{"id":"74323ea1-8f06-4e8e-85c3-86147fcf1fc3","arxiv_id":"2607.06244","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":5.5,"correctness_risk":"low","formal_verification":"none","parameter_count":3,"one_line_summary":"Radial beamforming with virtual-source transmits and pair-alternating line acquisition yields SoS images comparable to Cartesian methods while enabling >20 fps duplex imaging on conventional ultrasound systems.","lead":"The paper shows that radial beamforming plus fast pair-alternating transmits can produce pulse-echo speed-of-sound maps on ordinary line-based ultrasound hardware at >20 fps, matching Cartesian-grid quality. This removes a key barrier to real-time clinical SoS imaging on standard scanners.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The Reader correctly isolates residual inter-pair motion and the still-offline transfer as the softest points, yet correctly treats them as residual engineering caveats rather than reasons to reject the claim. The manuscript already supplies the necessary comparative evidence (Figs. 3–7) and an explicit, checkable timing model. My stress-test finds no stronger or more load-bearing concern; therefore the ACCEPT verdict with high confidence stands unchanged.","tokens_in":12667,"tokens_out":404,"duration_ms":5633,"concrete_test":"Re-run the phantom protocol of Section III-A with the exact PAB sequence of Fig. 2 (no 37.5 ms buffer transport) on the same Fukuda scanner, reconstructing SoS from the resulting line-pair displacements; if RMSE and ΔSoS remain within 10 % of the offline Cartesian baseline reported in Fig. 4, the residual-motion and hardware-transfer caveats are empirically closed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's central claim is an engineering feasibility claim: radial R-grid beamforming (especially R9) plus pair-alternating acquisition yields SoS maps comparable to conventional Cartesian software beamforming while fitting line-based on-the-fly hardware and enabling >20 fps duplex timing. Comparative results on k-Wave, CIRS phantoms (RMSE and ΔSoS) and three biopsy-confirmed breast lesions support image-quality parity; the timing argument in Section II-F / Fig. 2 is a transparent arithmetic calculation under stated parallel-beamforming and reconfiguration assumptions. The residual single-cycle motion assumption and the offline PC transfer used in the present experiments are acknowledged limitations of the current implementation rather than hidden contradictions; they do not undermine the demonstrated comparability or the resource-efficiency argument. No load-bearing internal inconsistency or unsupported leap is present.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The manuscript proposes radial (R-grid) beamforming with virtual-source transmits, together with fast pair-alternating beamforming (PAB), for pulse-echo speed-of-sound (SoS) imaging from apparent speckle shifts. The goal is compatibility with conventional line-based, on-the-fly beamformers of mid-range ultrasound systems. Two radial variants (R9-grid with virtual Rx origin behind the array and R0-grid with origin on the surface) are compared against conventional Cartesian-grid (C-grid) software beamforming. Evaluation uses k-Wave simulations of a circular inclusion, CIRS tissue-mimicking phantoms with positive and negative SoS contrast, and three biopsy-confirmed invasive ductal carcinomas. Metrics are RMSE against manufacturer ground truth (phantoms) and ΔSoS contrast (phantoms and in vivo). A sample timing calculation (Fig. 2) argues that the sequences support >20 fps SoS data acquisition interleaved with B-mode under modest parallel-beamforming assumptions.","tokens_in":12904,"tokens_out":1217,"duration_ms":30421,"significance":"If the engineering claim holds, the work is a concrete step toward clinical duplex SoS imaging on standard hardware rather than research-grade full-frame RF systems. Strengths include transparent arithmetic for the duplex timing budget (Section II-F / Fig. 2), explicit ablations of angular line density and addressable Rx channel count (Fig. 5), multi-domain validation (simulation, two-contrast phantoms, three in-vivo lesions), and open acknowledgment that reconstruction and buffer transport remain offline in the present experiments. The contribution is primarily systems-level feasibility rather than a new inverse-problem formulation; the L1-regularized reconstruction (Eq. 2) is taken from prior work. That is appropriate for the stated goal.","major_comments":[{"comment":"Section III-A and the timing argument of Section II-F / Fig. 2: the experimental pipeline still incurs a 37.5 ms buffer-transport overhead to a PC for every Tx event and performs reconstruction offline. The >20 fps duplex claim is therefore an arithmetic extrapolation under stated parallel-beamforming and reconfiguration assumptions, not a closed-loop measurement. The paper correctly uses the language “would allow” and “step towards,” yet the abstract and conclusion should more explicitly separate demonstrated image-quality parity from the still-unrealized on-system real-time pipeline so that readers do not over-read the frame-rate claim.","section":"Section III-A / II-F / Fig. 2"},{"comment":"Equation (2) and all reconstruction figures: the regularization weight λ is never reported, nor is a selection or cross-validation procedure