Pith. sign in

REVIEW 4 major objections 6 minor 21 references

MReg: A Novel Regression Model with MoE-based Video Feature Mining for Mitral Regurgitation Diagnosis

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read MReg reports 89.36 percent accuracy in grading mitral regurgitation from color Doppler video, using regression and MoE feature mining to outperform classification baselines.

desk verdict Sensible architecture and clear clinical motivation, but the reported F1 scores are arithmetically impossible given the paper's own Precision/Recall values, so the headline SOTA claim is currently unsupported. read the letter →

arxiv 2506.23648 v1 pith:3JILYH7Q submitted 2025-06-30 cs.CV

classification cs.CV
keywords mitralregurgitationcolorDopplerechocardiographyvideoregressionmixture-of-expertsfeatureselectionamplificationmultipleinstancelearningmedicalanalysis
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper introduces MReg, an automated model that grades mitral regurgitation (MR) from four-chamber color Doppler echocardiography videos into normal, mild, and moderate-to-severe categories. The authors argue that MR grading should be treated as a regression problem rather than classification, because severity is a continuous, ordinal spectrum. To capture the relevant visual evidence, MReg selects the cardiac-cycle instance most likely to contain regurgitation, amplifies its features by combining the original video features with a convolution applied to their L2 norm, and uses a Mixture-of-Experts module whose three experts each summarize features for one severity grade. On an in-house dataset of 1,868 videos, the model reports Accuracy 89.36%, Recall 85.93%, Precision 86.83%, Specificity 86.36%, and F1-score 94.28%, outperforming the compared supervised video classifiers and weakly supervised anomaly detectors.

What carries the argument

The central object is the two-stage X-CLIP-based pipeline. Stage I performs binary normal-versus-MR classification with focal loss and chooses the instance index $\alpha = \arg\max(f_{\text{2cls\_out}}[:, 1])$ with the highest MR probability. Stage II computes an amplified video feature $f_{\text{amp}} = N(f_{\text{video}} + \beta \times \text{Conv}(\|f_{\text{video}}\|_2))$ and feeds it, together with three-category text features, into a MoE module with one expert per grade; each expert's weight is $w_{E_i} = \text{SoftMax}(\|W_i\|_F)$, and the final regression output is mapped to three grades by thresholds (0.5 and 1.5). The total loss combines binary focal loss, MSE regression loss, an expert focal loss, and smoothing/sparsity regularizers.

What would settle it

Grade a held-out set of 200 videos by three independent cardiologists with consensus adjudication and compare MReg's predictions against the consensus grades, or against quantitative Doppler measurements such as effective regurgitant orifice area; if MReg's accuracy and F1 do not remain above the compared baselines on this reference labeling, the central claim is falsified.

Watch

Extended reading notes

Core claim

The central claim is that combining three feature mining strategies—instance selection via multiple instance learning, feature amplification, and MoE-based feature summary—within a regression formulation yields a clinically aligned and accurate MR diagnosis system. Specifically, the model first uses X-CLIP video and text features, then selects the instance with the highest MR probability to focus on the most severe regurgitation, amplifies regurgitation-related signals via a residual convolution on the L2-normalized feature map, and finally sums category-specific expert features to improve grading of complex cases. The regression loss with thresholds preserves ordinal continuity between grades. The paper's experiments show that each added module improves recall, especially for the underrepresented moderate-to-severe grade, and that the full model outperforms all compared SVC and WS-VAD methods on every reported metric.

Load-bearing premise

The evaluation assumes that the grade assigned by a single experienced sonographer is the correct diagnosis, with no second reader, no measurement of reader variability, and no external quantitative reference; if these labels are noisy or subjective, the reported accuracy and F1 are not a true measure of diagnostic performance.

