Pith. sign in

REVIEW 4 major objections 7 minor 32 references

Lightweight Method for Interactive 3D Medical Image Segmentation with Multi-Round Result Fusion

T0 review · 4 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read A lightweight CNN can match SAM-based models on unseen medical volumes

desk verdict A plausible lightweight CNN alternative to SAM for interactive medical segmentation, but the 'fewer interactions' claim is untestable until baseline click budgets are specified per slice or per volume. read the letter →

arxiv 2412.08315 v1 pith:TJ75OE4S submitted 2024-12-11 cs.CV

classification cs.CV
keywords interactivesegmentation3DmedicalimagelightweightCNNmulti-roundresultfusionmemory-augmentedpropagationzero-shotgeneralizationSegmentAnythingModelvolumetricimaging
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

LIM-Net is a compact CNN-based system for interactive 3D medical image segmentation that starts from a few user clicks on one slice and propagates a mask through the whole volume. The paper argues that this lightweight design, built from a 2D HRNet-32 interaction module, an XMem-style memory module, and a multi-round result fusion module, generalizes to unseen medical datasets as well as or better than SAM-based models while using fewer interactions. If true, this would give clinical annotation workflows a fast, low-memory alternative to prompt-based foundation models. The supporting evidence includes Dice scores on the unseen BTCV, MSD-Pancreas, and EndoVis18 benchmarks and an inference speed of 31.25 FPS with about 2.7 GB GPU memory.

What carries the argument

The method has three moving parts: a 2D Interaction Module that encodes user clicks into maps and produces an initial mask with an HRNet-32 backbone per RITM; a Memory-Augmented 3D Propagation module that takes that prompt mask and runs an XMem-style sensory/working/long-term memory model bidirectionally along the slice sequence; and the Multi-Round Result Fusion (MRF) module, a ResNet-50 quality-assessment network that scores whether the previous round's mask is better than the current round's for each slice and fuses the better one with a threshold rule. The MRF module is what stabilizes quality on slices far from the prompt and, the paper argues, makes improvement monotonic across interaction rounds.

What would settle it

One concrete test is to re-run LIM-Net, SAM-B, SAM-Med3D, and 3DSAM-adapter on BTCV, MSD-Pancreas, and EndoVis18 with exactly the same number of clicks per volume (e.g., six) and no target-dataset fine-tuning; if any SAM-based model matches or exceeds LIM-Net's reported Dice, the central claim fails. A second check is to evaluate the MRF module on an unseen modality and confirm that its fused mask never has a lower Dice than simply keeping the current round's mask; one violation would disprove the claimed monotonic improvement.

Watch

Extended reading notes

Core claim

The central discovery the paper is trying to establish is that traditional CNN networks can achieve strong zero-shot generalization in interactive 3D medical segmentation when trained on a large, diverse mixture of medical and natural images and equipped with long-term memory propagation plus slice-wise multi-round fusion. On unseen organs and modalities, LIM-Net with six clicks per volume reports Dice scores of 52.9% on the BTCV esophagus, 63.76% on MSD-Pancreas tumor, and comparable per-instrument scores on EndoVis18 surgical sequences, outperforming SAM-B, SAM-Med3D, and 3DSAM-adapter in several categories. The paper's claim is that a compact model with translation-equivariant convolutions can close the generalization gap to foundation models on medical structures that have distinct intensity characteristics.

Load-bearing premise

The comparison assumption is that every method gets an equivalent interaction budget: the paper leaves SAM-B's N undefined in one table and does not say whether 3DSAM-adapter's 10 points are per slice or per volume, so the 'stronger generalization with fewer interactions' claim depends on those budgets being equal or fairly specified.

Editorial extensions

If this is right

  • A lightweight CNN can serve as a deployable interactive segmenter on GPUs with roughly 2.7 GB of memory, opening real-time annotation on modest hardware.
  • Multi-round fusion raises Dice by 0.36 to 5.58 points on the four evaluated CT datasets and prevents distant slices from degrading after later interactions.
  • Training on a mix of natural, CT, and ultrasound images transfers to unseen organs and surgical video without fine-tuning, reducing the retraining burden.
  • LIM-Net can act as a fast baseline for interactive medical segmentation, complementing SAM-based tools and lowering the number of clicks needed per volume.

