Abstract

The intrinsic motion of the human uterus is key both for healthy physiology and for the etiology of diseases such as endometriosis and adenomyosis. However, non-invasive characterization remains a major challenge due to the location of the uterus, low signal-to-noise and overlapping motion from adjacent organs. Electrohysterography (EHG) is a method for monitoring uterine contractility using a surface electrode array. Previous studies that combined EHG and Magnetic Resonance Imaging (MRI) to analyze uterine peristalsis assumed an empty bladder, neglecting the influence of bladder filling. The overlap of uterine and bladder frequency ranges results however in frequency filtering alone being insufficient for uterine signal isolation. We introduce a novel multi-cluster beamforming framework based on simultaneous EHG and MRI to separate uterine and bladder electrical activity. Subject-specific 3D segmentations of the uterus, bladder, and electrodes enabled the generation of volumetric lead-field dictionaries, cluster-geometry-based DOA vectors, and spatially coherent organ-specific clusters. These clusters defined uterus-pass and bladder-null constraints in a multicluster linearly constrained minimum-variance (LCMV) beamformer with Bayesian covariance shrinkage, followed by overlap-add reconstruction of organ-resolved time series. Paired T1-weighted 3D whole abdomen MRI and multi-channel EHG data from 10 volunteers (7 females and 3 males), including longitudinal data from 3 women across menstrual phases was acquired. Raw EHG signals showed activity in the [5,50]mHz}band; in men only bladder components [25,41]mHz} were present, while female datasets exhibited uterine and bladder contributions. Across female volunteers, cluster compactness strongly predicted beamformer stability (Pearson r=0.91. Leakage subtraction consistently improved bladder isolation. The proposed anatomy-guided multicluster LCMV enables robust uterus–bladder signal separation in low-channel EHG.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026/paper/5794_paper.pdf

SharedIt Link: Not yet available

SpringerLink (DOI): Not yet available

Supplementary Material: https://papers.miccai.org/miccai-2026/supp/5794_supp.zip

Link to the Code Repository

https://github.com/CamilaBustos258/mcLCMV_EHG_MRI.git

Link to the Dataset(s)

N/A

BibTex

@InProceedings{BusMar_Disentangling_MICCAI2026,
        author = { Bustos-Vivas, Maria Camila AND Tripathy, Smiti AND Aviles Verdera, Jordina AND Alves de Castro, Nyvenn AND Hutter, Jana},
        title = { { Disentangling Uterine and Bladder Activity in Paired EHG-MRI Data Using Anatomy-Guided Multicluster Beamforming } },
        booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026},
        year = {2026},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 16896},
        month = {September},
        page = {pending}
}


Reviews

Review #1

  • Please describe the contribution of the paper

    This paper describes a multicluster beamforming framework developed for the specific task of separating uterine and bladder electrical activity in paired electrohysterography and MRI data, providing a pathway for non-invasive evaluation of uterine muscular activity. This study targets an interesting, clinically relevant task with promising preliminary data presented in a cohort of 10 healthy volunteers.

  • Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    1.A core strength of this paper is the specificity of the task and the success of the tailored approach to a challenging problem, given the heterogeneity and diversity in the population studied. 2.This paper also includes a comprehensive evaluation with good statistical analysis and clear discussion of the the actual implications of the findings in the clinical environment, acknowledging the present limitations of the study. 3.The paper is well-written and logical with a clear motivation and story presented. 4.The composition of the dataset, while small, shows thoughtfulness in the inclusion of male participants for a “control” comparison without a uterus and the assessment of different menstrual phases in a small portion of the cohort.

  • Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.

    1.A primary weakness of the paper is the size of the human dataset which contains only 10 participants of which only 7 had a uterus and only 3 were evaluated longitudinal to consider the effect of menstrual phase, as acknowledged in the Discussion. 2.There is no mention in this paper of reproducibility factors, such as source code or dataset access.

  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission does not mention open access to source code or data, but provides a clear and detailed description of the algorithm to ensure reproducibility.

  • Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?

    N/A

  • Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html

    The most prominent comments are featured above. A few minor additional comments:

    1.Because your “Results” section also features the description of the experimental dataset, it may be best to re-name this section along the lines of Experiments and Results. 2.In the Results, there is a comment about “repeated sessions from the same volunteer aligned” but it would be good to clarify whether this was repeated at the same timepoint or referring to the 3 longitudinal subjects.

