that should make the resource broadly reusable. The paper also gives a detailed account of the segmentation and geometry extraction pipeline. However, the apparent sign error in the strain-energy function in Eq. (1) currently undermines the central verification claim, because the synthetic ground truth is the only part of the dataset explicitly intended as a benchmark for method verification. This must be resolved before the dataset can be relied upon as a benchmark.","major_comments":[{"comment":"Equation (1) prints W = α(I_B−3) − β(I_B−3)^2 with α = 0.174 MPa and β = 1.881 MPa. With the minus sign, dW/dI_B = α − 2β(I_B−3), which becomes negative once I_B−3 exceeds α/(2β) ≈ 0.046, so the strain energy is non-convex and the material model is not a stable hyperelastic solid. The cited Raghavan–Vorp model [10] uses a plus sign: W = α(I_B−3) + β(I_B−3)^2. Thus either the Abaqus simulation was run with the printed nonphysical model, making the synthetic ground truth non-physiological and unsuitable as a benchmark, or it was run with the correct positive sign, making Eq. (1) a misdescription that breaks the reproducibility claim for the ground truth. Please correct the sign, state explicitly which material definition was used in Abaqus, and confirm that the FE displacement field was generated with a convex, stable material model.","section":"Ground Truth, Eq. (1)"}],"minor_comments":[{"comment":"The claim 'To the best of our knowledge, this dataset is the first publicly available 4D-CTA resource for AAA' is not supported by a systematic literature search; please either provide the search strategy or soften the statement to avoid an unsupported novelty claim.","section":"Background"},{"comment":"The abbreviation I_B should be defined as the first invariant of the left Cauchy–Green tensor at first use, since it appears in Eq. (1) before any definition is given.","section":"Ground Truth"},{"comment":"There is a typo in the Segmentation subsection: 'MATALB' should be 'MATLAB'.","section":"Experimental Design, Materials and Methods"},{"comment":"Per-patient acquisition parameters (voxel spacing, matrix size, reconstruction kernel, contrast protocol, and phase availability) are not tabulated; providing them would materially improve reproducibility and reusability of the image data.","section":"Data Description"},{"comment":"The mesh statistics are limited to element type C3D20H; reporting element and node counts and a mesh-sensitivity or convergence check would strengthen confidence in the FE displacement field used as ground truth.","section":"Ground Truth"},{"comment":"The FDA and ASME guidelines invoked to justify plausibility of the synthetic ground truth are not cited; please provide the specific guidance documents.","section":"Ground Truth"},{"comment":"The statement that each 4D-CTA dataset 'typically' contains ten frames is followed by the information that patients 1, 2, 4, and 8 have fewer frames; please clarify whether missing phases are absent from the released dataset or simply not used in the companion study.","section":"Abstract and Data Description"}],"recommendation":"major_revision","confidential_remarks":"The paper is tightly coupled to companion paper [1], and the dataset's benchmark value depends directly on the correctness of the synthetic ground truth. The sign error in Eq. (1) is the main blocker: the authors should be asked to correct the equation, confirm the actual Abaqus material definition, and either regenerate the ground truth with a convex model or demonstrate that the existing displacement field was produced with the correct positive-sign model. The 'first dataset' claim is an overreach without a systematic search, but it is not by itself disqualifying."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The short version: this is a genuinely useful data descriptor for what appears to be the first public 4D-CTA AAA dataset, and the real-image portion is worth having. But the synthetic-ground-truth section has a likely sign error in Eq. (1) that needs resolving before anyone relies on the generated deformations.\n\nWhat is new and good: ten patient 4D-CTA image sets, systolic surface geometries, and a Zenodo archive with a careful file inventory. The paper is transparent about acquisition, the AI-segmentation-plus-post-processing pipeline, and the limitations (single site, limited phases for some patients). Open formats (NRRD, STL) and clear folder structure make the dataset easy to reuse. If the images and geometries are as described, this fills a real gap.\n\nThe soft spots are real but not uniformly fatal. The sign in Eq. (1) is wrong as printed: W = alpha*(I_B - 3) - beta*(I_B - 3)^2 is not the Raghavan–Vorp model, and it becomes non-convex once I_B - 3 exceeds roughly 0.046, which is essentially immediate under 13 kPa loading. So either the FE simulation ran with a nonphysical material, or Eq. (1) misdescribes the simulation. Either way, the synthetic ground truth's \"physiologically plausible\" status is unsubstantiated. The FDA/ASME plausibility argument cannot repair an internal inconsistency between the printed model and the cited material. This is a load-bearing issue for the ground-truth component, because the displacements used to warp the diastolic image come from that FE model. It is not load-bearing for the real images and geometries, which stand on their own.