Pith. sign in

REVIEW 4 major objections 6 minor 43 references

Leveraging Segment Anything Model for Source-Free Domain Adaptation via Dual Feature Guided Auto-Prompting

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read A frozen MedSAM, prompted by bounding boxes found through dual feature guided search, outperforms existing source-free domain adaptation methods on medical segmentation benchmarks.

desk verdict Real SFDA gains from a frozen MedSAM, but the headline margin needs multi-seed runs and a principled validation split before I'd trust the exact numbers. read the letter →

arxiv 2505.08527 v3 pith:JF7AJCJ3 submitted 2025-05-13 cs.CV

classification cs.CV
keywords source-freedomainadaptationmedicalimagesegmentationSegmentAnythingModelauto-promptingboundingboxpromptpseudo-labelrefinementfeatureaggregationMedSAM
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

The paper tries to show that a frozen general-purpose segmentation model, MedSAM, can be turned into a reliable pseudo-labeler for source-free domain adaptation by automatically finding good bounding-box prompts. Existing SFDA methods produce defective boxes because their predictions are biased by the domain gap, and naively passing those boxes to MedSAM adds little or hurts accuracy. The proposed Dual Feature Guided (DFG) method first aggregates target features toward class prototypes, then expands a box prompt through two feature spaces—the target model's and MedSAM's—until MedSAM's output enters a stable interval, and finally keeps only the largest connected component of each predicted organ. If the method is right, a site with only a source model and unlabeled target scans can get segmentation accuracy close to fully supervised training, without sending data back to the source site.

What carries the argument

The carrying mechanism is the dual feature guided box prompt search (DBS). It starts from pixels whose target-model probability is within $p_{\Delta}$ of the class maximum, then repeatedly adds spatially adjacent pixels whose features are close (cosine similarity above $\tau_f = 0.99$) to expand a box; in parallel it expands from MedSAM's segmentation seed using a distance threshold proportional to the feature divergence within that seed. At each expansion, MedSAM is run on the current box and the number of changed pixels between consecutive predictions, $\Delta_M$, is compared; the search stops when a long enough interval with low $\Delta_M$ is found, meaning the box sits around an organ boundary. A connectivity-based post-processing (CP) then keeps only the largest connected component per class to remove enlarged false-positive regions.

What would settle it

Take a held-out target slice with ground-truth labels, expand a ground-truth bounding box outward by 1, 2, and 3 pixels, and count changed MedSAM prediction pixels at each step; if the change is not consistently low while the box stays inside the organ and sharply higher once it crosses the boundary, the stable-interval stopping rule that DFG relies on is not reliable in that domain.

Watch

Extended reading notes

Core claim

The paper's central claim is that MedSAM's segmentation output is locally stable when the box prompt is perturbed near the true organ boundary, and that this stability can be used as a stopping signal while growing a box from confidently predicted pixels. To use this signal, the paper observes two feature-distribution properties: target-model features of one class remain partially clustered after feature aggregation, while MedSAM features, though lacking class semantics, separate foreground from background well. DFG therefore grows the box using propagation over the target model's feature space (TBS) when features are clustered, and over MedSAM's feature space (MBS) when features are dispersed, stopping at the first long stable interval in the change of MedSAM prediction pixels. On four benchmarks the refined pseudo-labels and connectivity-based post-processing raise average Dice above all compared SFDA methods, with the prostate result (93.3%) nearly matching the fully supervised upper bound (94.4%).

Load-bearing premise

The entire pipeline depends on MedSAM's output remaining stable while the box prompt fluctuates near the true organ boundary and changing sharply when the box crosses it; if a target domain produces accidental stable intervals or unstable outputs near boundaries, the search stops at the wrong place and the pseudo-labels do not improve.

Editorial extensions

If this is right

  • On CHAOS to BTCV (MRI to CT), average Dice rises from 74.4% with ProtoContra to 84.9% with DFG, and average ASSD drops from 4.87 to 1.99.
  • On BTCV to CHAOS (CT to MRI), DFG reaches 85.2% average Dice and 1.00 average ASSD, surpassing all compared methods.
  • On CHAOS to CURVAS (MRI to CT), DFG reaches 85.7% average Dice and 2.27 average ASSD.
  • On NCI-ISBI to QUBIQ prostate adaptation, DFG attains 93.3% Dice and 2.18 ASSD, close to the fully supervised upper bound of 94.4% Dice.
  • Finetuning SAM on bad pseudo-labels (ProtoContra plus SAMed) drops average Dice to 67.3%, so keeping MedSAM frozen while searching for boxes is the more effective way to use the foundation model.

