{"id":"c8084939-3246-442f-b63f-8507ef25f694","arxiv_id":"2504.19326","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"A thermally drawn 16-electrode catheter combined with a new dynamic time warping pipeline localizes itself inside a vascular phantom in real time using only electric impedance and distance estimates.","lead":"Researchers built a custom 16-electrode catheter using 3D printing and thermal drawing, then used its electric impedance signals to track its position inside a model blood vessel in real time without X-rays. The system was tested by an experienced surgeon navigating the catheter toward a target in the phantom, achieving millimeter-level accuracy in the best runs.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Reported 3-8 mm accuracies depend on electrical distance estimates the paper finds biased by up to -10.2% and 4.4x noisier than [22]; whether these errors survive the resampling and anchor-propagation steps is not quantified.","rationale":"The paper is an honest engineering study with independent support: mechanical characterization, FEM-simulated signals matching recordings, and a working real-time pipeline in a phantom. The central claim, however, is not just that the catheter or the simulation works, but that the fused tracking achieves specific millimeter-level accuracies in real time. The fusion mechanism makes those accuracies directly dependent on distance estimates: the same estimates are used to condition the signal before DTW and to compute the final position in A2. The paper itself documents that these estimates are biased and noisy in exactly the large-diameter sections of the phantom where off-center catheter motion is possible. Without an ablation or sensitivity analysis, the reader cannot tell whether devDTW is correcting the distance errors or inheriting them. This concern is more load-bearing than the outlier-exclusion issue, because it affects the validity of the accuracy numbers even for the retained runs and also bears on the authors' own statement that distance underestimation is expected in tissue. The proposed test is a direct offline ablation using already collected data, so it is cheap and decisive. This does not change the CONDITIONAL verdict; it identifies the condition that must be checked before the numbers can be trusted.","tokens_in":19250,"tokens_out":7732,"duration_ms":83111,"concrete_test":"Re-run the devDTW/A2 pipeline offline on the recorded navigation runs, substituting EM-tracker-derived distance for the electrical distance in Section 4.11 (monotonicity filtering, 1 mm resampling) and in A2 anchor propagation, leaving all other parameters and the reference database unchanged. Compare the resulting tracking and target RMS errors with the published Figure 8 values. If the errors improve by more than about 2 mm, the electrical distance errors are load-bearing and the reported accuracy is not robust; if the errors are unchanged within the run-to-run spread, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central quantitative claim is the 8.3 mm RMS tracking and 3 mm RMS target accuracy for the devDTW/A2 pipeline. Both numbers inherit the quality of the electrical distance estimates, because Section 4.11 uses those estimates to discard non-monotonous segments and resample the live impedance signal at 1 mm intervals, and approach A2 adds the estimated distance dS to an anchor's matched reference index to produce the final position. Section 2.4.2 reports mean distance errors between +1.1% and -10.2% of traveled distance, standard deviations up to 4.4x higher than [22], and forward/backward differences up to 6.9%; errors are strongest in large-diameter sections S3-S5 where off-center catheter motion is most likely. The authors also note systematic underestimation is expected in real tissue. The paper does not report an ablation replacing electrical distance with EM ground-truth distance in the preprocessing and anchor steps, so it is unknown whether the published accuracies are robust to these distance errors or instead absorb them. If the distance errors propagate, the 3 mm target accuracy is a property of this phantom and this catheter configuration, not of the fusion method, and the 'first demonstration' claim becomes considerably weaker.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript presents two contributions. First, it describes a manufacturing pipeline for bespoke 6 Fr electrode catheters using 3D-printed preforms, thermal drawing, femtosecond laser micromachining, and laser spot welding, yielding 16 electrodes, a guidewire channel, and an auxiliary sensor lumen; the catheter is mechanically characterized against a commercial device and electrically validated in simple and anatomically realistic phantoms. Second, it introduces a real-time localization pipeline that fuses tetrapolar impedance-based vascular feature matching via a new deviation-based dynamic time warping (devDTW) algorithm with electric displacement estimates from prior work [22], evaluated in an AAA phantom with an expert endovascular surgeon and a lay person navigating without fluoroscopy. The claimed headline results are 8.3 mm RMS tracking