Reading between the lines

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

  • If every baseline were given exactly the same number of clicks per volume, the reported margins over SAM-B and 3DSAM-adapter might narrow; the paper's tables do not fully specify interaction budgets, so a matched-budget comparison is an open test.
  • The MRF quality network is trained on defect masks synthesized from four CT datasets, so its fusion decisions on ultrasound or other modalities are an extrapolation; a self-supervised quality estimator could make the module more transferable.
  • The bidirectional memory propagation could be attached to a SAM prompt encoder, potentially letting foundation models match LIM-Net's efficiency; the paper does not test this combination.
  • If a slice-wise oracle (choosing the mask with higher ground-truth Dice) is used in place of the quality network, one could measure how much headroom remains in the fusion strategy; the paper does not report this ceiling.
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 / 7 minor

Summary. The paper proposes LIM-Net, a lightweight CNN-based interactive 3D medical image segmentation method composed of a 2D click-driven interaction module, an XMem-inspired memory-augmented propagation module, and a Multi-Round Result Fusion (MRF) module that selectively merges masks across interaction rounds. The authors train on a diverse mix of medical and natural image datasets and evaluate on seen datasets (KiTS19, MSD-Colon, MSD-Lung) and unseen datasets (BTCV, MSD-Pancreas, EndoVis18). The abstract claims that LIM-Net exhibits stronger generalization to unseen data than SAM-based models while requiring fewer interactions, with low GPU memory consumption. The paper reports Dice scores, runtime, and GPU memory usage, and includes an ablation of the MRF module.

Significance. If the central claims were fully supported, the paper would make a useful contribution: a lightweight interactive segmentation model that can run in resource-constrained settings at 31 FPS with 2.7 GB GPU memory, while matching or exceeding prompt-based foundation models on unseen medical volumes, would be practically valuable. The MRF idea of selectively fusing masks from multiple interaction rounds is also worth investigating, and the training recipe on a diverse mixture of medical and natural images is a legitimate design choice. However, the current evidence has several load-bearing gaps: the interaction budgets for SAM baselines are ambiguously defined, the 'unseen' evaluation in Table I is undermined by Stage 3 training on the same datasets, the MRF ablation is measured on the same datasets used to construct its training data, and no error bars or significance tests are reported. These gaps prevent the abstract's claims from being verified as written.

