{"id":"3b43719d-32c8-4985-a529-9b4fbec20e41","arxiv_id":"2409.13477","paper_version":8,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"PnP-CoSMo is a modular plug-and-play iterative reconstruction technique that disentangles content and style in multi-contrast MR images to guide reconstruction from reference scans without k-space training data.","lead":"The paper introduces PnP-CoSMo, a plug-and-play iterative method for multi-contrast MRI reconstruction that learns a content/style model from partially paired image-domain data and replaces aliased content with reference content during reconstruction. This approach could reduce reliance on large paired k-space training datasets, making guided reconstruction more practical in clinical MRI workflows.","discovery_kind":"new_method","skeptic_critique":{"model":"grok-4.3","headline":"Content/style model trained only on image-domain pairs may fail to disentangle factors when input images contain aliasing from k-space undersampling","rationale":"The reader's weakest_assumption correctly isolates the critical modeling assumption. The proposed concrete_test directly probes whether that assumption survives the transition from clean training images to aliased inputs encountered at inference; a large discrepancy would falsify the claim that the PnP replacement plus data consistency yields stable high-quality results without paired k-space training.","tokens_in":1821,"tokens_out":347,"duration_ms":12983,"concrete_test":"Generate synthetic aliasing by retrospectively undersampling the reference images used to train the content/style model at the same acceleration factors as the test set; re-encode these aliased images and measure the L2 distance between their content codes and the codes of the corresponding fully-sampled images. If the distance exceeds the intra-subject variability observed on fully-sampled pairs, the replacement step cannot be assumed to isolate contrast-independent factors.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central construction replaces aliased content with reference content derived from the learned model, then enforces data consistency. This implicitly assumes the content encoder produces the same latent representation for both fully-sampled reference images and aliased reconstructions. Because the model is trained exclusively on (partially) paired image-domain data without exposure to aliasing patterns, the content codes extracted from undersampled iterates may mix contrast-specific aliasing artifacts into the 'content' channel; direct replacement then propagates those errors rather than removing them. The subsequent corrective procedure for the content estimate is described only at high level and does not provably restore consistency with the original k-space measurements.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The manuscript introduces PnP-CoSMo, a modular plug-and-play iterative method for guided multi-contrast MRI reconstruction. It first learns a content/style disentanglement model from partially paired image-domain datasets (no k-space data required), then applies the model in reconstruction by replacing aliased content in the target contrast with high-quality content from a reference scan, followed by data consistency and a corrective step for the content estimate. The approach is claimed to offer cross-contrast generalizability by design and is evaluated on the NYU fastMRI DICOM dataset (equivalent/superior quality to end-to-end methods) plus two in-house multi-coil datasets (up to 32.6% greater acceleration at fixed SSIM versus non-guided reconstruction).","tokens_in":1970,"tokens_out":615,"duration_ms":31718,"significance":"If the central assumption on stable disentanglement holds, the result is significant for reducing reliance on scarce paired k-space training data while retaining an interpretable generative-factor framework; the plug-and-play modularity and explicit separation of shared versus contrast-specific factors distinguish it from black-box end-to-end networks. The reported acceleration gains on in-house data and generalizability claims would be practically relevant if reproducible.","major_comments":[{"comment":"The core assumption that the content encoder yields consistent latent representations for fully-sampled reference images and aliased iterates (trained exclusively on clean image-domain pairs) is load-bearing for the replacement operation and subsequent convergence. No experiment or analysis demonstrates invariance of the content code to undersampling artifacts; the corrective procedure is described only at high level without a convergence guarantee or ablation on aliasing severity.","section":"Method (content replacement and corrective procedure)"},{"comment":"Table reporting the 32.6% acceleration gain on the two in-house multi-coil datasets lacks details on coil sensitivity estimation, data exclusion criteria, number of subjects, or statistical testing; without these, it is impossible to verify whether the gain is robust or driven by dataset-specific factors.","section":"Experiments (in-house dataset results)"},{"comment":"The claim of greater generalizability versus end-to-end methods is supported only by qualitative or aggregate SSIM comparisons on fastMRI; no cross-dataset transfer experiment (e.g., model trained on one contrast pair applied to a