accuracy for the robust-feature approach A2 and an average 3 mm RMS target accuracy for the anchor-based technique. This is framed as the first demonstration of continuous, real-time, fluoroscopy-free, bioelectric-only catheter localization.","tokens_in":19516,"tokens_out":3809,"duration_ms":43716,"significance":"If the quantitative claims hold, this would be a meaningful step for bioelectric navigation: it moves from post-hoc branch classification to real-time continuous centerline localization, and it demonstrates that an experienced surgeon can act on the feedback. The engineering contribution is substantial: the thermally drawn, wire-embedded catheter with bespoke electrode placement is a genuine advance in prototyping capability, and the electrical validation against FEM simulations is a strength. The authors are also transparent about distance-estimation errors, radial-position-dependent feature visibility, and the single-phantom nature of the evaluation, all of which are useful for the community. However, the central accuracy claims rest on favorable evaluation conditions and on an unquantified dependence on the electric distance estimator, so the significance is currently more at the level of a promising feasibility demonstration than an established performance level.","major_comments":[{"comment":"The headline tracking accuracy of 8.3 mm RMS and the 3 mm RMS target accuracy are computed over the best four of five runs after excluding the worst run, but no outlier criterion is specified and the distribution of the excluded runs is not reported. Since the manuscript explicitly notes that some runs had high target errors, especially with approach A1, the reported numbers may substantially overstate the typical performance. The abstract and discussion state these numbers without qualification, which is misleading. Please report per-run errors, the outlier exclusion rule, all-run statistics, and per-user results, and qualify the headline numbers accordingly.","section":"Section 2.4.3, Figure 8A/B"},{"comment":"The tracking pipeline critically depends on electric distance estimates from [22]: Section 4.11 uses them to discard non-monotonous signal segments and resample live impedance signals at 1 mm intervals, and approach A2 propagates anchor positions by adding the estimated distance dS to the matched reference index. Yet Section 2.4.2 reports mean distance errors up to -10.2% of traveled distance, standard deviations up to 4.4 times those in [22], forward/backward differences up to 6.9%, and the strongest underestimation in the large-diameter sections S3-S5. The paper does not quantify how these distance errors propagate into the final position estimates; no ablation replacing the electric distance with EM ground-truth distance in the preprocessing and anchor steps is provided. Without such an analysis, it is unclear whether the reported 8.3 mm and 3 mm accuracies are intrinsic to the fusion method or are absorbing favorable phantom-specific distance estimates. This is the central load-bearing gap in the tracking evaluation.","section":"Sections 4.11, 2.4.3 (approach A2), and 2.4.2"},{"comment":"The reference database used for devDTW is built from FEM simulations of the same silicon AAA phantom used in the evaluation, with off-centerline trajectories hand-designed from that same phantom. The reported accuracies are therefore in-distribution measures: they demonstrate that the pipeline can localize a catheter inside a phantom whose simulated reference signals were created from the same geometry, conductivity, and catheter trajectory assumptions. No test on a different phantom geometry, a different saline conductivity, or a different catheter configuration is presented. The Discussion acknowledges some related limitations, but the quantitative claims in the abstract and conclusion should be explicitly scoped as single-phantom, in-distribution results, or supported by at least one additional validation condition.","section":"Section 2.4.3, Figure 7D-a"},{"comment":"The real-time navigation evaluation involves one expert endovascular surgeon and one lay person, with no indication of how many repeated runs per user were performed beyond the 'five runs' used for the best-four analysis. With a single expert participant and no statistical model of run-to-run variability, the comparison between approaches A1 and A2 and the reported RMS values have very limited inferential power. This is acceptable for a feasibility demonstration, but the text should avoid presenting the numerical accuracies as stable performance characteristics; per-run trajectories, medians, and confidence intervals should be shown, and the single-participant nature of the usability claims (including the NASA-TLX results) should be stated prominently.","section":"Section 2.4.3, Figure 8"}],"minor_comments":[{"comment":"There is a typo: 'nozzel temperature' should be 'nozzle temperature', and the superscripts in 'Tg=112 o− 113oC' should be formatted consistently.","section":"Section 4.1"},{"comment":"The manufacturer name 'Elastrat S` arl' appears garbled; please correct the company name and add details of the phantom