major comments (4)
  1. [§III.C, Tables IV–VI] The headline claim that LIM-Net requires 'fewer interactions' than SAM-based models is not testable from the reported tables. In Table IV, SAM-B is evaluated with 'N points' and '6N points' but N is never defined; if N is points per slice, the total budget could be orders of magnitude larger than LIM-Net's '6 points'. In Table V, '3DSAM-adapter(10 points)' is compared with 'Ours(6 clicks/volume)' without stating whether the adapter's 10 points are per slice or per volume. In Table VI, SAM-B(N points) is again compared with 'Ours(6 points)' without defining N or the per-slice/per-volume convention. Because the central claim is defined by the ratio of interaction counts, the authors must either define these budgets precisely or rerun the baselines with exactly 6 total clicks/points per volume and report Dice at matched interaction counts.
  2. [§III.B, Table I] The implementation details state that after Stage 2, 'Stage 3 further trained the model on these specific datasets' for Table I, and that the 3D model for Table I was trained on MSD-Colon, MSD-Lung, and KiTS19. Table I then reports results on exactly those datasets. Therefore Table I cannot be used as evidence for zero-shot generalization or for the claim of 'stronger generalization to unseen data'. The paper should clearly separate in-domain benchmark results from truly unseen evaluations, and the abstract and conclusion should not imply that Table I supports generalization claims.
  3. [§III.B, §II.C3, Table II] The MRF quality network is trained on a defect dataset 'constructed using a baseline model without MRF on four datasets: KiTS19-Organ, KiTS19-Tumor, MSD-lung, and MSD-colon'. The MRF ablation in Table II and the round-by-round analysis in Figure 3 are then evaluated on those same four datasets. This means the MRF gain is measured on the distribution of defect masks that the quality network was trained on, which can overstate the module's benefit. Please report the MRF ablation on held-out datasets (e.g., on the unseen BTCV, MSD-Pancreas, and EndoVis18 evaluations) and also specify the value of the fusion threshold τ in Eq. (4), with a sensitivity analysis over τ.
  4. [Tables IV–VI] All quantitative results are reported as single Dice numbers without standard deviations, confidence intervals, or the number of test volumes per cell. This is especially problematic in Table VI, where each category may correspond to one or very few sequences, making differences such as 9.0 vs 3.2 or 71.9 vs 22.9 difficult to interpret. The authors should report mean ± std over cases or folds, state how many volumes/sequences each number is based on, and indicate whether any significance testing was performed.
minor comments (7)
  1. [§II.C3] The sentence 'it guarantees monotonic improvement in overall segmentation quality' is too strong: Eq. (4) uses a learned quality network and a threshold τ, and if the quality estimate is wrong, the fusion can select a worse mask. Please replace 'guarantees' with a more cautious statement or provide a formal proof under explicit assumptions.
  2. [§II.D, Eq. (10)] The notation in Eqs. (10)–(12) is not fully defined in the text (e.g., S, A, Ck, the subscripts on c, and the dimensions of the matrices). Although the authors refer to XMem for details, the equations should be self-contained enough for a reader to follow the fusion computation.
  3. [Table V] Table V lists both 'Colon Cancer' and 'Pancreas Tumor' columns, but the text only discusses the pancreas tumor result on MSD-Pancreas. Please clarify what the 'Colon Cancer' column reports, which dataset it comes from, and whether that dataset was seen or unseen during training.
  4. [Tables IV–VI] The interaction units are inconsistent across tables: Table IV uses '6 points', Table V uses '6 clicks/volume', and Table VI uses '6 points'. Please define 'click' and 'point' precisely and use one convention throughout so that the interaction budgets are directly comparable.
  5. [Table III, §III.B] In Table III, the In-house CT dataset has no test split, while the text says the in-house CT and US data were used in training. Please clarify the split for the in-house data and whether any in-house test set exists.
  6. [Figure 3, Table I] The text mentions that on KiTS19-Organ, improvements become limited when the number of interaction rounds exceeds 7, while Table I reports results after six rounds. Please state the exact number of rounds used for each reported result and align the figure axes and the table caption.
  7. [§III.B] The term 'XMem-s012' is introduced without explanation. Please define which variant of XMem this is and why it was chosen for the EndoVis18 experiments.

Circularity Check

1 steps flagged · score 4.0 of 10

MRF gain is partly in-sample because the quality network is trained on the same KiTS19/MSD datasets used in Tables I–II; the unseen-data generalization claim is not circular.

  1. fitted input called prediction [Section III-B (Implementation details); evaluated in Table I and ablation Table II on KiTS19-Organ, KiTS19-Tumor, MSD-Lung, MSD-Colon]
    "The MRF module is based on a pretrained ResNet-18 model, which was trained on a defect dataset. The defect dataset was constructed using a baseline model without MRF on four datasets: KiTS19-Organ, KiTS19-Tumor, MSD-lung, and MSD-colon, which comprised both defect masks and ground truth masks."

    The MRF quality-assessment network is trained on defect masks generated by the authors' own no-MRF baseline on exactly KiTS19-Organ, KiTS19-Tumor, MSD-Lung, and MSD-Colon. The paper then reports MRF's benefit on the same four datasets: Table I reports Kidney (Organ/Tumor), Lung, and Colon results, and Table II compares Ours w/o MRF against Ours-MRF on the identical columns. The reported Dice gains of MRF on those datasets are therefore in-sample measurements of a classifier fitted to the error distribution of the very baseline it is supposed to improve; they are not independent evidence of a general fusion benefit. The unseen-data comparisons (BTCV, MSD-Pancreas, EndoVis18) do not share this flaw, so the central generalization claim is not reduced by construction.