third unseen contrast) is shown to isolate the benefit of the disentanglement design.","section":"Experiments (generalizability evaluation)"}],"minor_comments":[{"comment":"Notation for the content and style encoders is introduced without an explicit equation defining the replacement operator; adding a compact equation would improve clarity.","section":"Method"},{"comment":"Figure captions for the simulation experiments on interpretability and convergence do not state the acceleration factor or noise level used, making direct comparison to the in vivo results difficult.","section":"Simulations"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their constructive and detailed comments on our manuscript. We address each major comment point by point below, indicating where revisions will be made.","responses":[{"response":"We agree that explicit validation of content-code invariance to undersampling is important for the replacement step. The manuscript already includes simulations exploring interpretability and convergence of the overall scheme, but these do not directly ablate aliasing severity on the encoder. In the revision we will add a targeted ablation that measures content-code stability across increasing acceleration factors and will expand the description of the corrective procedure with additional analysis of its effect on convergence.","revision_made":"yes","referee_comment":"[Method (content replacement and corrective procedure)] The core assumption that the content encoder yields consistent latent representations for fully-sampled reference images and aliased iterates (trained exclusively on clean image-domain pairs) is load-bearing for the replacement operation and subsequent convergence. No experiment or analysis demonstrates invariance of the content code to undersampling artifacts; the corrective procedure is described only at high level without a convergence guarantee or ablation on aliasing severity."},{"response":"We acknowledge that these experimental details are necessary for reproducibility. The revised manuscript will include the coil-sensitivity estimation method, subject count, exclusion criteria, and any statistical tests performed on the reported acceleration gains.","revision_made":"yes","referee_comment":"[Experiments (in-house dataset results)] Table reporting the 32.6% acceleration gain on the two in-house multi-coil datasets lacks details on coil sensitivity estimation, data exclusion criteria, number of subjects, or statistical testing; without these, it is impossible to verify whether the gain is robust or driven by dataset-specific factors."},{"response":"The cross-contrast generalizability follows directly from the explicit content/style factorization, which isolates shared anatomical content from contrast-specific style and thereby permits the same trained model to be used on any contrast pair without k-space retraining. The fastMRI results already compare performance across multiple contrast combinations. While an additional cross-dataset transfer experiment would provide further isolation of the design benefit, the current aggregate results together with the modular formulation already substantiate the claim; we will clarify this distinction in the discussion.","revision_made":"partial","referee_comment":"[Experiments (generalizability evaluation)] The claim of greater generalizability versus end-to-end methods is supported only by qualitative or aggregate SSIM comparisons on fastMRI; no cross-dataset transfer experiment (e.g., model trained on one contrast pair applied to a third unseen contrast) is shown to isolate the benefit of the disentanglement design."}],"tokens_in":1591,"tokens_out":559,"duration_ms":26169,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The one thing to know is that this paper turns content/style disentanglement into a plug-and-play operator for iterative reconstruction. They train the model on partially paired images, then replace the content code of an aliased estimate with the reference content and enforce data consistency, with a corrective step afterward. This avoids the large paired k-space datasets that end-to-end methods require.","headline":"PnP-CoSMo gives a modular way to guide multi-contrast MRI reconstruction from image-domain content/style models without needing paired k-space training data.","tokens_in":2483,"tokens_out":148,"would_cite":false,"duration_ms":32165,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[],"headline":"MRI content/style disentanglement for PnP reconstruction has no relation to RS distinction-forcing or J-cost machinery","alignment":"orthogonal","rationale":"Paper's core is a two-stage image-domain MUNIT-style model (content encoders Ec_i, style encoders Es_i, decoders G_i) plus content-consistency operator g_M and CR gradient step on AG2(c, ŝ). This operates entirely in the domain of inverse problems and unpaired image translation; it neither invokes nor parallels any RS theorem (reality_from_one_distinction, Jcost uniqueness, phi-ladder, 8-tick periodicity, AlexanderDuality D=3 forcing, etc.). No cosh-cost, ratio symmetry, or parameter-free constant derivation appears.","tokens_in":58075,"confidence":"high","tokens_out":169,"duration_ms":6191,"cache_read_input_tokens":32896,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A content/style model enables plug-and-play