{"id":"87b6294a-a846-46b1-a9fb-fb17014c9d1d","arxiv_id":"2505.12485","paper_version":2,"verdict":"REJECT","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"high","formal_verification":"none","parameter_count":4,"one_line_summary":"A combined 1D-2D axisymmetric CFD workflow estimates FFR and iFR for intermediate coronary stenosis, predicting decreasing values with higher severity and longer lesion length.","lead":"This paper combines a fast one-dimensional model of blood flow through the coronary arteries with a more detailed two-dimensional model of a narrowed segment, aiming to estimate two clinical indices of blockage severity: FFR and iFR. It reports that both indices drop as the degree of stenosis rises and as the blockage gets longer.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 2D model receives its outlet pressure from the 1D stenosis simulation, so FFR evaluated at that outlet is a 1D result; no FFR definition or extraction point is given, and reported values conflict with the imposed pressure.","rationale":"The reader's weakest assumption identifies the same load-bearing concern: the 2D model's outlet pressure is inherited from the 1D model, making FFR either predetermined or undefined. My analysis confirms this and adds the internal inconsistency with the reported FFR values. The paper provides no FFR definition, no extraction location, and no independent validation against invasive FFR measurements. The grid independence and literature validation of velocity/pressure waveforms are positive but do not address the diagnostic claim. Therefore, the REJECT verdict remains appropriate. I see no additional objection that would shift the verdict further; the central issue is sufficient.","tokens_in":6972,"tokens_out":4930,"duration_ms":52169,"concrete_test":"Perform a sensitivity study on the 70% severity case: rerun the 2D axisymmetric simulation with the outlet pressure changed from 93.303 mmHg to 102.6 mmHg (a 10% increase), keeping all other BCs identical. Then compute FFR using the authors' (unstated) definition and extraction point, and compare to the original. If FFR shifts by approximately 10% (in absolute pressure terms), the FFR is directly controlled by the imposed outlet pressure, demonstrating that the 2D model provides no independent diagnostic information. If FFR changes only minimally, then the FFR is extracted upstream and the authors must specify that location; either way, the test forces clarification of the FFR definition and the role of the 2D model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that the 1D-2D coupling reliably predicts FFR/iFR for intermediate stenosis. For the 2D model to add independent diagnostic information, the distal pressure (the numerator of FFR) must be computed from the local flow solution, not imposed. However, the paper imposes the outlet pressure directly from the 1D model (Table 2, e.g., 93.303 mmHg at 70% severity). Since the 1D model already includes the stenosed segment, this outlet pressure already reflects the stenosis-induced pressure drop. If FFR is taken at that outlet, it is inherited verbatim from the 1D solution, making the 2D model redundant for the primary index. If FFR is taken upstream, the paper never defines the extraction location or the FFR formula, so the reported values cannot be reproduced or verified. Moreover, the numbers themselves are suspicious: with an aortic pressure near 100 mmHg, an outlet pressure of 93.303 mmHg would imply FFR ≈ 0.93, yet the paper reports FFR ≈ 0.68–0.76 at 70% severity. This suggests either a different aortic pressure, a different FFR location, or an error. Without a precise definition and extraction point, the headline FFR/iFR values are not anchored to the simulation setup. The conclusion that the method is a 'reliable and non-invasive diagnostic tool' therefore rests on an unstated and likely circular boundary-condition dependency.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a reduced-order CFD approach combining a 1D arterial tree model with a 2D axisymmetric model of a stenosed coronary artery, using CT-derived geometry and non-Newtonian rheology, to estimate Fractional Flow Reserve (FFR) and instantaneous wave-free ratio (iFR) for intermediate-grade stenoses (40%, 50%, 70%). The 1D model provides global flow/pressure boundary conditions for the 2D model; FFR and iFR are reported as decreasing with severity and with longer lesion length. The authors claim the method is a reliable and non-invasive diagnostic tool for coronary stenosis evaluation.","tokens_in":7286,"tokens_out":6554,"duration_ms":65475,"significance":"If substantiated, a