model.","section":"Section 2.4.2"},{"comment":"Please state explicitly how many runs each user performed in each scenario, and whether 'best four' means best four of five per user/scenario or best four of the pooled runs; the current text is ambiguous.","section":"Section 2.4.3 and Figure 8"},{"comment":"The phrase '17 times per second' is system-specific; clarify that this is the sampling rate of the Eliko system with the chosen multiplexer configuration.","section":"Section 4.11"},{"comment":"The manuscript uses 'S` arl.', 'Dassult', 'continously', and other small typographical errors; a careful proofreading pass is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid feasibility study with a credible manufacturing contribution, but the headline accuracy claims are not yet supported at the level asserted. The missing ablation that replaces electric distance with EM ground truth in the preprocessing and anchor propagation is the key technical issue, and the best-four-of-five reporting compounds it. I would encourage the editor to prioritize these points in revision. The self-citation to [22] is appropriate, but the distance estimator variability reported in Section 2.4.2 should be treated as a first-order uncertainty source for the tracking claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid engineering paper with a genuinely new contribution: a 16-electrode catheter made by thermal drawing with wire feeding, laser micromachining, and laser welding. The manufacturing pipeline is detailed, the mechanical characterization is thorough, and the impedance signals match FEM simulations well. The devDTW fusion of vascular feature matching with electric distance estimation is also new as a real-time pipeline, and the authors are honest about the radial dependence of renal detection and the limitations of their distance estimator. The soft spots are real but mostly in proportion. The headline 3 mm RMS target accuracy is based on five runs, excluding the worst run in some analyses, with one surgeon and one lay person, and the same phantom used to build the reference database. That is a thin evidence base for a claim of general capability. More importantly, the stress-test concern is correct: the pipeline's preprocessing (Section 4.11) and the A2 anchor propagation both rely on electric distance estimates, and Section 2.4.2 reports those estimates are biased up to -10.2% and 4.4 times noisier than the prior work, especially in the large-diameter sections. The paper never reports an ablation replacing those distance estimates with EM ground truth, so you cannot tell whether the published accuracies are robust to the distance errors or already absorbing them. That is a load-bearing gap, not a minor omission. On the other hand, the circularity concern is not serious: tracking accuracy is measured against an external EM tracker, so the numbers are not fitted to the output. The distance estimator is an input, not the target. The paper does not invent entities or hide results; the distance errors are reported in the text, which earns credit. Bottom line: this paper deserves a serious referee. The engineering is impressive and reproducible, and the tracking concept is worth testing further. A strong revision would add the ground-truth-distance ablation, more runs, and ideally independent validation in a different phantom. I would accept it for peer review and push the authors on those points, but I would not block it on the fabrication work, which stands on its own.","headline":"A genuine engineering advance in bioelectric catheter fabrication and real-time tracking, but the headline accuracy figures rest on fragile distance estimates and in-phantom-only validation.","tokens_in":661,"tokens_out":821,"would_cite":true,"duration_ms":29190,"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":"A thermally drawn 16-electrode catheter can localize itself continuously in a vascular phantom by fusing electric distance estimates with vessel-geometry impedance features, letting a surgeon and a lay person navigate to a target within 3…","keywords":["Bioelectric navigation","Thermal drawing","Impedance sensing","Catheter localization","Dynamic time warping","Endovascular surgery","Abdominal aortic aneurysm","Proprioceptive navigation"],"falsifier":"Run the same phantom navigation while deliberately alternating wall-pressed and centred catheter paths through the aneurysm, comparing the anchor-based output against an electromagnetic ground truth. If the 3 mm RMS target accuracy appears only when the wall-pressed renal feature is visible, or if centring the catheter introduces a drift larger than the reported tracking error because the distance estimate underestimates travel in wide vessels, the central claim does not survive outside the phantom.","tokens_in":19053,"feed_emoji":"⚡","tokens_out":9176,"duration_ms":91303,"temperature":0.7,"pith_summary":"The paper sets out to prove that a catheter can navigate inside blood vessels using only the electric impedance of its surroundings, with no X-ray imaging. To that end the authors build a custom 6 Fr catheter with 16 electrodes using 3D printing, thermal drawing, and laser