full rationale

The central claim — stronger generalization to unseen data than SAM-based models — rests on Tables IV, V, and VI, which evaluate on BTCV, MSD-Pancreas, and EndoVis18. Those datasets are not used to train the MRF quality network, and the comparison methods are external published baselines or challenge entries rather than self-citations. No load-bearing self-citation chain or imported uniqueness theorem is present. The main circularity concern is localized to the MRF ablation: the quality-assessment network is trained on defect masks produced by the authors' own baseline without MRF on KiTS19-Organ, KiTS19-Tumor, MSD-Lung, and MSD-Colon, and the reported MRF improvements in Table I and Table II are measured on exactly those same datasets. That makes the MRF contribution on seen data partly a fitted result rather than a transferable finding. Because the paper's headline generalization evaluation is on genuinely held-out datasets and the comparison to SAM-based models does not reduce to any fitted parameter by construction, the overall circularity is moderate rather than severe. The undefined interaction budgets for SAM baselines (e.g., 'N points' in Tables IV and VI, '10 points' in Table V) are a serious reporting and validity problem for the 'fewer interactions' claim, but they are not a circularity and are not counted in this score.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

No new physical or ontological entities are introduced. The MRF module and memory stores are algorithmic components, not entities requiring external falsifiable handles. The main free parameters are the fusion threshold, the quality-network weights, and the hand-chosen corruption settings for defective mask generation.

free parameters (4)
  • MRF fusion threshold tau = unspecified (likely 0.5)
    Equation (4) selects the previous mask when Pi is greater than tau; the value is never stated and no ablation is provided. The reported fusion results depend on it.
  • MRF quality network parameters theta = trained on defect dataset
    The network weights are fit to predict which of two masks is better. The defect dataset is generated by the authors' baseline on KiTS19, MSD-Lung, and MSD-Colon, so the module is tuned to those distributions.
  • Defective mask corruption probabilities = [0.1, 0.1, 0.3, 0.2, 0.1, 0.2]
    Hand-chosen probabilities for random shapes, morphology, boundary displacement, smoothing, removal, and merging; they affect interaction-module training and are not justified by sensitivity analysis.
  • Morphology and displacement ranges = dilation 10-30 iterations, erosion 10, displacement 10-30 pixels
    Hand-chosen ranges for defective mask generation, with no reported sensitivity study.
assumptions (5)
  • domain assumption Medical images have distinctive intensity variations and well-defined regions of interest, so CNN inductive biases transfer across modalities.
    Invoked in the Introduction to justify the hypothesis that CNN-based approaches can generalize to unseen medical data.
  • domain assumption The quality assessment network's probability Pi correctly identifies the better of two masks for each slice.
    Equations (2) to (4) and the claimed monotonic improvement require the learned Pi to be accurate on unseen data.
  • domain assumption XMem's memory consolidation and matching equations, shown in Equations (7) to (12), are correct and applicable to medical image slices.
    The paper adopts XMem without re-derivation and applies it bidirectionally to volumetric medical data.
  • domain assumption Training on a large mixed dataset, including AMOS, COCO-LVIS, DAVIS, and in-house CT and ultrasound, yields zero-shot generalization to unseen medical domains.
    This is the central learning assumption behind the generalization claim in Sections III-A and III-B.
  • domain assumption The MSD-Pancreas, BTCV, and EndoVis18 datasets were not seen during training.
    The paper asserts these are unseen, but training includes MSD-Colon, MSD-Lung, KiTS19, and AMOS from similar CT distributions, and no patient-level overlap check is reported.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Lightweight Method for Interactive 3D Medical Image Segmentation with Multi-Round Result Fusion." pith.science (2026). https://pith.science/paper/TJ75OE4S

@misc{pith2026241208315,
  author       = {Pith},
  title        = {Pith review of: Lightweight Method for Interactive 3D Medical Image Segmentation with Multi-Round Result Fusion},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TJ75OE4S}},
  note         = {Machine review of arXiv:2412.08315}
}
read the original abstract