low-cost patient-specific FFR/iFR prediction tool of this type would be clinically valuable. The paper's strengths include grid-independence checks for both solvers and validation against published waveforms and steady-flow stenosis data. The 1D-to-2D hierarchical coupling is a reasonable idea in principle. However, the central claim of reliable FFR/iFR assessment is not supported: FFR is not defined or located, the 2D model inherits its outlet pressure from the 1D model, no hyperemic simulation is provided (which is standard for FFR), and there is no clinical validation against invasive FFR. The manuscript is a methodological sketch rather than a demonstrated diagnostic tool.","major_comments":[{"comment":"FFR is never defined and the location where distal pressure is extracted is never stated. The paper reports numerical FFR values but gives no formula such as FFR = Pd/Pa and no specification of the temporal averaging (mean over the cycle, at a particular phase, etc.). This makes the headline results unreproducible. More importantly, the 2D model receives its outlet pressure as a Dirichlet boundary condition generated directly from the 1D model (Table 2). If FFR is evaluated at that outlet, its value is inherited verbatim from the 1D pressure solution, so the higher-order 2D model adds no independent information about the primary clinical index. The manuscript must define FFR/iFR precisely, state the extraction point, and show how the local 2D pressure field (if used) enters the calculation.","section":"3. Result and discussion; Table 2"},{"comment":"The reported FFR values are numerically inconsistent with the imposed outlet pressures. For 70% severity, the outlet pressure is 93.303 mmHg. With an aortic pressure near 100 mmHg (as suggested by the inlet waveform), FFR at that outlet would be about 0.93, yet the text reports FFR values of 0.68 (Newtonian), 0.76 (power-law), 0.73 (Carreau), and 0.65 (Casson). If the reference aortic pressure differs from 100 mmHg, or if FFR is computed at an interior location (where pressure would be higher, not lower, than the outlet), this must be stated explicitly. As written, the reported FFR values cannot be reconciled with the stated boundary conditions.","section":"Table 2; 3. Result and discussion"},{"comment":"FFR is physiologically defined under maximal hyperemia, where coronary microvascular resistance is minimized. The model applies a resting-style sigmoid pressure waveform at the inlet and a zero-reflection resistance at the outlet, with no representation of hyperemia (e.g., reduced distal resistance or adenosine-induced vasodilation). Computing FFR from a resting-state simulation therefore does not correspond to the index used clinically. The authors should either introduce a hyperemia model or explicitly argue why resting conditions are sufficient; without this, the claim that the computed values are 'FFR' is physiologically unsupported.","section":"2.1.2 Boundary conditions; 3. Result and discussion"},{"comment":"The manuscript repeatedly claims that the geometry is patient-specific and derived from 'Multi-slice CT scans', yet no details of the imaging-to-model pipeline are given: no segmentation method, no patient characteristics, no imaging parameters, and no demonstration that the analytical stenosis equation (which models area variation with axial position) matches any real patient's anatomy. The analytical equation appears to be a generic parameterization, not a patient-derived geometry. The reliability of the diagnostic prediction cannot be assessed without this information. Either provide the full patient-specific pipeline or temper the claim to 'representative stenosis geometries'.","section":"2.1 Computational domain"},{"comment":"The 2D axisymmetric model imposes a single time-independent pressure value at the outlet (Table 2), while the inlet flowrate is a pulsatile waveform (Figure 5). The paper does not state whether the outlet pressure is a mean, peak, end-diastolic, or time-varying value, nor how a constant distal pressure is physically consistent with pulsatile inflow. This is not merely a presentation detail: the pressure drop computed by the 2D model (and any FFR derived from it) depends critically on the temporal nature of the outlet boundary condition. The outlet condition must be fully specified and justified.","section":"2.2.2 Boundary Conditions"}],"minor_comments":[{"comment":"The text claims that all non-Newtonian models predict higher FFR than the Newtonian model, but the reported Casson value at 70% (0.65) is lower than the Newtonian value (0.68). The Conclusion even states that 'the Carreau and Casson