  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.

    (5) Accept — should be accepted, independent of rebuttal

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    Overall, this study is well-done, interesting, and includes a thorough evaluation to a specific problem in an underrepresented domain. The only caveat is that while the specific task leads to a very focused study, the proposed methodologies appear to have somewhat limited generalizability to other tasks which may limit the interest within the MICCAI community.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.

    N/A

  • [Post rebuttal] Please justify your final decision from above.

    N/A



Review #2

  • Please describe the contribution of the paper

    This paper studies a meaningful and underexplored problem: separating uterine and bladder activity in simultaneous low-channel EHG-MRI acquisitions. The authors propose an anatomy-guided multicluster LCMV beamforming pipeline that uses subject-specific MRI segmentations to build uterus/bladder steering dictionaries, applies organ-specific pass/null constraints with covariance shrinkage, and reconstructs organ-resolved time series with leakage subtraction. The study includes 10 paired EHG-MRI datasets, including female and male volunteers and some longitudinal menstrual-phase acquisitions. The paper’s main claimed finding is that uterine cluster compactness, more than gross uterus-bladder separation angle, predicts beamformer stability.

  • Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    The paper targets uterus–bladder source disentanglement in paired EHG–MRI, a niche but meaningful problem where simple spectral filtering is insufficient because the signals overlap in frequency.

    The anatomy-guided multicluster LCMV formulation is well motivated: subject-specific MRI geometry is used to define clustered uterus/bladder steering directions and organ-specific pass/null constraints, which is a sensible way to inject anatomy into a difficult inverse problem.

    The method includes clustered steering, constrained beamforming, and leakage subtraction in a logically consistent framework, with subject-level diagnostics showing strong target response and cross-organ suppression.

    A notable finding is that uterine cluster compactness, rather than gross uterus–bladder separation angle, appears to be the main driver of pass-direction stability, which could be useful for future sensor-placement and modeling design.

    The paper provides cohort-level and subject-level analyses, including PSD trends, compactness/stability correlations, leakage metrics, and dynamic examples, making the proposed mechanism relatively interpretable.

    The paper demonstrates stable constrained beamforming with only 8 electrodes is practically appealing and supports the claim that anatomical priors can compensate for limited sensors.

  • Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.

    The paper lacks convincing comparisons with meaningful baselines, such as simpler frequency-based separation, standard beamforming without multicluster constraints, no-null versions, or no-leakage-subtraction versions. This makes it hard to judge the true added value of each component.

    Most reported evidence is internal or model-based, such as pass/null gains, residuals, correlations, and energy ratios. These support engineering consistency, but they do not fully verify that the recovered signals are truly organ-specific physiological sources.

    The dataset contains only 10 paired recordings, with 7 female and 3 male volunteers and only 3 longitudinal acquisitions. All appear to be healthy participants, which limits the strength of the biological and clinical claims.

    The paper explains the framework reasonably well, but several practical details appear insufficiently specified for full reproduction, including some clustering/regularization choices and preprocessing specifics.

    Some claims may be overstated. Very strong suppression numbers and leakage improvements are reported, but without independent physiological validation, they should be interpreted more cautiously as method-internal evidence rather than definitive organ-level separation.

  • Please rate the clarity and organization of this paper

    Satisfactory

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission does not provide sufficient information for reproducibility.

  • Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?

    N/A.

  • Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html

    N/A.

  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.

    (4) Weak Accept — marginally above the acceptance threshold, but would not mind if rejected, dependent on rebuttal

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    The manuscript presents promising separation performance on 10 paired EHG–MRI datasets, offers the valuable insight that uterine cluster compactness may be a stronger predictor of beamformer stability than the uterus–bladder separation angle, and demonstrates practical feasibility even with an 8-channel setup. While the study is still subject to several important constraints—including a relatively small cohort, data collected only from healthy participants, menstrual-phase timing based on opportunistic acquisition, and the lack of independent physiological validation—these factors primarily point to meaningful opportunities for future validation and expansion rather than detracting from the significance of the current findings.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.

    N/A

  • [Post rebuttal] Please justify your final decision from above.