In medical imaging, precise annotation of lesions or organs is often required. However, 3D volumetric images typically consist of hundreds or thousands of slices, making the annotation process extremely time-consuming and laborious. Recently, the Segment Anything Model (SAM) has drawn widespread attention due to its remarkable zero-shot generalization capabilities in interactive segmentation. While researchers have explored adapting SAM for medical applications, such as using SAM adapters or constructing 3D SAM models, a key question remains: Can traditional CNN networks achieve the same strong zero-shot generalization in this task? In this paper, we propose the Lightweight Interactive Network for 3D Medical Image Segmentation (LIM-Net), a novel approach demonstrating the potential of compact CNN-based models. Built upon a 2D CNN backbone, LIM-Net initiates segmentation by generating a 2D prompt mask from user hints. This mask is then propagated through the 3D sequence via the Memory Module. To refine and stabilize results during interaction, the Multi-Round Result Fusion (MRF) Module selects and merges optimal masks from multiple rounds. Our extensive experiments across multiple datasets and modalities demonstrate LIM-Net's competitive performance. It exhibits stronger generalization to unseen data compared to SAM-based models, with competitive accuracy while requiring fewer interactions. Notably, LIM-Net's lightweight design offers significant advantages in deployment and inference efficiency, with low GPU memory consumption suitable for resource-constrained environments. These promising results demonstrate LIM-Net can serve as a strong baseline, complementing and contrasting with popular SAM models to further boost effective interactive medical image segmentation. The code will be released at \url{https://github.com/goodtime-123/LIM-Net}.

Figures

Figures reproduced from arXiv: 2412.08315 by the authors.

Figure 5
Figure 5. It can be observed that although our model was [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 15 canonical work pages

  1. [1]

    nnu-net: a self-configuring method for deep learning-based biomedical image segmentation,

    F. Isensee, P. F. Jaeger, S. A. Kohl, J. Petersen, and K. H. Maier-Hein, “nnu-net: a self-configuring method for deep learning-based biomedical image segmentation,” Nature methods, vol. 18, no. 2, pp. 203–211, 2021

  2. [2]

    Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images,

    A. Hatamizadeh, V . Nath, Y . Tang, D. Yang, H. R. Roth, and D. Xu, “Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images,” in International MICCAI Brainlesion Workshop . Springer, 2021, pp. 272–284. 10 IEEE TRANSACTIONS AND JOURNALS TEMPLA TE

  3. [3]

    Unetr++: delving into efficient and accurate 3d medical image segmentation,

    A. Shaker, M. Maaz, H. Rasheed, S. Khan, M.-H. Yang, and F. S. Khan, “Unetr++: delving into efficient and accurate 3d medical image segmentation,” arXiv preprint arXiv:2212.04497 , 2022

  4. [4]

    nn- former: V olumetric medical image segmentation via a 3d transformer,

    H.-Y . Zhou, J. Guo, Y . Zhang, X. Han, L. Yu, L. Wang, and Y . Yu, “nn- former: V olumetric medical image segmentation via a 3d transformer,” IEEE Transactions on Image Processing , 2023

  5. [5]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y . Loet al., “Segment anything,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 4015–4026

  6. [6]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly et al. , “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv:2010.11929 , 2020

  7. [7]

    Medical sam adapter: Adapting segment anything model for medical image segmentation,

    J. Wu, R. Fu, H. Fang, Y . Liu, Z. Wang, Y . Xu, Y . Jin, and T. Arbel, “Medical sam adapter: Adapting segment anything model for medical image segmentation,” arXiv preprint arXiv:2304.12620 , 2023

  8. [8]

    3dsam-adapter: Holistic adaptation of sam from 2d to 3d for promptable medical image segmentation,

    S. Gong, Y . Zhong, W. Ma, J. Li, Z. Wang, J. Zhang, P.-A. Heng, and Q. Dou, “3dsam-adapter: Holistic adaptation of sam from 2d to 3d for promptable medical image segmentation,” arXiv preprint arXiv:2306.13465, 2023

Show all 32 references
  1. [9]

    Sam-med3d,

    H. Wang, S. Guo, J. Ye, Z. Deng, J. Cheng, T. Li, J. Chen, Y . Su, Z. Huang, Y . Shen et al. , “Sam-med3d,” arXiv preprint arXiv:2310.15161, 2023

  2. [10]

    Are transformers more robust than cnns?