Editorial extensions

If this is right

  • If MReg's results hold, automatic MR screening can run on standard A4C color Doppler videos without manual keyframe selection, since the MIL-based instance picker replaces the sonographer's search for the worst jet.
  • The regression formulation with thresholds (0.5 and 1.5) implies that the same network can be recalibrated for different clinical cutoffs without retraining, because the output is a continuous severity score rather than a discrete class.
  • Feature amplification increases the gap between decision frames and other frames in an instance, making the model's decision frame dominant and improving interpretability of which part of the cardiac cycle drove the prediction.
  • The MoE feature summary assigns each expert to a severity grade; this category-level decoupling is what handles eccentric and multi-jet regurgitation, where jet area alone is ambiguous.
  • Equipping the model with $L_{\text{expert}}$, a focal loss on the unmixed expert outputs, resolves the performance drop that occurs when adding MoE without supervision, bringing the full model to its best metrics.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The single-sonographer labels are treated as ground truth with no inter-observer variability reported; if label noise varies by grade, the reported confusion-matrix improvements for Grade 2 (56.12 to 76.53 recall) could partly reflect the labeler's bias rather than true disease severity.
  • The thresholds 0.5 and 1.5 on the regression output are fixed hyperparameters; a clinically deployed system would likely need them calibrated against quantitative reference standards such as effective regurgitant orifice area or regurgitant volume.
  • The model currently merges moderate and severe MR into one intervention class; extending to four grades would require retraining the MoE to four experts and would likely need additional data for the severe class.
  • Because X-CLIP is pretrained on Kinetics-400 action videos, the method should transfer to other echocardiographic views (e.g., parasternal long axis) with only fine-tuning, but this is untested in the present study.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper introduces MReg, a regression-based framework for automated mitral regurgitation (MR) diagnosis from four-chamber color Doppler echocardiography videos. The method combines MIL-based instance selection, feature amplification, and an MoE-based feature summary module on top of X-CLIP video features, and is trained with a combination of binary focal loss, MSE regression loss, expert focal loss, and smoothing/sparsity losses. Experiments are reported on an in-house dataset of 1,868 A4C-CDV cases with three grades (normal, mild, moderate-severe), comparing against supervised video classification and weakly supervised video anomaly detection baselines. The authors report state-of-the-art accuracy (89.36%), recall, precision, specificity, and F1 on a held-out test split, with code released.

Significance. If the reported results are reliable, MReg would be a practically useful automated MR grading tool with a sensible clinical motivation (regression over ordinal grades, selection of the most informative cardiac cycle, and amplification of regurgitation features). The paper provides a held-out test split, a clear architecture description, and comparisons to several baselines. However, the evaluation section has unresolved issues: the F1 values are not consistent with the reported precision/recall under common averaging schemes, the ground-truth labels come from a single sonographer with no variability analysis, and all results are single-run without confidence intervals. These issues directly affect the strength of the central SOTA claim.