\n\nAlso minor: the \"first public 4D-CTA AAA dataset\" claim is supported only by \"to the best of our knowledge,\" not a systematic search. That is a small overreach, not a fatal flaw.\n\nWho this is for: researchers in AAA biomechanics and image registration who need shared patient data to benchmark algorithms. The paper deserves a serious referee; a data descriptor with a plausible first-of-its-kind resource should not be desk-rejected. I would recommend conditional acceptance: the authors must correct the sign in Eq. (1) (if it is a typo) or rerun the FE model with the correct material law and regenerate the ground truth, and they should also demonstrate the resulting displacement field is plausible. The real-image dataset can proceed regardless, but the synthetic ground truth needs to be trustworthy before it is used as a benchmark.","headline":"Useful 4D-CTA AAA dataset with a solid real-image component, but the synthetic ground truth is built on a likely sign error in Eq. (1) that must be fixed before the benchmark is trustworthy.","tokens_in":9086,"tokens_out":2131,"would_cite":true,"duration_ms":21688,"reading_group":"maybe","serious_thinker":"unclear","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper releases a public 4D-CTA dataset for abdominal aortic aneurysm kinematics, with ten patients' cardiac-gated CT images, geometries, and finite-element synthetic ground truth.","keywords":["Abdominal aortic aneurysm","Patient-specific analysis","Wall displacement","Wall strain","Image registration","Computed tomography angiography","Biomechanics","Non-invasive method"],"falsifier":"Download the dataset and run an independent image-registration algorithm on Patient 1's diastolic and synthetic-systolic images; if the recovered displacement field does not match the provided FE nodal displacements within a few voxels, the ground truth's internal consistency is called into question. Separately, a systematic search of public repositories for any earlier public 4D-CTA AAA dataset would directly test the 'first publicly available' claim.","tokens_in":8042,"feed_emoji":"🫀","tokens_out":5879,"duration_ms":42101,"temperature":0.7,"pith_summary":"This paper establishes a dataset rather than a new measurement or model: it provides, to the authors' knowledge, the first publicly available time-resolved CT angiography (4D-CTA) resource for abdominal aortic aneurysm (AAA). The data cover ten patients, with ECG-gated 3D-CTA frames sampled across the cardiac cycle, patient-specific external-wall geometries, and a synthetic ground-truth package built from a finite element simulation of one patient's aneurysm. A sympathetic reader would care because the dataset lets independent groups develop and test non-invasive, image-based methods for measuring AAA wall displacement and strain, an area that has received far less attention than wall-stress analysis. The paper also supplies the materials needed to reproduce the kinematic results reported in its companion biomechanics study.","feed_headline":"First public 4D-CTA dataset tracks aneurysm wall motion","feed_subtitle":"Ten patients' cardiac-cycle CT scans plus synthetic ground truth let researchers test AAA strain measurement.","key_machinery":"The mechanism that carries the argument is the pairing of real multi-phase imaging with a synthetic known-deformation target. For each patient, ECG-gated CTA frames are acquired at ten-percent intervals of the R-R interval (with some patients having fewer phases), and the systolic frame is segmented through an automated pipeline to produce a triangulated external-wall surface. For the ground truth, a finite element simulation on a hexahedral mesh supplies a displacement field for Patient 1; a B-spline scattered transform built from the FE nodal coordinates warps the diastolic image into a synthetic systolic image, so the true wall displacement is known by construction.","core_discovery":"The paper's central claim is that this collection—4D-CTA images and AAA geometries for ten patients plus a finite-element-derived synthetic systolic image with known wall displacements—constitutes a reusable benchmark for AAA kinematic analysis. To the authors' knowledge it is the first publicly available 4D-CTA resource for AAA. The ground-truth portion is generated by creating a patient-specific hexahedral finite element model of Patient 1's AAA, applying a uniform 13 kPa internal pressure with rigidly constrained ends and literature-based hyperelastic wall parameters, and using the computed nodal displacement field to warp the diastolic-phase CTA image into a synthetic systolic image. The stated purpose is method verification: image-registration algorithms can be checked against known displacements, and the companion study's results can be reproduced.","pith_inferences":["This benchmark could make AAA wall strain a practical endpoint for multi-center studies, in the way that public datasets have accelerated other image-analysis fields.","The simplified FE ground