    Y . Bai, J. Mei, A. L. Yuille, and C. Xie, “Are transformers more robust than cnns?” Advances in neural information processing systems , vol. 34, pp. 26 831–26 843, 2021

  3. [11]

    Interactive medical image segmentation using deep learning with image-specific fine tuning,

    G. Wang, W. Li, M. A. Zuluaga, R. Pratt, P. A. Patel, M. Aertsen, T. Doel, A. L. David, J. Deprest, S. Ourselin et al. , “Interactive medical image segmentation using deep learning with image-specific fine tuning,” IEEE transactions on medical imaging , vol. 37, no. 7, pp. 156...

  4. [12]

    Quality-aware memory network for interactive volumetric image segmentation,

    T. Zhou, L. Li, G. Bredell, J. Li, and E. Konukoglu, “Quality-aware memory network for interactive volumetric image segmentation,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2021: 24th International Conference, Strasbourg, France, September 27–October...

  5. [13]

    A hybrid propagation network for interactive volumetric image segmentation,

    L. Shi, X. Zhang, Y . Liu, and X. Han, “A hybrid propagation network for interactive volumetric image segmentation,” in International Conference on Medical Image Computing and Computer-Assisted Intervention . Springer, 2022, pp. 673–682

  6. [14]

    Exploring cycle consistency learning in interactive volume segmentation,

    Q. Liu, M. Zheng, B. Planche, Z. Gao, T. Chen, M. Niethammer, and Z. Wu, “Exploring cycle consistency learning in interactive volume segmentation,” arXiv preprint arXiv:2303.06493 , 2023

  7. [15]

    Rethinking space-time networks with improved memory coverage for efficient video object segmentation,

    H. K. Cheng, Y .-W. Tai, and C.-K. Tang, “Rethinking space-time networks with improved memory coverage for efficient video object segmentation,” Advances in Neural Information Processing Systems , vol. 34, pp. 11 781–11 794, 2021

  8. [16]

    Deepigeos: a deep interactive geodesic framework for medical image segmentation,

    G. Wang, M. A. Zuluaga, W. Li, R. Pratt, P. A. Patel, M. Aertsen, T. Doel, A. L. David, J. Deprest, S. Ourselin et al., “Deepigeos: a deep interactive geodesic framework for medical image segmentation,” IEEE transactions on pattern analysis and machine intelligence , vol. 41, ...

  9. [17]

    Xmem: Long-term video object segmentation with an atkinson-shiffrin memory model,

    H. K. Cheng and A. G. Schwing, “Xmem: Long-term video object segmentation with an atkinson-shiffrin memory model,” in European Conference on Computer Vision . Springer, 2022, pp. 640–658

  10. [18]

    Reviving iterative training with mask guidance for interactive segmentation,

    K. Sofiiuk, I. A. Petrov, and A. Konushin, “Reviving iterative training with mask guidance for interactive segmentation,” in 2022 IEEE Inter- national Conference on Image Processing (ICIP) . IEEE, 2022, pp. 3141–3145

  11. [19]

    Deep high-resolution representation learning for visual recognition,

    J. Wang, K. Sun, T. Cheng, B. Jiang, C. Deng, Y . Zhao, D. Liu, Y . Mu, M. Tan, X. Wang et al. , “Deep high-resolution representation learning for visual recognition,” IEEE transactions on pattern analysis and machine intelligence , vol. 43, no. 10, pp. 3349–3364, 2020

  12. [20]

    ishape: A first step towards irregular shape instance segmentation,

    L. Yang, Y . Z. Wei, Y . He, W. Sun, Z. Huang, H. Huang, and H. Fan, “ishape: A first step towards irregular shape instance segmentation,” arXiv preprint arXiv:2109.15068 , 2021

  13. [21]

    Amos: A large-scale abdominal multi-organ benchmark for versatile medical image segmentation,

    Y . Ji, H. Bai, C. Ge, J. Yang, Y . Zhu, R. Zhang, Z. Li, L. Zhanng, W. Ma et al. , “Amos: A large-scale abdominal multi-organ benchmark for versatile medical image segmentation,” Advances in Neural Information Processing Systems, vol. 35, pp. 36 722–36 732, 2022

  14. [22]

    Focalclick: Towards practical interactive image segmentation,

    X. Chen, Z. Zhao, Y . Zhang, M. Duan, D. Qi, and H. Zhao, “Focalclick: Towards practical interactive image segmentation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 1300–1309

  15. [23]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 . Springer,...

