{"id":"1dc3979e-c234-41ee-8a16-bbdd0e3461ae","arxiv_id":"2507.04345","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A structured engineering review of robot-assisted TMS finds that clinical applicability, operational complexity, and cost are the main barriers to adoption.","lead":"This paper reviews robot-assisted transcranial magnetic stimulation (Robo-TMS), covering hardware, calibration and registration, neuronavigation, and control systems. It concludes that broad clinical adoption is blocked by unverified clinical benefit, high operational complexity, and implementation cost.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Table I's cross-system accuracy and force comparison rests on a few unbenchmarked, co-cited sources; the Discussion's field-level claims need a citation audit.","rationale":"The review is a survey, so its load-bearing risk is not an internal derivation but whether its field-level claims are built on representative evidence. The reader's weakest assumption identifies this correctly. I focused on Table I because the same three references support all Accuracy cells, and the values are explicitly system-specific; that combination makes the 'typical' labels and the Discussion's reproducibility and accuracy claims more fragile than the rest of the paper. I did not find a mathematical error in the SE(3) control derivations in Sections III.A and V.A, and the review is transparent about unverified clinical efficacy, so this is not a reject-level flaw. The priority claim, 'first comprehensive review,' would also benefit from a documented search protocol, but the Table I audit is the more concrete, testable vulnerability. Because the reader already assigned CONDITIONAL, my read does not change the verdict.","tokens_in":25701,"tokens_out":10282,"duration_ms":118024,"concrete_test":"Audit Table I cell-by-cell against references [31]-[35]. For each cited source, extract the reported accuracy and force values, experimental setup (phantom, healthy subject, or patient), number of trials, error definition, and tracker model. Rebuild the table as per-study ranges with protocols listed. If any cell cannot be recovered from its citation, or the protocols are non-comparable, revise Table I and soften the Discussion claims that rely on the numerical comparison.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The review's central engineering synthesis is carried by Table I: conventional TMS is quoted at about 6 mm / 3 deg accuracy and 6-10 N contact force, while both robot classes are quoted at about 2 mm / 1.5 deg and 2.5 N. Section VI then uses these numbers to argue that Robo-TMS delivers reproducible, accurate treatment. The evidence base for this comparison is thin: the Accuracy row cites the same three references ([31]-[33]) for all three system types, and the footnote concedes the values come from specific commercial systems and experimental prototypes. No standardized benchmark or measurement protocol, such as phantom versus human, error definition, tracker noise, or trial count, is provided. If those sources used different protocols or small samples, the claimed 2 mm / 1.5 deg and 2.5 N figures are not 'typical' but a convenience summary, and the Discussion's performance advantage is not established. The authors are honest that clinical benefit is unverified, but the engineering performance comparison is the load-bearing premise for that framing.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This review paper surveys robot-assisted transcranial magnetic stimulation (Robo-TMS) from an engineering perspective. It is organised around four subsystems: hardware and integration, calibration and registration, neuronavigation systems (optical tracking and E-field modelling), and control systems (motion compensation, contact force control, and safety). The authors argue that clinical adoption of Robo-TMS is limited by unverified clinical applicability, high operational complexity, and substantial implementation costs, and they propose future directions including marker-less tracking, non-rigid registration, learning-based E-field modelling, individualised MRI generation, and robot-assisted multi-locus TMS. The paper explicitly claims to be the first comprehensive engineering-focused review of Robo-TMS.","tokens_in":25860,"tokens_out":5151,"duration_ms":57803,"significance":"The manuscript fills a genuine gap: prior reviews have focused on TMS fundamentals, clinical applications, or specific components, while an integrated engineering perspective on Robo-TMS has been lacking. The paper's value lies in its systematic organisation, the explicit mathematical formulations of calibration and control (Eqs. 1-5), and a concrete research agenda. It is also honest in acknowledging