guided reconstruction of undersampled MRI by replacing aliased content from a reference contrast scan.","keywords":["multi-contrast MRI","plug-and-play reconstruction","content-style modeling","guided reconstruction","undersampled MRI","image-domain training","generalizable reconstruction","iterative reconstruction"],"falsifier":"If the replacement operation is tested on paired contrasts that share no common anatomical content (for example, brain versus knee scans), the iterative scheme should produce no improvement or outright degradation in SSIM relative to non-guided reconstruction.","tokens_in":2744,"feed_emoji":"🧲","tokens_out":735,"duration_ms":26228,"temperature":0.7,"pith_summary":"The paper establishes a modular reconstruction method that learns a content/style decomposition from partially paired image-domain MR data and then uses it as a plug-and-play operator inside an iterative loop. The decomposition separates contrast-independent anatomical content from contrast-specific appearance, so that reconstruction reduces to swapping the aliased content estimate with high-quality reference content, enforcing data consistency, and applying a corrective step. This design removes the requirement for large paired k-space training sets that end-to-end methods need. On public and in-house multi-coil data the approach reaches higher acceleration factors while matching or exceeding the quality of fully supervised baselines and showing better cross-contrast transfer. The method also supplies an explicit generative-factor account of why one contrast can guide another.","feed_headline":"Content-style replacement guides faster multi-contrast MRI","feed_subtitle":"A modular method swaps aliased content with reference information using image data alone and reaches higher acceleration than non-guided or端","key_machinery":"The content/style model that explicitly separates contrast-independent content from contrast-specific style, turning prior incorporation into a simple replacement operation inside the iterative reconstruction.","core_discovery":"A content/style model of two-contrast MR images is learned from partially paired image-domain data; the disentanglement allows the reconstruction step to become a direct replacement of the aliased content estimate by reference content, which is then combined with an MR data-consistency projection and a content-correction step to produce an iterative scheme that requires no k-space training data and remains generalizable across contrasts.","pith_inferences":["The modular replacement operator could be swapped for other learned priors, allowing hybrid reconstruction pipelines that mix multiple reference scans.","Because the content replacement is explicit, the method could support visual inspection of transferred information for clinical review.","The reduction in training-data requirements suggests the approach may lower barriers to deploying guided reconstruction on new scanner hardware or protocols.","Similar content-style separation might be tested on other multi-modal medical imaging tasks where one modality can serve as a reference for another."],"forward_implications":["Up to 32.6 percent greater acceleration is possible at a fixed SSIM on multi-coil datasets compared with non-guided reconstruction.","Image quality matches or exceeds that of end-to-end trained methods while requiring only image-domain training data.","The same trained model transfers across different contrasts without retraining on raw k-space.","The scheme supplies an explicit account of shared versus non-shared generative factors between the two contrasts.","Convergence behavior and interpretability can be examined directly through the replacement and correction steps."],"fun_headline_variants":["Content-style model plugs into multi-contrast MRI reconstruction","Plug-and-play via content-style disentanglement in MRI","Reference content replaces aliases in guided MRI","Content-style replacement enables modular MRI reconstruction"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The content and style factors learned from partially paired image-domain datasets remain accurately separable even when applied to aliased, undersampled images during reconstruction.","fun_headline_variants_meta":{"raw":{"variants":["Content-style model plugs into multi-contrast MRI reconstruction","Plug-and-play via content-style disentanglement in MRI","Reference content replaces aliases in guided MRI","Content-style replacement enables modular MRI reconstruction"]},"model":"grok-4.3","cost_usd":0.00476,"raw_usage":{"total_tokens":2389,"prompt_tokens":755,"num_sources_used":0,"completion_tokens":54,"cost_in_usd_ticks":47599500,"prompt_tokens_details":{"text_tokens":755,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1580,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":755,"tokens_out":54,"duration_ms":12693,"temperature":1.0,"reasoning_tokens":1580,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-23T20:59:01.716701+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"If the replacement operation is tested on paired contrasts that share no common anatomical content (for example, brain versus knee scans), the iterative scheme should produce no improvement or outright degradation in SSIM relative to non-guided reconstruction.","supporting_citations":[],"review_version":1}