micro-machining, and they design a tracking pipeline called devDTW that fuses two electric signals: vessel-geometry features and distance travelled. In a silicone model of an abdominal aortic aneurysm, an endovascular surgeon and a lay person steered the catheter to a simulated stent-graft landing site using only the system's display, reaching an average target accuracy of 3 mm RMS with the anchor-based refinement and 8.3 mm RMS live tracking. If these numbers transfer beyond the phantom, this would be the first demonstration of continuous, real-time, fluoroscopy-free catheter localization from bioelectric sensing alone, reducing radiation and contrast-agent exposure in procedures like stent-graft repair.","feed_headline":"No X-rays: bioelectric catheter tracks itself to 3 mm","feed_subtitle":"Surgeon and lay user steered a phantom catheter to a simulated graft landing using only its electric feedback.","key_machinery":"The load-bearing mechanism is the devDTW pipeline: a three-dimensional dynamic time warping algorithm that matches the live impedance signal, resampled to 1 mm steps using electric distance estimates, against a database of simulated reference signals built from a base centerline plus off-centerline deviations. The key addition is the anchor-region refinement: high-variance stretches of the live signal are matched robustly, and the catheter position is computed from the reference index of the most recent anchors plus the estimated distance moved since passing them. This fusion is what turns a scalar impedance trace into a continuous centerline position while allowing pull-backs and re-advances, because the distance estimates strip non-monotonic motion before matching. The other half of the machinery is the manufacturing pipeline of 3D-printed polycarbonate preforms, wire feeding during the draw, thermal drawing, femtosecond laser micromachining for electrode windows, and laser spot welding.","core_discovery":"The paper's central claim is that bioelectric navigation can be promoted from branch classification to continuous real-time localization by fusing two previously separate electric signals: tetrapolar (four-electrode) impedance features that change with vessel geometry, and electric distance estimates that measure how far the catheter has travelled. On a prototype 6 Fr polycarbonate catheter with 16 electrodes made by thermal drawing, the authors show that the two signals can be resampled onto a common millimetre abscissa and matched against finite-element simulated reference signals using deviation-based dynamic time warping (devDTW), with an anchor-based refinement that locks onto high-variation regions of the signal. In an anatomically realistic phantom, this pipeline tracked the catheter with 8.3 mm RMS error during live navigation and let an endovascular surgeon and a lay user place a simulated stent-graft landing zone within 3 mm RMS of the target using only the system's display. The authors also show that the catheter's mechanical properties are comparable to or better than a commercial 6 Fr catheter in bending and torsion, and that its impedance recordings reproduce the expected inverse relationship between impedance and vessel cross-section.","pith_inferences":["Editorial inference: the anchor-region idea is not specific to bioelectric impedance; any catheter sensor that yields strong local landmarks plus a weak odometry signal could use the same deviation-based matching and anchor propagation to localize itself.","Editorial inference: the paper's own distance-error data point to the main transfer risk, so the natural next test is to fuse distance estimates from several electrode configurations and reject inconsistent estimates before resampling, which could remove the systematic underestimation the authors expect in tissue.","Editorial inference: because renal-branch features appear only when the catheter touches the wall, a practical system may need to deliberately seek wall contact at known landmarks or use radially arranged electrodes, rather than assuming the feature is always visible.","Editorial inference: the spare 250 µm sensor channel could carry a shape sensor, turning the single-point centerline estimate into a curve and resolving the off-centerline position ambiguity that currently creates phantom artifacts."],"forward_implications":["If the 8.3 mm RMS live tracking and 3 mm RMS target accuracy hold, endovascular navigation can be done in real time with no fluoroscopy, removing radiation exposure and the need for nephrotoxic contrast agents in the phantom context.","The back-and-forth handling that surgeons use to reposition devices is supported, because the distance-estimate preprocessing strips non-monotonic motion before matching, removing a key limitation of prior bioelectric branch classification.","The 16-electrode layout with a proximal tetrapolar configuration lets one tracking run address two EVAR landmarks at once, such as the renal-artery ostium and the iliac bifurcation, on a single catheter.","The manufacturing route of 3D-printed preforms plus thermal drawing can produce 21 m of catheter body in one draw, so electrode distributions could be