major comments (4)
  1. [Section 3, Tables 1 and 2] The F1 column is not derivable from the reported Precision and Recall values under micro or weighted averaging. For example, MReg has Precision=86.83 and Recall=85.93, whose harmonic mean is approximately 86.38, yet the table reports F1=94.28. The same pattern appears for Video-Swin (84.39 harmonic vs. reported 93.18) and Baseline (78.35 harmonic vs. reported 90.09). If F1 is macro-averaged, this can be reconciled because macro F1 is not constrained by global precision/recall, but the paper must state the averaging scheme explicitly and report per-class precision, recall, and F1. As written, the headline F1 numbers are not reproducible from the reported metrics and undermine the SOTA comparison.
  2. [Section 3, Dataset and Implementations] The ground-truth labels for all 1,868 cases were provided by a single experienced sonographer, with no inter-observer or intra-observer variability reported, no second-reader adjudication, and no external reference standard (e.g., cardiac MRI or quantitative Doppler parameters). Since every reported accuracy, recall, precision, specificity, and F1 value is computed against these labels, any label subjectivity or noise directly limits the validity of the central clinical claim. The authors should report label variability on a subset or at minimum thoroughly discuss this limitation in the paper.
  3. [Tables 1 and 2] All metrics are reported from a single training run with no confidence intervals, standard deviations, or repeated trials. The P-values in the tables are only comparisons of each method against MReg (Chi-square test), not pairwise significance tests among ablations or baselines. Given the small number of Grade 2 test cases (98) and the observed sensitivity of results to module changes (e.g., Accuracy 88.74 vs. 88.37 in Table 2), the claim that MReg 'significantly outperforms' all competing methods needs variance estimates and appropriate statistical testing to be convincing.
  4. [Section 2.2 and Table 2] The contribution of the MoE-based feature summary module is not isolated in the ablation study. Adding MoE without the auxiliary expert loss (Baseline + FS + Amp + MoE) slightly decreases Accuracy and F1 compared to Baseline + FS + Amp (88.37 vs. 88.74 and 93.90 vs. 93.98, respectively), and only the full MReg with Lexpert improves accuracy to 89.36. The paper should provide an ablation that separates the effect of the MoE module from the effect of Lexpert, and explain why MoE alone degrades performance. This is important because the MoE module is presented as a key contribution.
minor comments (6)
  1. [Section 2, Methodology] The feature dimension notation for f3_img,text is confusing: it is written as R3×16×196×3×512, which likely contains a typo. Please clarify the intended tensor shape.
  2. [Section 3, Metrics] Please define how F1-score is computed (micro, macro, or weighted) and report per-class precision/recall/F1, especially for the Grade 2 category, which is critical for clinical use.
  3. [Figure 4, caption] The phrase 'as largest gaps: ∼0.05 vs. ∼4 in C' is unclear. Please clarify what the numbers refer to and how the gaps were measured.
  4. [Tables 1 and 2, P-Value column] The caption and text should state exactly what statistical test is used (e.g., Chi-square on which contingency table) and what hypothesis is being tested. Currently the P-value column is underspecified.
  5. [Section 2.1, Hyperparameters] The thresholds (thre1=0.5, thre2=1.5), amplification coefficient β=2, and loss weights λ1=0.01, λ2=0.001 are set manually. A sensitivity analysis for these key hyperparameters would strengthen the paper.
  6. [Section 3, Ablation description] In the text describing Table 2, 'rows 3-4' should clearly map to the Baseline+FS and Baseline+FS+Amp rows; the current wording ('rows 3-4, Table 2') is ambiguous because the table rows include additional variants.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: MReg is a supervised empirical pipeline evaluated on a held-out split; no derivation reduces to its inputs.

full rationale

The claimed derivation chain is empirical and supervised end-to-end: video features from X-CLIP are passed through MIL instance selection, feature amplification, and a MoE summary module; all modules are optimized with losses L2cls, L3cls, Lexpert, Lsmooth, and Lsparsity against the same manual labels, and the final grading thresholds (thre1=0.5, thre2=1.5) are fixed hyperparameters rather than quantities fitted to the test labels. Table 1's SOTA result is obtained by training on a training split and evaluating on a separate test split with external baselines; no equation in the paper defines a target quantity in terms of the model output or vice versa. The only self-citation ([11]) appears in the introduction as related disease-detection work and is not used to justify the architecture, loss, or evaluation. The ROI detector and Pair annotation tool are preprocessing utilities, not part of the MR-grading derivation. The reported F1/Precision/Recall inconsistencies in Table 1 are internal-consistency/correctness concerns, not circularity. Hence no circular step is present.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The evaluation rests on label quality and hand-set design choices. The key unverified premise is that a single sonographer's labels are reliable ground truth; if label noise is high or moderate/severe collapse is clinically unjustified, the performance numbers lose meaning. The model also assumes the automatically selected clip contains the most severe regurgitation. Hyperparameters beta, thresholds, and loss weights are manual and not sensitivity-tested.