Reading between the lines

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

  • If the box-prompt stability property transfers to point prompts or to faster promptable networks, the same search loop could be adapted for real-time or full 3D-volumetric settings.
  • Because MedSAM is frozen, the framework should improve automatically when a stronger promptable segmenter is released, without changing the adaptation algorithm.
  • The largest-connected-component rule embeds a single-object-per-class assumption; organs split into disconnected pieces would be penalized, so a multi-component variant is a natural testable extension.
  • The reported failure cases suggest the stable-interval detector can be fooled by accidental stability, so an ensemble across box sizes or a learned boundary detector could be tested against the same benchmarks.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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 proposes DFG, a source-free domain adaptation (SFDA) method for medical image segmentation that uses the frozen medical foundation model MedSAM as an external teacher. A feature aggregation (FA) phase adapts the source model to the target feature distribution, after which two box-prompt search procedures, TBS over target model features and MBS over MedSAM features, progressively expand a bounding box until a stable interval in MedSAM output is detected via Eq. (5). The resulting pseudo-labels are post-processed by connectivity analysis (CP) and used to train the target model with a Dice loss. Experiments on three abdominal MRI/CT adaptation tasks and one prostate MRI task report average Dice of 84.9%, 85.2%, 85.7%, and 93.3%, outperforming the compared SFDA baselines.

Significance. If the empirical results are robust, the paper offers a practically useful way to inject a frozen medical foundation model into SFDA, and it does so with an original combination: using both MedSAM features and MedSAM outputs for auto-prompting, rather than fine-tuning SAM on pseudo-labels. The paper includes an ablation study, an efficiency analysis, and an unusually candid failure-case section, and the code is released. However, the headline margin over strong baselines is not yet robustly established: all results are single-run, the many hyperparameters appear to be selected using the test benchmarks, and the central stable-interval assumption is explicitly shown to fail in some cases. The contribution is therefore plausible but needs stronger empirical support before the SOTA claims can be accepted.

major comments (4)
  1. [V-A3, Fig. 6, Tables I-III] The hyperparameters of DFG (tau_f=0.99, tau_Div=2.5, p_Delta=0.005, r=4, tau_max=0.35/0.3, tau_Delta=15/30/45, kappa=1/10, and the margin m) appear to be selected with access to the test labels of the same benchmarks, since Section V-A3 gives no label-free validation criterion and Fig. 6 sweeps values on the CHAOS-to-BTCV task and reports final test Dice. Because SFDA assumes target labels are unavailable, this selection protocol makes the reported margins over the baselines optimistic and their size unquantified. Please introduce a validation protocol that does not use target test labels (e.g., a frozen target split, pseudo-label stability, or an unsupervised criterion), and report the sensitivity of the headline margins over the Fig. 6 ranges on all three benchmarks.
  2. [Tables I-VIII] All reported numbers are single-run results with no standard deviations or seed counts. Some of the claimed advantages are modest (for example, CT-to-MRI average Dice is 85.2% for DFG versus 80.5% for ProtoContra, and several component gains in Table IV are only one to three points), so run-to-run variance could affect the ranking. Please report mean and standard deviation over at least three seeds for DFG and the reimplemented baselines, and include the top two or three hyperparameter configurations from Fig. 6.
  3. [V-C7, Fig. 7, Eq. (5)] The stopping rule in Eq. (5) assumes that MedSAM output is stable exactly when the box prompt is near the true organ boundary, but Section V-C7 and Fig. 7 document three failure modes: organ features too variable to be included in the propagation, MedSAM output not stable enough near the ground truth, and a false stable interval appearing before the boundary. Because this stability property is the load-bearing mechanism that converts box search into pseudo-label refinement, the manuscript should quantify its failure rate (e.g., the proportion of organ instances for which the selected box is consistent with the ground-truth box on the target training data) and demonstrate that the average gains are not driven primarily by slices where the heuristic succeeds by chance.
  4. [V-A3, Tables I-III] The FVP baseline is not reimplemented: Section V-A3 states that FVP results are taken directly from the original paper, while DPL, AdaMI, UPL, and ProtoContra are reimplemented with tuned batch sizes under the same network and augmentations. If FVP used a different backbone, training schedule, or evaluation protocol, the comparison in Tables I-III is not fully controlled. Please either reimplement FVP under the identical protocol or list the protocol differences and their expected effect on the comparison; also report the chosen batch sizes for all reimplemented baselines and the validation criterion used to select them.
minor comments (6)
  1. [Abstract, Section I] The phrase 'for the first time explore the potentials of Segment Anything Model for SFDA' should be softened, since the related-work section cites auto-prompting adaptation methods such as MedSAM-U, and the novelty claim is not necessary for the paper's contribution.
  2. [IV-C1, Eq. (4)] In Eq. (4), the condition 'for all i' in I_j' is too strong if read literally, because a candidate pixel i would need to satisfy both conditions with every pixel already in I_j rather than with at least one neighboring pixel; please clarify the intended neighbor criterion.
  3. [V-C7, Fig. 7] The failure-case examples in Fig. 7 are useful, but the figure would be easier to interpret if the final selected box prompt and the ground-truth box prompt were drawn with distinct line styles in a single panel; currently the blue search boxes and the red ground-truth box are difficult to distinguish in the middle row.
  4. [V-A3, Fig. 6] The temperature kappa is set to 10 for the prostate dataset, but its sensitivity is not included in Fig. 6; please add an ablation or state the range over which the final prostate result is stable.
  5. [V-A4] Please specify how 3D Dice and ASSD are computed (e.g., per-volume averaging and whether slices without foreground are excluded) so that the comparison with FVP and ProtoContra, which may use different averaging conventions, is unambiguous.
  6. [IV-D, Table VII] The connectivity-based post-processing assumes each foreground class is a single connected component; this is reasonable for the tested organs, but the manuscript should state the limitation for anatomies with multiple disconnected parts, since CP could remove true positive regions.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: DFG's reported Dice/ASSD gains are evaluated on external, held-out target-domain test labels, and the core prompt-search mechanism relies on an empirical property of MedSAM rather than on the paper's own fitted outputs.