tuned to a patient's anatomy rather than fixed by extrusion dies.","The surgeon's reported workload, with low temporal demand, very low frustration, and lower effort with the anchor-based display, suggests the proprioceptive interface does not add heavy cognitive load."],"supporting_citations":[{"why":"Supplies the original bioelectric navigation concept and the DTW-based branch classification this work extends to continuous tracking.","marker":"[20]"},{"why":"Shows electrode configurations and feature-based localization from bioelectric sensing, informing the catheter's tetrapolar layouts.","marker":"[21]"},{"why":"Provides the electric displacement and direction estimation that the devDTW pipeline uses to resample signals and propagate anchors.","marker":"[22]"},{"why":"Establishes the inverse relation between measured impedance and vessel cross-sectional area, the physical basis for matching features.","marker":"[49]"},{"why":"Defines dynamic time warping, which devDTW generalizes to several deviating reference signals.","marker":"[50]"},{"why":"Supplies the open-begin and open-end matching technique that lets the tracker localize partially traversed vessel trees.","marker":"[52]"}],"fun_headline_variants":["Catheter navigates vessels via electric sense, hits 3 mm in phantom","X-ray-free catheter uses electric signals to track its path accurately","Bioelectric catheter: self-tracking navigation without ionizing radiation","Electric signals guide catheter to 3 mm accuracy, no X-rays needed","Proprioceptive catheter: feels its way through vasculature, 3 mm error"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The navigation result rests on the assumption that the electrically estimated distance the catheter has travelled is accurate enough to reshape the impedance signal and carry the anchor positions forward; the paper reports mean distance errors up to -10.2% and expects systematic underestimation in real tissue, so if that bias is not controlled the 3 mm target accuracy cannot transfer beyond the phantom.","fun_headline_variants_meta":{"raw":{"variants":["Catheter navigates vessels via electric sense, hits 3 mm in phantom","X-ray-free catheter uses electric signals to track its path accurately","Bioelectric catheter: self-tracking navigation without ionizing radiation","Electric signals guide catheter to 3 mm accuracy, no X-rays needed","Proprioceptive catheter: feels its way through vasculature, 3 mm error"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000648,"raw_usage":{"total_tokens":3005,"prompt_tokens":1008,"completion_tokens":1997,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":624,"completion_tokens_details":{"reasoning_tokens":1901}},"tokens_in":624,"tokens_out":1997,"duration_ms":15868,"temperature":1.0,"reasoning_tokens":1901,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T05:55:11.855926+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the same phantom navigation while deliberately alternating wall-pressed and centred catheter paths through the aneurysm, comparing the anchor-based output against an electromagnetic ground truth. If the 3 mm RMS target accuracy appears only when the wall-pressed renal feature is visible, or if centring the catheter introduces a drift larger than the reported tracking error because the distance estimate underestimates travel in wide vessels, the central claim does not survive outside the phantom.","supporting_citations":[{"cited_title":"Biologically inspired catheter for endovascular sensing and navigation","cited_arxiv_id":null,"evidence_quote":"Supplies the original bioelectric navigation concept and the DTW-based branch classification this work extends to continuous tracking."},{"cited_title":"Feature-based electromagnetic tracking registration using bioelectric sensing.IEEE Robotics and Automation Letters, 8(6):3286–3293, 2023","cited_arxiv_id":null,"evidence_quote":"Shows electrode configurations and feature-based localization from bioelectric sensing, informing the catheter's tetrapolar layouts."},{"cited_title":"Extending bioelectric navigation for displace- ment and direction detection","cited_arxiv_id":null,"evidence_quote":"Provides the electric displacement and direction estimation that the devDTW pipeline uses to resample signals and propagate anchors."},{"cited_title":"Accurate nonfluoroscopic guidance and tip location of peripherally inserted central catheters using a conductance guidewire system","cited_arxiv_id":null,"evidence_quote":"Establishes the inverse relation between measured impedance and vessel cross-sectional area, the physical basis for matching features."},{"cited_title":"Dynamic programming algorithm optimization for spoken word recog- nition","cited_arxiv_id":null,"evidence_quote":"Defines dynamic time warping, which devDTW generalizes to several deviating reference signals."},{"cited_title":"Matching incomplete time series with dynamic time warping: an algorithm and an application to post-stroke rehabilitation","cited_arxiv_id":null,"evidence_quote":"Supplies the open-begin and open-end matching technique that lets the tracker localize partially traversed vessel trees."}],"review_version":1}