    N/A



Review #3

  • Please describe the contribution of the paper

    The paper’s main contribution is an anatomy-guided signal separation framework for disentangling uterine and bladder activity from paired EHG-MRI recordings. The authors use subject-specific MRI segmentations of the uterus, bladder, and electrode positions to build organ-specific spatial models, and then use these models to guide a multicluster LCMV beamformer. The goal is to separate uterine activity from bladder-related activity, which is difficult because both sources can overlap in the same low-frequency EHG range.

    A key technical contribution is that the method does not rely only on frequency filtering. Instead, it uses anatomical information to define uterus-pass and bladder-null constraints, followed by overlap-add reconstruction and a leakage subtraction step to reduce residual cross-organ contamination.

    Also the paper also provides an initial in-vivo feasibility study using paired abdominal EHG and T1-weighted MRI data from 10 volunteers, including 7 women and 3 men.

  • Please list the major strengths of the paper: you should highlight a novel formulation, an original way to use data, demonstration of clinical feasibility, a novel application, a particularly strong evaluation, or anything else that is a strong aspect of this work. Please provide details, for instance, if a method is novel, explain what aspect is novel and why this is interesting.

    the clinical problem is well motivated and relatively underexplored. Separating uterine activity from bladder-related activity in non-invasive EHG is important because both organs can contribute low-frequency signals, and simple frequency filtering may not be sufficient.

    the use of subject specific MRI anatomy is a strong and interesting aspect of the work. Instead of treating EHG as only a multichannel surface signal, the authors use MRI-derived segmentations of the uterus, bladder, and electrodes to guide the source separation. to me this is a nice formulation because the uterus and bladder are deep, spatially close organs, so anatomical priors are important for good separation.

    next the authors do not represent each organ as a single source, but instead define multiple spatial clusters for the uterus and bladder.

    overall framework is coherent and shows how anatomical imaging and surface electrophysiology can be combined in practice.

    the work provides an initial in-vivo feasibility demonstration. Although the cohort is small, the authors include 10 paired EHG-MRI datasets, including both female and male volunteers, and show that the method can be applied with only 8 EHG channels.

  • Please list the major weaknesses of the paper. Please provide details: for instance, if you state that a formulation, way of using data, demonstration of clinical feasibility, or application is not novel, then you must provide specific references to prior work.

    i do have a few minor concerns;

    for example the validation of true uterus–bladder separation is still indirect. The authors show that the beamformer satisfies pass/null constraints and that leakage subtraction reduces correlation between reconstructed signals, but these are mainly internal technical checks. There is no independent physiological reference, such as bladder filling, urodynamics, uterine pressure, cine MRI motion, or expert-labeled uterine activity, to confirm that the separated signals truly come from the intended organs. This is important because the paper claims robust in-vivo uterus–bladder separation.

    some key implementation details are missing or unclear. The paper describes MRI-based steering dictionaries, clustering, LCMV beamforming, covariance shrinkage, and overlap-add reconstruction, but does not clearly specify details such as the number of clusters, clustering method, window length, overlap, preprocessing steps, shrinkage selection, and the exact bladder-pass/uterus-null formulation. These details are important for reproducibility. But i think this limitations can be easily resolved by making the code public.

    simple baselines or ablations are missing. As the main claim is that anatomy-guided multicluster beamforming is needed, it should be compared with simpler alternatives such as frequency filtering alone or single-cluster beamforming lcmv without null constraints

  • Please rate the clarity and organization of this paper

    Good

  • Please comment on the reproducibility of the paper. Please be aware that providing code and data is a plus, but not a requirement for acceptance.

    The submission does not provide sufficient information for reproducibility.

  • Based on your review and your understanding of the MICCAI Scientific Code of Ethics, do you believe this submission may involve a potential ethics concern or violation?

    N/A

  • Optional: If you have any additional comments to share with the authors, please provide them here. Please also refer to our Reviewer’s guide on what makes a good review and pay specific attention to the different assessment criteria for the different paper categories: https://conferences.miccai.org/2026/en/REVIEWER-GUIDELINES.html

    How were the uterus, bladder, and electrode segmentations obtained from MRI, and was segmentation variability assessed?

    Were electrode positions assumed to be fixed throughout the EHG-MRI acquisition, or could motion, abdominal deformation affect the electrode-to-organ geometry?