models yield slightly lower values than the Newtonian model'. Please correct this inconsistency in the interpretation of the model comparison.","section":"3. Result and discussion; Conclusion"},{"comment":"The iFR results are presented only as a qualitative description with a figure; no numerical values or a definition of the wave-free period are provided. Since iFR is a central claim of the paper, a formula and a table or precise values from the figure are needed.","section":"Figure 9"},{"comment":"The captions of Figures 8 and 11 are identical ('1D boundary condition given inlet to arterial tree') and do not match the content; Figure 11 appears to plot pressure drop versus severity. Please correct the captions.","section":"Figures 8 and 11"},{"comment":"There is a typographical error: 'Carreau Yesuda' should be 'Carreau–Yasuda'.","section":"2.2.1 2D Numerical details"},{"comment":"References [20] and [21] are on microfluidic particle sorting and do not appear to be relevant to the 2D stenosis validation; please cite appropriate studies for the stenotic flow validation.","section":"References"},{"comment":"The 'grey zone' for FFR is described as 0.76–0.80, whereas the cited literature (e.g., reference [16]) commonly uses 0.75–0.80; please align the threshold with the cited sources.","section":"3. Result and discussion"}],"recommendation":"reject","confidential_remarks":"The manuscript reads as a conference extended abstract and does not meet the evidence standards of a full journal paper. The load-bearing deficiencies (no FFR definition, circular boundary-condition dependency, no hyperemia, internal numerical inconsistency, and no clinical validation) cannot be fixed by a minor revision; they require a reworking of the modeling strategy and additional validation data. I also note that several central components rely heavily on the corresponding author's own previous work (references [5], [10], [11], [18]), so the novelty claim should be carefully positioned. I do not see a viable path to acceptance in its current form."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a classic case of a reasonable engineering workflow overreaching into a clinical claim. The 1D arterial network is standard and the waveform validation against Low et al. checks out. The 2D axisymmetric validation against Varghese et al. is also fine. What's new is the specific 1D-generated pressure/flow boundary conditions fed into a 2D local model with several non-Newtonian rheology options, plus lesion-length sweeps. If the paper framed itself as \"a reduced-order model for local hemodynamics with inherited distal pressure,\" I'd have few complaints.\n\nThe soft spots are load-bearing. First, FFR and iFR are never defined. No formula, no extraction location, no aortic pressure for normalization. With the outlet pressure set to 93.303 mmHg at 70% severity, an FFR of ~0.68–0.76 implies a mean aortic pressure of 122–137 mmHg, which is never stated. Second, the 2D model takes the outlet pressure directly from the 1D model (Table 2). Because the 1D model already includes the stenosed segment, the distal pressure and therefore any FFR computed at the outlet is inherited, not independently predicted. The \"higher-order\" model adds local velocity detail but no independent diagnostic information. Third, the numbers are internally inconsistent: at 70% severity the text says the Newtonian model gives 0.68 and non-Newtonian models higher (power-law 0.76, Carreau 0.73), but then lists Casson at 0.65—below the Newtonian value. The conclusion then claims Carreau and Casson yield \"slightly lower values than the Newtonian model,\" contradicting the results. Fourth, the abstract promises patient-specific CT, but no CT images or reconstruction details appear; the geometry is an analytical stenosed tube.\n\nThe presentation is sloppy—Figure 8 caption says \"1D boundary condition given inlet to arterial tree\" while it is presumably a velocity profile, and Figure 11 has the same caption for a pressure-drop plot. These are minor but consistent with a rushed submission.\n\nBottom line: the underlying 1D framework is solid, and the coupling concept is worth exploring, but the clinical parameter claim is unsupported as written. I would send this to a referee who knows coronary hemodynamics, because a solid referee might extract a publishable paper after major revision: define FFR/iFR explicitly, compute distal pressure from the 2D solution rather than imposing it, resolve the model ranking, and dial the claims down. In its current form, it does not support \"reliable diagnostic tool.