  16. [24]

    Lvis: A dataset for large vocabulary instance segmentation,

    A. Gupta, P. Dollar, and R. Girshick, “Lvis: A dataset for large vocabulary instance segmentation,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 5356– 5364

  17. [25]

    A large annotated medical image dataset for the development and evaluation of segmentation algorithms,

    A. L. Simpson, M. Antonelli, S. Bakas, M. Bilello, K. Farahani, B. Van Ginneken, A. Kopp-Schneider, B. A. Landman, G. Litjens, B. Menze et al. , “A large annotated medical image dataset for the development and evaluation of segmentation algorithms,” arXiv preprint arXiv:1902.0...

  18. [26]

    The kits19 challenge data: 300 kidney tumor cases with clinical context, ct semantic segmentations, and surgical outcomes,

    N. Heller, N. Sathianathen, A. Kalapara, E. Walczak, K. Moore, H. Kaluzniak, J. Rosenberg, P. Blake, Z. Rengel, M. Oestreich et al. , “The kits19 challenge data: 300 kidney tumor cases with clinical context, ct semantic segmentations, and surgical outcomes,” arXiv preprint arX...

  19. [27]

    2018 Robotic Scene Segmentation Challenge,

    M. Allan, S. Kondo, S. Bodenstedt, S. Leger, R. Kadkhodamohammadi, I. Luengo, F. Fuentes, E. Flouty, A. Mohammed, M. Pedersen, A. Kori, V . Alex, G. Krishnamurthi, D. Rauber, R. Mendel, C. Palm, S. Bano, G. Saibro, C.-S. Shih, H.-A. Chiang, J. Zhuang, J. Yang, V . Iglovikov, A...

  20. [28]

    Miccai multi-atlas labeling beyond the cranial vault–workshop and challenge,

    B. Landman, Z. Xu, J. Igelsias, M. Styner, T. Langerak, and A. Klein, “Miccai multi-atlas labeling beyond the cranial vault–workshop and challenge,” in Proc. MICCAI Multi-Atlas Labeling Beyond Cranial V ault—Workshop Challenge, vol. 5, 2015, p. 12

  21. [29]

    Transbts: Mul- timodal brain tumor segmentation using transformer,

    W. Wang, C. Chen, M. Ding, H. Yu, S. Zha, and J. Li, “Transbts: Mul- timodal brain tumor segmentation using transformer,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2021: 24th International Conference, Strasbourg, France, September 27–October 1, 2021,...

  22. [30]

    Self-supervised pre-training of swin transformers for 3d medical image analysis,

    Y . Tang, D. Yang, W. Li, H. Roth, B. Landman, D. Xu, V . Nath, and A. Hatamizadeh, “Self-supervised pre-training of swin transformers for 3d medical image analysis,” 06 2022, pp. 20 698–20 708

  23. [31]

    3d ux-net: A large kernel volumetric convnet modernizing hierarchical transformer for medical image segmentation,

    H. H. Lee, S. Bao, Y . Huo, and B. A. Landman, “3d ux-net: A large kernel volumetric convnet modernizing hierarchical transformer for medical image segmentation,” ArXiv, vol. abs/2209.15076, 2022

  24. [32]

    Sam- med2d,

    J. Cheng, J. Ye, Z. Deng, J. Chen, T. Li, H. Wang, Y . Su, Z. Huang, J. Chen, L. Jiang, H. Sun, J. He, S. Zhang, M. Zhu, and Y . Qiao, “Sam- med2d,” 2023

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.