that clinical benefit remains unverified. However, the quantitative performance comparison in Table I, which underpins the narrative that Robo-TMS improves accuracy and contact force over conventional TMS, is not robustly sourced; if the reported values are only illustrative, the strength of the discussion claims must be tempered. With a revised and properly contextualised comparison, the review would be a useful reference for both engineers and clinicians.","major_comments":[{"comment":"The Accuracy and Contact Force rows of Table I cite the same three references ([31]-[33]) for conventional TMS and for both Robo-TMS classes, and the table footnote concedes that the values come from specific commercial systems and experimental prototypes. Because [31]-[33] are Robo-TMS system papers rather than comparative studies, and because no measurement protocol (phantom versus human, error definition, tracker noise, trial count) is given, the table cannot support the 'typical' performance characterization. This matters for the paper's central claim: Section VI-A1 uses these numbers to argue that Robo-TMS delivers reproducible and accurate treatment, and the 'typical' framing is a load-bearing premise for that argument. Please either present the values as reported ranges for individual systems with explicit protocol details, or replace the table with a source-by-source comparison that separates conventional, industrial, and specialised systems.","section":"Table I and Section VI-A1"},{"comment":"The paper identifies accessibility as 'the foremost barrier' to clinical adoption, but the review does not report a method (e.g., thematic analysis, stakeholder survey, frequency of barrier mentions in the literature, or cost analysis) that would justify a ranking among the three named barriers. As written, the ranking is an authorial judgment rather than a demonstrated synthesis outcome. Please either soften the claim to 'a primary barrier' or briefly document the basis for the ordering.","section":"Section VI-B"}],"minor_comments":[{"comment":"The corresponding author email contains a typo: 'zhenhong.li@anchester.ac.uk' should presumably be 'zhenhong.li@manchester.ac.uk'.","section":"Section I (footnote)"},{"comment":"The citation for SlicerTMS is incomplete; it lacks a venue, volume/pages, or DOI, which makes it difficult for readers to locate the work.","section":"Reference [139]"},{"comment":"The sentence 'The accuracy of Robo-TMS significantly reduces the risk of side effects' is stronger than the cited evidence supports; references [6] and [16] address the safety of navigated TMS in brain tumour mapping, not a comparative Robo-TMS-versus-conventional side-effect rate. Please soften to reflect that the evidence is indirect.","section":"Section VI-A3"},{"comment":"The claim of being 'the first review to comprehensively survey the status and challenges of Robo-TMS from an engineering perspective' would be more verifiable if the authors briefly described their literature search strategy (databases, years, inclusion criteria) used to identify prior reviews.","section":"Section I"},{"comment":"The depth-focality metrics d_{1/2} and S_{1/2} are defined in the caption of Fig. 3; consider defining them in the main text before the figure reference, since they are used again later.","section":"Section II-B4"}],"recommendation":"major_revision","confidential_remarks":"The priority claim in the Introduction should be verified by the editor against the existing literature. The manuscript's technical synthesis is sound in its qualitative parts, but the Table I evidence base is a load-bearing weakness that should be addressed before acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is a competent, well-organized engineering review of robot-assisted TMS, and it is the first I know of that treats the integrated system—hardware, calibration, navigation, control—as a single engineering problem. If someone asks you for a starting point on Robo-TMS, this is a good one to send.\n\nWhat it does well: the four-part structure is sensible; the calibration and registration sections are technically accurate; the control-system math (standard SE(3) homogeneous transformations) is correctly presented. The authors are also honest about clinical uncertainty—they state plainly that clinical benefit over conventional TMS is unverified and that accessibility is the main adoption barrier. The future-directions section is concrete, and the reference list is broad and mostly primary sources.