\"","headline":"A reasonable 1D-2D coupling study whose clinical FFR/iFR claim is unsupported: the distal pressure is inherited from the 1D model, FFR/iFR are never defined, and the reported numbers are internally inconsistent.","tokens_in":7851,"tokens_out":4212,"would_cite":false,"duration_ms":42004,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper argues that a 1D-2D axisymmetric CFD hierarchy, fed by CT-derived geometry and non-Newtonian blood rheology, can non-invasively quantify intermediate coronary stenosis by reproducing FFR/iFR trends from above 0.92 at 40%…","keywords":["coronary artery stenosis","fractional flow reserve","instantaneous wave-free ratio","reduced-order CFD","1D arterial network","2D axisymmetric model","non-Newtonian blood rheology","lesion length"],"falsifier":"Take the 70% stenosis case and run the 2D axisymmetric model with a coupled outflow resistance so the distal pressure is computed rather than prescribed at 93.303 mmHg; if the resulting FFR at the standard distal measurement point moves materially across the 0.75–0.80 decision zone, the higher-order model is not contributing independent clinical information.","tokens_in":6762,"feed_emoji":"🫀","tokens_out":8339,"duration_ms":75748,"temperature":0.7,"pith_summary":"This paper argues that a reduced-order computational fluid dynamics chain — a one-dimensional model of the whole arterial tree feeding boundary conditions to a two-dimensional axisymmetric model of the narrowed segment — can turn CT-derived geometry into a non-invasive estimate of coronary stenosis severity. Fractional Flow Reserve (FFR) and instantaneous wave-free ratio (iFR), two pressure-based indices clinicians use to decide between stenting and medical therapy, both fall as stenosis severity rises: above 0.92 at 40% narrowing, down to about 0.65–0.76 at 70%, with the exact value depending on how blood's shear-thinning viscosity is modelled. The paper also reports that a 3 cm lesion produces lower FFR and iFR than a 1 cm lesion at the same percentage blockage. If the approach holds, it would offer a fast, non-invasive alternative to invasive pressure-wire measurement in the intermediate-grade grey zone where treatment decisions are hardest.","feed_headline":"FFR and iFR drop with stenosis severity in a CT-fed flow model","feed_subtitle":"A two-stage 1D-to-2D simulation estimates the two clinical indices without an invasive pressure wire.","key_machinery":"The engine is the 1D-2D axisymmetric reduced-order chain. The 1D layer solves mass and momentum conservation on a 61-segment arterial tree using forward and backward characteristic variables $W_1$ and $W_2$, reconstructing cross-sectional area $A$ and velocity $u$, with a sigmoid pressure waveform at the inlet and a zero-reflection resistance model at the outlet. That layer produces the time-varying flow rate and the outlet pressure (for instance 93.303 mmHg at 70% stenosis) imposed on a 2D axisymmetric Navier-Stokes solver for the stenosed segment, which uses Newtonian, power-law, Carreau-Yasuda, or Casson viscosity to compute the local velocity and pressure fields from which FFR and iFR are derived.","core_discovery":"On the paper's own terms, the discovery is that the two-tier reduced-order model reproduces the expected physiology of intermediate stenosis while remaining cheap enough to run over one cardiac cycle. FFR and iFR decline monotonically with severity under every rheology model, FFR staying above 0.92 at 40% and reaching roughly 0.65–0.76 at 70%; the power-law fluid offers the least resistance and the highest FFR/iFR, while the Casson and Newtonian models sit at the low end. For a fixed severity, lengthening the lesion from 1 cm to 3 cm lowers both indices, evidence that lesion length should be read alongside percentage stenosis. The paper presents these trends as the basis for a non-invasive diagnostic tool that captures global flow conditions from the 1D tree and local pressure-flow detail from the 2D axisymmetric stenosis model.","pith_inferences":["The pressure hand-off from 1D to 2D means the 2D model's FFR is largely inherited from the 1D distal pressure, so re-running the same stenosis with a self-consistently computed outlet pressure is a direct way to test how much independent information the higher-order model adds.","The model's monotone severity–FFR curve invites a calibration study against invasive FFR/iFR on the same CT data, which could turn these computed values into decision thresholds with confidence intervals.","Because the local model is axisymmetric, a natural extension is a same-patient comparison against full 3D geometry to identify the plaque shapes and curvatures where the