free parameters (5)
  • Amplification coefficient beta = 2
    Hand-set in Eq. (1); controls how strongly local regurgitation features are amplified; no sensitivity analysis reported.
  • Regression-to-category thresholds = 0.5 and 1.5
    Manual thresholds in Section 2.1 map continuous regression output to Grade 0/1/2; the reported accuracy depends on this mapping.
  • Loss weights lambda1 and lambda2 = 0.01 and 0.001
    Manual weights for smoothing and sparsity losses in the total loss; not ablated.
  • MIL clip configuration = 3 clips x 16 frames
    Manual sampling choice; each clip is assumed to cover more than one cardiac cycle, and Stage I selects one clip for final diagnosis.
  • Number of MoE experts = 3
    One expert per severity class; chosen because there are three grades, not learned from data.
assumptions (5)
  • domain assumption Stage I instance with highest MR probability contains the cardiac cycle with the most severe regurgitation.
    All Stage II diagnosis is routed through selected instance alpha (Eqs. 1-2); the ablation MReg** with random selection drops accuracy from 89.36 to 81.06, showing the assumption is load-bearing.
  • domain assumption Labels from one experienced sonographer are reliable ground truth.
    No inter-observer variability, second reader, or external reference is reported; the model is trained and evaluated against these labels.
  • domain assumption Moderate and severe MR can be collapsed into a single grade because both require intervention.
    This defines the 3-class target and may obscure ordinal differences between moderate and severe cases.
  • domain assumption Each 16-frame clip contains more than one cardiac cycle.
    Stated in the dataset protocol; if false, the selected clip may not include the systolic regurgitation jet needed for grading.
  • domain assumption X-CLIP features pretrained on Kinetics-400 transfer to Doppler echocardiography after fine-tuning.
    The backbone is initialized on natural video and used for medical ultrasound; the authors fine-tune but do not validate feature transfer independently.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MReg: A Novel Regression Model with MoE-based Video Feature Mining for Mitral Regurgitation Diagnosis." pith.science (2026). https://pith.science/paper/3JILYH7Q

@misc{pith2026250623648,
  author       = {Pith},
  title        = {Pith review of: MReg: A Novel Regression Model with MoE-based Video Feature Mining for Mitral Regurgitation Diagnosis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3JILYH7Q}},
  note         = {Machine review of arXiv:2506.23648}
}
read the original abstract

Color Doppler echocardiography is a crucial tool for diagnosing mitral regurgitation (MR). Recent studies have explored intelligent methods for MR diagnosis to minimize user dependence and improve accuracy. However, these approaches often fail to align with clinical workflow and may lead to suboptimal accuracy and interpretability. In this study, we introduce an automated MR diagnosis model (MReg) developed on the 4-chamber cardiac color Doppler echocardiography video (A4C-CDV). It follows comprehensive feature mining strategies to detect MR and assess its severity, considering clinical realities. Our contribution is threefold. First, we formulate the MR diagnosis as a regression task to capture the continuity and ordinal relationships between categories. Second, we design a feature selection and amplification mechanism to imitate the sonographer's diagnostic logic for accurate MR grading. Third, inspired by the Mixture-of-Experts concept, we introduce a feature summary module to extract the category-level features, enhancing the representational capacity for more accurate grading. We trained and evaluated our proposed MReg on a large in-house A4C-CDV dataset comprising 1868 cases with three graded regurgitation labels. Compared to other weakly supervised video anomaly detection and supervised classification methods, MReg demonstrated superior performance in MR diagnosis. Our code is available at: https://github.com/cskdstz/MReg.

Figures

Figures reproduced from arXiv: 2506.23648 by the authors.