\n\nWhere it is soft: Table I compares conventional, industrial-robot, and specialized-robot systems on accuracy and contact force, but the accuracy rows all cite the same three references ([31]–[33]) and the force rows rest on just [34] and [32], [35]. The footnote concedes these are specific commercial systems and experimental prototypes, so the table is an illustration, not a systematic benchmark. The Discussion leans on \"reproducible and accurate\" framing without restating that caveat. To me this is a minor flaw, not a serious one: the paper's main claims—that adoption is limited by unverified clinical benefit, complexity, and cost—do not depend on the exact numbers being generalizable. The \"first engineering review\" claim is also asserted without a documented search strategy, so treat it as a plausible priority claim rather than a proven one. The abstract and introduction could easily add a sentence describing how the literature was gathered.\n\nI checked the derivations and the cited error values; they look faithful to source. The paper does not overreach, and the authors explicitly flag what is unknown.\n\nBottom line: worth sending to peer review. If I were the referee, I would request minor revisions—move the Table I caveat into the table title or repeat it in the Discussion, and add one sentence on literature selection. The rest is solid.\n\nRecommendation: engage with this, and cite it if you are working in this area.","headline":"A solid first-engineering-synthesis of Robo-TMS that is worth reading and citing, provided the reader treats its performance comparison table as illustrative rather than benchmarked.","tokens_in":26391,"tokens_out":2695,"would_cite":true,"duration_ms":30400,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This review organizes robot-assisted TMS into four subsystems and identifies accessibility, not engineering capability, as the main barrier to clinical adoption.","keywords":["transcranial magnetic stimulation","robot-assisted TMS","neuronavigation","calibration and registration","optical tracking","electric field modelling","multi-locus TMS","clinical adoption"],"falsifier":"A public benchmark that measures targeting accuracy, contact force, and setup time of several Robo-TMS systems on the same phantom would falsify the review's representative performance profile if results vary widely across systems; similarly, a randomized trial showing Robo-TMS clinical superiority over manual navigated TMS would undercut the 'unverified clinical applicability' claim.","tokens_in":25504,"feed_emoji":"🤖","tokens_out":3359,"duration_ms":37718,"temperature":0.7,"pith_summary":"The paper claims to be the first comprehensive engineering survey of robot-assisted transcranial magnetic stimulation (Robo-TMS), a technology that mounts a TMS coil on a robot arm to keep stimulation precisely on target during long sessions. It maps the field into four subsystems: hardware and integration, calibration and registration, neuronavigation systems, and control systems. The central finding is that broader clinical adoption is limited by accessibility, driven by unverified clinical applicability, high operational complexity, and substantial implementation costs. The review then points to emerging engineering tools, such as marker-less tracking, non-rigid registration, learning-based electric field modelling, individualised MRI generation, and robotic multi-locus TMS, as the routes to overcome these barriers.","feed_headline":"Robot TMS works, but cost and complexity block clinics","feed_subtitle":"First engineering survey maps four subsystems and says accessibility, not accuracy, decides clinical uptake.","key_machinery":"The organizing device is the coordinate-transformation chain expressed as homogeneous matrices in $SE(3)$, which relates the head, MRI, camera, robot base, end-effector, and coil frames. The clinical targeting requirement $^\\text{head}T^*_{\\text{coil}} = {}^{\\text{head}}T_{\\text{MRI}} \\cdot {}^{\\text{MRI}}T^*_{\\text{coil}}$ is realized in real time through the tracked pose chain involving the optical camera, and the paper uses this framework to locate error sources in calibration and registration and to compare typical system performance, around $2\\,\\text{mm}/1.5^\\circ$ accuracy and roughly $2.5\\,\\text{N}$ contact force.","core_discovery":"The paper establishes a four-part engineering taxonomy for Robo-TMS and argues that the field's central technical challenge is not raw accuracy but the smooth, low-cost integration of calibration, registration, neuronavigation, and control into a clinically usable workflow. Its load-bearing diagnostic claim is that accessibility is the foremost barrier to clinical adoption, more decisive than any single accuracy metric. The review further asserts that the clinical value of Robo-TMS over conventional TMS remains unverified, so the engineering community lacks evidence-based guidelines for when and how the technology should be used.","pith_inferences":["The