described. Because the same inverse problem is used for every grid comparison, an undocumented λ choice can affect absolute RMSE/ΔSoS values and the visual appearance of inclusions. For reproducibility and for fair comparison across C-/R-grids, λ (or the procedure used to set it) should be stated for each experiment series.","section":"Eq. (2) / Results"}],"minor_comments":[{"comment":"Figure 3 caption and color-bar note: the statement that color ranges were offset by ≈8 m/s to aid visual comparison of contrast should also appear in the main text or a methods subsection so that quantitative readers do not misinterpret absolute SoS levels.","section":"Fig. 3"},{"comment":"Section II-C: the choice of the 9 mm virtual Rx origin depth for the R9-grid is stated as “without loss of generality” matching the Tx focal depth; a short sensitivity remark (or reference to the ablation in Fig. 5) would clarify whether this depth is critical.","section":"Section II-C"},{"comment":"In vivo evaluation uses only three lesions and reports ΔSoS without uncertainty or inter-observer variability on the radiologist masks. For a methods paper this is acceptable, but a sentence noting the limited sample size would temper any clinical-generalization reading.","section":"Section IV-C / Fig. 6"},{"comment":"Notation: σ vs. c and σ₀ vs. c₀ are introduced cleanly, yet the conversion between delay maps Δτ (ns) and the reconstructed SoS maps (m/s) is left implicit; a one-line reminder of the scaling would help readers moving between DMs and SoS images.","section":"Section II-A"},{"comment":"Minor typography: “e.g.,” and “i.e.,” spacing is inconsistent; “R 9-grid” vs. “R9-grid” appears in both forms.","section":"Throughout"}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is a solid systems/engineering contribution that fits an ultrasound or biomedical imaging journal. Novelty is incremental relative to the authors’ own prior VS-sequence and reconstruction papers, but the radial + PAB packaging for line-based hardware is useful and cleanly demonstrated. I see no citation or scope concerns that would require editorial intervention beyond ordinary review."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is a practical engineering paper that closes a real gap. Speckle-shift SoS has been stuck on research systems that can dump full RF frames and do Cartesian software beamforming. Schweizer, Bezek and Goksel show that a radial (R9) grid with virtual-source transmits plus pair-alternating line beamforming produces maps that look essentially the same as the Cartesian baseline, while fitting the constraints of a conventional on-the-fly, line-based front-end.\n\nWhat is new is the combination tailored to those constraints: radial origins placed between the Tx pair so that the tracking windows sit on the natural displacement whiskers, and the fast alternation that reduces inter-pair motion to a single Tx–Rx cycle. The experiments are clean—k-Wave, two-contrast CIRS phantoms with RMSE and ΔSoS, three biopsy-confirmed carcinomas—and the side-by-side figures make the parity claim easy to check. Ablations on angular density and Rx-channel count are included. The reconstruction itself is the same L1-regularized inverse problem they (and others) have used before; the contribution is the acquisition geometry, not a new inverse solver.\n\nSoft spots are minor and openly stated. The >20 fps duplex number is arithmetic under stated parallel-beamforming and reconfiguration assumptions (Fig. 2), not a live closed-loop demo; the present data still went through offline PC transfer because of the Fukuda buffer. Residual single-cycle motion is assumed negligible. None of that undercuts the demonstrated image-quality parity or the resource-efficiency argument. Free parameters (λ, angular density, f-number) are the usual ones and do not appear to be over-tuned.\n\nThis is for people who actually build or deploy quantitative ultrasound on mid-range scanners. It is not a new physics result, but it is the kind of careful systems work that moves a method from “possible on a research box” to “possible on the machines already in clinics.” I would send it to peer review without hesitation; the evidence is there and the claims are proportionate.","headline":"Solid engineering paper that makes SoS imaging practical on ordinary line-based scanners; quality matches Cartesian software beamforming and the >20 fps claim is a clean timing calculation.","tokens_in":13455,"tokens_out":524,"would_cite":true,"duration_ms":7425,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Radial beamforming with fast pair alternation lets standard ultrasound scanners acquire speed-of-sound maps above 20 frames per second while matching Cartesian accuracy on breast lesions.","keywords":["speed-of-sound imaging","radial beamforming","virtual-source transmits","pair-alternating beamforming","speckle tracking","real-time ultrasound","breast lesion imaging"],"falsifier":"Acquire the identical phantom or breast-lesion data on a conventional line-based scanner that implements the proposed radial sequence in hardware; if the reconstructed speed-of-sound contrast or RMSE systematically falls below the Cartesian software baseline reported here, the claim of parity fails.","tokens_in":13603,"feed_emoji":"🔊","tokens_out":882,"duration_ms":11157,"temperature":0.7,"pith_summary":"Standard