Figure 1
Figure 1. Visualizations of normal (A) and various MR cases: (B) mild MR, (C)-(D) [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Overview of our proposed method. frame instances. The MIL-based operation is designed to facilitate the subse￾quent feature selection to use the most representative clip for diagnosis, avoid￾ing interference from others. Then, X-CLIP [13] is used to extract video features fvideo ∈ R3×16×512, and two/three category text features: f 2 img,text ∈ R3×2×512 and f 3 img,text ∈ R3×16×196×3×512, respectively. Last, these fe… view at source ↗
Figure 3
Figure 3. Recall confusion matrix of the proposed method. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: A and B present instance selection for MReg (row 1) and MReg** (row 2) based diagnosis, with the index on the first image. C-F display per-frame regression results, including cases w and w/o Amp (C-D), and using feature summary (E-F). Instance labels are marked on the …
Figure 5
Figure 5. Figure 5: The activation diagram of MReg in the three categories, with the original image at the top and the corresponding activation maps at the bottom. and MReg further reveals the importance of accurate MR instance selection. Furthermore, as shown in [PITH_FULL_IMAGE:figures…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

21 extracted references · 14 canonical work pages

  1. [1]

    In: Proceedings of the IEEE/CVF international confer- ence on computer vision

    Arnab, A., Dehghani, M., Heigold, G., Sun, C., Lučić, M., Schmid, C.: Vivit: A video vision transformer. In: Proceedings of the IEEE/CVF international confer- ence on computer vision. pp. 6836–6846 (2021)

  2. [2]

    Linear algebra and its applications429(8-9), 1864–1885 (2008)

    Böttcher, A., Wenzel, D.: The frobenius norm and the commutator. Linear algebra and its applications429(8-9), 1864–1885 (2008)

  3. [3]

    In: proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Carreira, J., Zisserman, A.: Quo vadis, action recognition? a new model and the kinetics dataset. In: proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 6299–6308 (2017)

  4. [4]

    Procedia Computer Science171, 820–828 (2020)

    Chandra, V., Sarkar, P.G., Singh, V.: Mitral valve leaflet tracking in echocardiog- raphy using custom yolo3. Procedia Computer Science171, 820–828 (2020)

  5. [5]

    Journal of the American Heart Association11(16), e024168 (2022)

    Cheng, L.H., Bosch, P.B., Hofman, R.F., Brakenhoff, T.B., Bruggemans, E.F., van der Geest, R.J., Holman, E.R.: Revealing unforeseen diagnostic image features with deep learning by detecting cardiovascular diseases from apical 4-chamber ul- trasounds. Journal of the American Heart Association11(16), e024168 (2022)

  6. [6]

    Frontiers in Cardiovascular Medicine8, 735587 (2021)

    Herz, C., Pace, D.F., Nam, H.H., Lasso, A., Dinh, P., Flynn, M., Cianciulli, A., Golland, P., Jolley, M.A.: Segmentation of tricuspid valve leaflets from transtho- racic 3d echocardiograms of children with hypoplastic left heart syndrome using deep learning. Frontiers in Cardiovascular Medicine8, 735587 (2021)

  7. [7]

    Computers in Biology and Medicine156, 106705 (2023)

    Li, H., Wang, Y., Qu, M., Cao, P., Feng, C., Yang, J.: Echoefnet: multi-task deep learning network for automatic calculation of left ventricular ejection fraction in 2d echocardiography. Computers in Biology and Medicine156, 106705 (2023)

  8. [8]

    Medical image analysis79, 102461 (2022) 10 Liu et al

    Liang, J., Yang, X., Huang, Y., Li, H., He, S., Hu, X., Chen, Z., Xue, W., Cheng, J., Ni, D.: Sketch guided and progressive growing gan for realistic and editable ultrasound image synthesis. Medical image analysis79, 102461 (2022) 10 Liu et al

Show all 21 references
  1. [9]

    Lin, T.Y., Goyal, P., Girshick, R., He, K., Dollár, P.: Focal loss for dense object detection.In:ProceedingsoftheIEEEinternationalconferenceoncomputervision. pp. 2980–2988 (2017)

  2. [10]