typical accuracy and contact-force figures in the review's comparison table come from a small set of specific commercial systems and experimental prototypes without a standardized benchmark, so the comparative performance profile could shift under a common evaluation protocol.","The review implies, but does not test, that Robo-TMS offers no proven clinical benefit over conventional navigated TMS; a randomized head-to-head trial would be the decisive test of that assumption.","The expectation that broader adoption will lower costs through economies of scale is speculative until clinical demand is demonstrated.","Combined advances in automated registration and learning-based E-field solvers could eventually make fully unattended TMS sessions feasible, a step the paper only hints at."],"forward_implications":["If the four-subsystem taxonomy is correct, engineering effort should concentrate on the least automated steps, namely calibration and registration, which dominate operational complexity.","If accessibility is the true adoption barrier, reducing cost and workflow burden matters more than further improving targeting accuracy.","If marker-less tracking and automated registration mature, setup time and operator dependence will drop, easing clinical translation.","If learning-based E-field modelling delivers real-time field estimates, closed-loop robotic targeting with online dose visualization becomes feasible.","If Robo-mTMS integration succeeds, stimulation can be re-targeted electronically within small cortical regions, reducing the need for slow robot-arm motion during multi-target sessions."],"supporting_citations":[{"why":"Foundational book-length treatment of robotized TMS that supplies the early system configurations and force-control concepts the review builds on.","marker":"[29]"},{"why":"Provides the calibration methodology and control approach whose reported accuracy and contact-force values anchor the review's typical performance figures.","marker":"[32]"},{"why":"Quantifies accuracy and precision of landmark-based versus surface-based registration, the central error-source analysis for neuronavigation.","marker":"[79]"},{"why":"Safety and recommendation guidelines that justify neuronavigation use and define hardware/software safety constraints for Robo-TMS.","marker":"[6]"},{"why":"Introduces the fully automated setup procedure combining AutoHS hotspot localization with Robo-TMS, the basis for reducing manual intervention.","marker":"[90]"},{"why":"Reviews learning-based real-time E-field modelling algorithms that the paper cites as a promising path to real-time performance.","marker":"[23]"},{"why":"Establishes multi-locus TMS theory, which underlies the Robo-mTMS integration proposed as a workflow-streamlining future direction.","marker":"[54]"},{"why":"Demonstrates a robotic-electronic platform for autonomous TMS targeting, supporting the Robo-mTMS outlook.","marker":"[56]"}],"fun_headline_variants":["Robo-TMS review: precision isn't the issue, integration is","Engineering survey: cost and complexity block Robo-TMS clinics","Robot TMS: accessibility, not accuracy, gates clinical adoption","Review maps Robo-TMS subsystems, says usability decides uptake"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The review's barrier diagnosis assumes that the handful of commercial systems and research prototypes it cites are representative of Robo-TMS as a whole, since no standardized benchmark or systematic head-to-head comparison backs the typical accuracy and contact-force figures.","fun_headline_variants_meta":{"raw":{"variants":["Robo-TMS review: precision isn't the issue, integration is","Engineering survey: cost and complexity block Robo-TMS clinics","Robot TMS: accessibility, not accuracy, gates clinical adoption","Review maps Robo-TMS subsystems, says usability decides uptake"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000244,"raw_usage":{"total_tokens":1483,"prompt_tokens":849,"completion_tokens":634,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":465,"completion_tokens_details":{"reasoning_tokens":561}},"tokens_in":465,"tokens_out":634,"duration_ms":6772,"temperature":1.0,"reasoning_tokens":561,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T19:50:22.896509+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A public benchmark that measures targeting accuracy, contact force, and setup time of several Robo-TMS systems on the same phantom would falsify the review's representative performance profile if results vary widely across systems; similarly, a randomized trial showing Robo-TMS clinical superiority over manual navigated TMS would undercut the 'unverified clinical applicability' claim.","supporting_citations":[],"review_version":1}