truth (uniform pressure, rigid ends, literature material constants) is a simplification; a natural extension would be to generate multiple synthetic frames across the cardiac cycle and to vary the FE assumptions, so methods can be stress-tested against realistic modeling uncertainty.","The irregular temporal sampling between patients (10, 7, or only 2 phases) is itself a stress test: methods validated on this dataset will need to handle sparse and uneven cardiac-phase coverage, which mirrors real clinical variability."],"forward_implications":["Anyone can download the dataset and reproduce the wall-displacement and strain results of the companion study [1].","The synthetic ground truth gives registration and strain-measurement methods a quantitative target to be checked against, without requiring invasive markers.","Because the images are stored as NRRD and geometries as STL, the data can be used across different open-source and commercial analysis platforms.","The pipeline's reliance on standard clinical CT scanners means the approach could be applied wherever ECG-gated CT angiography is available."],"supporting_citations":[{"why":"Companion study defining the image-registration kinematic method this dataset supports and whose results can be reproduced.","marker":"[1]"},{"why":"Software system used to build the patient-specific AAA surface geometries from segmentations.","marker":"[2]"},{"why":"Open-source platform used for surface extraction and for applying the warping transform.","marker":"[3]"},{"why":"Scattered transform technique that converts the FE displacement field into an image warping transform.","marker":"[4]"},{"why":"Method for generating the patient-specific structured hexahedral mesh used in the ground-truth FE model.","marker":"[7]"},{"why":"Source of the hyperelastic constitutive model and the material parameters used in the FE simulation.","marker":"[10]"},{"why":"Additional source for the AAA wall material parameters adopted in the model.","marker":"[11]"},{"why":"In vivo wall-stress study supporting the same material parameter values.","marker":"[12]"}],"fun_headline_variants":["First public 4D-CTA dataset for AAA wall motion tracking","Open 4D-CTA collection reveals aneurysm pulsation in 10 patients","New 4D-CTA benchmark comes with synthetic ground truth for validation","Validate AAA strain with first public 4D-CTA dataset"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a finite element model using uniform 13 kPa pressure, rigidly constrained ends, and literature-based tissue parameters produces a 'physiologically plausible' displacement field good enough to stand in for real AAA wall motion when verifying new methods; if that simplification fails to represent in-vivo motion, algorithms tuned to this benchmark may not transfer to patients.","fun_headline_variants_meta":{"raw":{"variants":["First public 4D-CTA dataset for AAA wall motion tracking","Open 4D-CTA collection reveals aneurysm pulsation in 10 patients","New 4D-CTA benchmark comes with synthetic ground truth for validation","Validate AAA strain with first public 4D-CTA dataset"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000822,"raw_usage":{"total_tokens":3645,"prompt_tokens":1039,"completion_tokens":2606,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":655,"completion_tokens_details":{"reasoning_tokens":2529}},"tokens_in":655,"tokens_out":2606,"duration_ms":15095,"temperature":1.0,"reasoning_tokens":2529,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:42:57.051809+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Download the dataset and run an independent image-registration algorithm on Patient 1's diastolic and synthetic-systolic images; if the recovered displacement field does not match the provided FE nodal displacements within a few voxels, the ground truth's internal consistency is called into question. Separately, a systematic search of public repositories for any earlier public 4D-CTA AAA dataset would directly test the 'first publicly available' claim.","supporting_citations":[{"cited_title":"Joldes, K","cited_arxiv_id":null,"evidence_quote":"Software system used to build the patient-specific AAA surface geometries from segmentations."},{"cited_title":"Joldes, A","cited_arxiv_id":null,"evidence_quote":"Scattered transform technique that converts the FE displacement field into an image warping transform."},{"cited_title":"Generation of Patient-specific Structured Hexahedral Mesh of Aortic Aneurysm Wall","cited_arxiv_id":"2206.06175","evidence_quote":"Method for generating the patient-specific structured hexahedral mesh used in the ground-truth FE model."},{"cited_title":"Raghavan, D.A","cited_arxiv_id":null,"evidence_quote":"Source of the hyperelastic constitutive model and the material parameters used in the FE simulation."},{"cited_title":"Raghavan, D.A","cited_arxiv_id":null,"evidence_quote":"Additional source for the AAA wall material parameters adopted in the model."},{"cited_title":"Spectral neighbor joining for reconstruction of latent tree models","cited_arxiv_id":"2002.12547","evidence_quote":"In vivo wall-stress study supporting the same material parameter values."}],"review_version":1}