axial symmetry assumption changes FFR by more than the grey-zone tolerance.","The lesion-length data could support a composite index, such as severity weighted by lesion length or pressure drop per unit length, that the paper does not define but its results suggest would sharpen the intermediate-grey-zone decision."],"forward_implications":["Clinicians could obtain FFR and iFR for intermediate lesions directly from CT data, bypassing invasive pressure-wire measurement.","At equal percentage stenosis, lesion length shifts the functional result: a 3 cm lesion lowers FFR and iFR relative to a 1 cm lesion, so length should enter the intervention decision.","The choice of blood rheology changes the clinical reading: at 70% severity the Newtonian model gives FFR about 0.68 while the power-law model gives 0.76, a spread that crosses the 0.75–0.80 decision boundary.","Because the 1D tree supplies the boundary conditions, the approach avoids full 3D coronary simulation and can run the local 2D model with about 200,000 elements over one cardiac cycle."],"supporting_citations":[{"why":"Supplies the one-dimensional mass and momentum equations and the pressure-area wall relation used for the arterial tree.","marker":"[3]"},{"why":"Provides the locally conservative Galerkin method and the coronary circulation relations that build the 61-segment 1D network and its boundary-condition treatment.","marker":"[4]"},{"why":"Is the benchmark whose right-carotid pressure and flow waveforms validate the 1D solver.","marker":"[12]"},{"why":"Is the direct numerical simulation of stenotic flow used to validate the 2D axisymmetric velocity field.","marker":"[14]"},{"why":"Establishes the 1D-versus-3D FFR comparison that the reduced-order 1D-2D approach extends.","marker":"[5]"},{"why":"Supplies the earlier numerical study of occlusion percentage and lesion length that motivates the lesion-length analysis.","marker":"[10]"},{"why":"Provides the prior reduced-order stenosed-coronary model whose stenosis representation and haemodynamic setup are extended here.","marker":"[11]"},{"why":"Supplies the non-Newtonian viscosity coefficients used for the power-law, Carreau-Yasuda, and Casson models.","marker":"[13]"}],"fun_headline_variants":["FFR and iFR fall as stenosis worsens in CT-fed model","Non-invasive 1D-2D CFD predicts FFR and iFR from CT","Stenosis severity and lesion length lower FFR, iFR in silico","CT-based reduced-order model estimates FFR and iFR without wire","Two-tier CFD ties stenosis severity to FFR and iFR drop"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the 1D arterial-network pressure at the stenosis outlet is accurate enough to be imposed unchanged on the 2D model, since FFR is defined from that pressure and the 2D solver cannot independently correct the main clinical number.","fun_headline_variants_meta":{"raw":{"variants":["FFR and iFR fall as stenosis worsens in CT-fed model","Non-invasive 1D-2D CFD predicts FFR and iFR from CT","Stenosis severity and lesion length lower FFR, iFR in silico","CT-based reduced-order model estimates FFR and iFR without wire","Two-tier CFD ties stenosis severity to FFR and iFR drop"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000199,"raw_usage":{"total_tokens":1389,"prompt_tokens":983,"completion_tokens":406,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":599,"completion_tokens_details":{"reasoning_tokens":307}},"tokens_in":599,"tokens_out":406,"duration_ms":4223,"temperature":1.0,"reasoning_tokens":307,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T20:33:13.297891+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the 70% stenosis case and run the 2D axisymmetric model with a coupled outflow resistance so the distal pressure is computed rather than prescribed at 93.303 mmHg; if the resulting FFR at the standard distal measurement point moves materially across the 0.75–0.80 decision zone, the higher-order model is not contributing independent clinical information.","supporting_citations":[{"cited_title":"grey zone,","cited_arxiv_id":null,"evidence_quote":"Supplies the one-dimensional mass and momentum equations and the pressure-area wall relation used for the arterial tree."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the prior reduced-order stenosed-coronary model whose stenosis representation and haemodynamic setup are extended here."},{"cited_title":"Numerical investigation of blood flow through stenosed coronary artery using reduced order model","cited_arxiv_id":null,"evidence_quote":"Supplies the non-Newtonian viscosity coefficients used for the power-law, Carreau-Yasuda, and Casson models."}],"review_version":1}