    How computationally expensive is the full pipeline, including MRI segmentation, steering dictionary construction, and beamforming? Is near-real-time or same-session analysis feasible?

  • Rate the paper on a scale of 1-6, 6 being the strongest (6-4: accept; 3-1: reject). Please use the entire range of the distribution. Spreading the score helps create a distribution for decision-making.

    (4) Weak Accept — marginally above the acceptance threshold, but would not mind if rejected, dependent on rebuttal

  • Please justify your recommendation. What were the major factors that led you to your overall score for this paper?

    The paper presents a novel and clinically relevant anatomy-guided beamforming approach for separating uterine and bladder activity from paired EHG-MRI data. although the validation is still indirect and some implementation details need clarification, the idea is original, technically thoughtful, and promising as a feasibility study.

  • Reviewer confidence

    Confident but not absolutely certain (3)

  • [Post rebuttal] After reading the authors’ rebuttal, please state your final opinion of the paper.

    N/A

  • [Post rebuttal] Please justify your final decision from above.

    N/A



Author Feedback

We thank all reviewers and the meta-reviewer for their constructive feedback and recognition of the work’s novelty and clinical relevance. Below we address the main concerns grouped by theme. P1 — Dataset size (R1, R2, R4): While the cohort consists of 10 volunteers, the 3 longitudinal participants each contributed 2 sessions (follicular/luteal), resulting in 13 recordings total. Simultaneous EHG-MRI acquisition is resource-intensive, requiring MRI facility access, custom synchronization, and extended patient preparation. Data collection is ongoing. P2 — Physiological validation (R2, R4): We acknowledge that independent physiological validation was not included in the current study and agree this is an important limitation. We clarify that the ~−155/−158 dB null suppression values quantify mathematical constraint satisfaction — how accurately the LCMV enforces pass/null directions — not physiological organ-level purity. The practically relevant cross-organ suppression is the null gain (0.10–0.48, Table 1). Similarly, Δr and energy ratios are engineering consistency metrics confirming the mathematical operations work as intended, and should not be interpreted as definitive evidence of organ-level separation. We will make this distinction more explicit in the camera-ready. The simultaneously acquired EPI dynamic MRI contains bladder filling information over time, which we plan to incorporate as an independent physiological reference in future work. P3 — Baselines and ablations (R2, R4): We agree explicit comparisons would strengthen the evaluation. While the Methods establish theoretically why frequency filtering alone is insufficient, we acknowledge this does not substitute for empirical comparison. We note that the leakage subtraction component is partially evaluated through the Δr and energy ratio columns in Table 1, which quantify its effect across all sessions. A full ablation against frequency filtering and single-cluster LCMV without null constraints remains an important direction for future work. P4 — Implementation details and reproducibility (R1, R2, R4): The full code will be released publicly upon publication. The camera-ready will clarify missing parameters (number of clusters, window length/overlap, shrinkage β selection). Participant data will be shared in accordance with institutional and ethical guidelines. We confirm the study received appropriate institutional ethical approval and all participants provided informed consent. P5 — Additional questions (R4): Segmentations were obtained manually from T1-weighted MRI; variability across sessions/operators was not assessed but is recognized as important for future work. Electrode positions are registered within the same simultaneous session, ensuring geometric consistency; soft-tissue deformation was not modeled and will be investigated in future work. Manual segmentation takes ~3 hours per subject (uterus being most challenging); full pipeline timing will be reported in the camera-ready. Near-real-time processing is not feasible due to prior MRI segmentation requirements, but same-session beamforming is achievable once the anatomical model is built. P6 — Minor clarifications (R1): We will rename “Results” to “Experiments and Results.” “Repeated sessions from the same volunteer” refers to the 3 longitudinal subjects scanned across menstrual phases, not same-day repeats. We thank the reviewers and meta-reviewer once more for their valuable input. The camera-ready will incorporate clarified implementation details and code release.




Meta-Review

Meta-review #1

  • Your recommendation

    Provisional Accept

  • Please justify your decision. In case you deviate from the reviewers’ recommendations, explain in detail the reasons why. In case of an invitation for rebuttal, clarify which points are important to address in the rebuttal.

    All reviewers are in favor of accepting this work, citing strong technical novelty, clinical application, and method validation. This is a strong submission, and I do not see any reason to contest the reviewers’ consensus.



back to top