ultrasound machines form images line-by-line with limited receive channels and cannot easily beamform full Cartesian frames for every transmit, so pulse-echo speed-of-sound imaging has stayed offline. This paper shows that the same machines can instead beamform on a radial grid whose origin sits between a pair of virtual-source transmits, then alternate those transmits line-by-line so that every tracking location is acquired only one pulse-echo cycle apart. The resulting displacement maps feed the same inverse problem used by earlier Cartesian methods and produce comparable speed-of-sound images on simulations, tissue-mimicking phantoms and in-vivo breast carcinomas. Because the sequence re-uses ordinary line-based hardware and can be interleaved with B-mode, dual-mode imaging becomes feasible at more than 20 frames per second. The practical consequence is that quantitative tissue-speed maps, previously confined to research scanners, can run on ordinary clinical systems without sacrificing contrast or accuracy.","feed_headline":"Standard scanners map tissue sound speed at 20+ fps","feed_subtitle":"Radial lines and single-cycle alternation match research-grade accuracy without new hardware","key_machinery":"Fast pair-alternating radial beamforming (R9-grid with Rx origin placed between the two virtual-source foci, lines alternated every Tx–Rx cycle) supplies displacement estimates that are motion-robust down to a single pulse-echo interval and lie fully inside the transmit-overlap region, so the same L1-regularized inverse problem recovers the local speed-of-sound map.","core_discovery":"Radial-grid beamforming of virtual-source transmit pairs, combined with fast pair-alternating acquisition of congruent beamforming lines, yields speed-of-sound reconstructions whose accuracy and lesion contrast match those of conventional Cartesian software beamforming, while remaining implementable on the line-based, limited-channel front-ends of ordinary ultrasound systems and supporting dual-mode frame rates above 20 fps.","pith_inferences":["The method may lower the barrier for multi-center clinical studies of SoS as a breast-cancer biomarker, because data can be collected on the same machines already used for routine B-mode.","Pair-alternating radial acquisition could be reused for other differential-delay modalities (e.g., attenuation or shear-wave tracking) that currently rely on full-frame Cartesian beamforming.","If residual single-cycle motion still dominates error in highly mobile organs, the same sequence could be further accelerated by increasing the parallel-beamforming factor already assumed in the timing budget."],"forward_implications":["Ordinary clinical ultrasound systems can display quantitative speed-of-sound maps interleaved with B-mode at real-time rates without hardware redesign.","Motion sensitivity of pulse-echo SoS imaging is reduced to the physical limit of a single pulse-echo cycle, improving robustness in freehand breast scanning.","The same radial geometry extends naturally to trapezoidal formats used by curved arrays and to pre-beamformed 2-D arrays for volumetric SoS imaging.","Because only tracked displacements (or a few tens of milliseconds of network inference) need leave the scanner, full RF transfer is no longer required."],"fun_headline_variants":["Radial beamforming maps tissue SoS at 20+ fps on standard scanners","Fast alternating radial lines yield real-time SoS without new hardware","Virtual-source radial grids match Cartesian SoS accuracy at over 20 fps","Congruent line alternation enables motion-robust SoS imaging on line-based systems","Radial on-the-fly beamforming supports dual-mode SoS frame rates above 20 fps"],"cache_read_input_tokens":128,"weakest_assumption_plain":"That the residual tissue motion occurring between two successive pulse-echo cycles is small enough not to corrupt clinical speed-of-sound reconstructions, and that a typical mid-range scanner can run the sequence without the offline data transfer still used in the present experiments.","fun_headline_variants_meta":{"raw":{"variants":["Radial beamforming maps tissue SoS at 20+ fps on standard scanners","Fast alternating radial lines yield real-time SoS without new hardware","Virtual-source radial grids match Cartesian SoS accuracy at over 20 fps","Congruent line alternation enables motion-robust SoS imaging on line-based systems","Radial on-the-fly beamforming supports dual-mode SoS frame rates above 20 fps"]},"model":"grok-4.5","effort":"low","cost_usd":0.0049,"raw_usage":{"total_tokens":1364,"prompt_tokens":773,"num_sources_used":0,"completion_tokens":109,"cost_in_usd_ticks":49000000,"prompt_tokens_details":{"text_tokens":773,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":482,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":773,"tokens_out":109,"duration_ms":5342,"temperature":1.0,"reasoning_tokens":482,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-11T01:00:37.634174+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Acquire the identical phantom or breast-lesion data on a conventional line-based scanner that implements the proposed radial sequence in hardware; if the reconstructed speed-of-sound contrast or RMSE systematically falls below the Cartesian software baseline reported here, the claim of parity fails.","supporting_citations":[],"review_version":2}