    In: Proceedings of the IEEE/CVF conference on computer vision and pat- tern recognition

    Liu, Z., Ning, J., Cao, Y., Wei, Y., Zhang, Z., Lin, S., Hu, H.: Video swin trans- former. In: Proceedings of the IEEE/CVF conference on computer vision and pat- tern recognition. pp. 3202–3211 (2022)

  3. [11]

    In: International Work- shop on Machine Learning in Medical Imaging

    Liu, Z., Zhu, X., Han, T., Huang, Y., Wang, J., Liu, L., Wang, F., Ni, D., Gou, Z., Yang, X.: Mitral regurgitation recogniton based on unsupervised out-of- distribution detection with residual diffusion amplification. In: International Work- shop on Machine Learning in Medical...

  4. [12]

    Circulation (2024)

    Long, A., Haggerty, C.M., Finer, J., Hartzel, D., Jing, L., Keivani, A., Kelsey, C., Rocha, D., Ruhl, J., vanMaanen, D., et al.: Deep learning for echo analysis, tracking, and evaluation of mitral regurgitation (delineate-mr). Circulation (2024)

  5. [13]

    In: European conference on computer vision

    Ni, B., Peng, H., Chen, M., Zhang, S., Meng, G., Fu, J., Xiang, S., Ling, H.: Expanding language-image pretrained models for general video recognition. In: European conference on computer vision. pp. 1–18. Springer (2022)

  6. [14]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Ren, J., Zhang, M., Yu, C., Liu, Z.: Balanced mse for imbalanced visual regression. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 7926–7935 (2022)

  7. [15]

    Cardiovascular Imaging18(1), 1–12 (2025)

    Sadeghpour, A., Jiang, Z., Hummel, Y.M., Frost, M., Lam, C.S., Shah, S.J., Lund, L.H., Stone, G.W., Swaminathan, M., Weissman, N.J., et al.: An automated ma- chine learning–based quantitative multiparametric approach for mitral regurgita- tion severity grading. Cardiovascular ...

  8. [16]

    Machine Learning and Knowledge Extraction5(4), 1680–1716 (2023)

    Terven J, Córdova-Esparza D M, R.G.J.A.: A comprehensive review of yolo archi- tectures in computer vision: From yolov1 to yolov8 and yolo-nas. Machine Learning and Knowledge Extraction5(4), 1680–1716 (2023)

  9. [17]

    Circulation150(12), 923–933 (2024)

    Vrudhula, A., Duffy, G., Vukadinovic, M., Liang, D., Cheng, S., Ouyang, D.: High- throughput deep learning detection of mitral regurgitation. Circulation150(12), 923–933 (2024)

  10. [18]

    Electronics10(4), 495 (2021)

    Wahlang, I., Maji, A.K., Saha, G., Chakrabarti, P., Jasinski, M., Leonowicz, Z., Jasinska, E.: Deep learning methods for classification of certain abnormalities in echocardiography. Electronics10(4), 495 (2021)

  11. [19]

    arXiv preprint arXiv:2109.08472 (2021)

    Wang, M., Xing, J., Liu, Y.: Actionclip: A new paradigm for video action recogni- tion. arXiv preprint arXiv:2109.08472 (2021)

  12. [20]

    Cardiovascular Imaging15(4), 551–563 (2022)

    Yang, F., Chen, X., Lin, X., Chen, X., Wang, W., Liu, B., Li, Y., Pu, H., Zhang, L., Huang, D., et al.: Automated analysis of doppler echocardiographic videos as a screening tool for valvular heart diseases. Cardiovascular Imaging15(4), 551–563 (2022)

  13. [21]

    Computer Vision and Image Understanding 249, 104163 (2024)

    Zanella, L., Liberatori, B., Menapace, W., Poiesi, F., Wang, Y., Ricci, E.: Delving into clip latent space for video anomaly recognition. Computer Vision and Image Understanding 249, 104163 (2024)

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.