full rationale

The paper's central claim is a measured performance comparison, not a derivation from its own definitions. DFG's pseudo-labels are produced by feeding automatically searched bounding boxes to the frozen external model MedSAM, and the reported Dice/ASSD numbers in Tables I-III are computed against unseen target-domain ground-truth labels according to standard train/test splits (e.g., 'randomly divide BTCV and CHAOS into training and test sets with a ratio of 4:1' in Section V-A1). The stable-interval property, 'MedSAM prediction keeps stable when the box prompt fluctuates near the ground truth' (Fig. 1, Section IV-C1), is presented as an empirical observation about MedSAM's behavior, not as a theorem proved from the method's own assumptions. The feature aggregation loss is explicitly borrowed from previously published work ([16], [35]), and the 'feature affinity property' is attributed to earlier external studies ([33], [9], [34]); none of these citations is a self-citation that supplies the paper's central conclusion. The only overlapping-author reference is [25], which appears in the related-work enumeration of pseudo-labeling methods and carries no load in the proposed framework. The heuristic thresholds (tau_f, tau_max, p_Delta, r, tau_Delta, tau_Div) are empirically chosen hyperparameters, which raises robustness and tuning concerns but does not make the benchmark outcome an input to the derivation by construction. In particular, no equation in the paper is equivalent to the reported Dice/ASSD numbers, and no predicted quantity is defined in terms of the target metric it is claimed to explain. The paper also explicitly discloses failure modes of its stable-interval detector in Section V-C7 and Fig. 7, which further confirms that the mechanism is an empirical heuristic with known limitations rather than a tautological construction. Therefore, there is no significant circularity.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The method depends on empirical properties of the source model and MedSAM feature spaces, plus a set of tuned thresholds. No free parameters are fitted in the sense of a mathematical derivation; the paper's claim is a measured benchmark improvement, so the ledger is dominated by domain assumptions rather than parameter fitting. The values listed under free_parameters are hand-set or tuned constants that are load-bearing for the reported numbers.

free parameters (8)
  • tau_f (feature similarity threshold) = 0.99
    Equation 4, propagation over target model features; set to 0.99 across all datasets; ablation Fig. 6(a) shows 0.99 or 0.999 best.
  • tau_Div (distance threshold coefficient) = 2.5
    Equation 7, propagation over MedSAM features; set across all datasets; Fig. 6(f) shows 2.5 optimal.
  • tau_max (upper bound on MBS distance threshold) = 0.35 abdominal, 0.3 prostate
    Equation 7; tuned per dataset group; Fig. 6(b) shows best range 0.31 to 0.39.
  • p_Delta (initial probability threshold) = 0.005
    Equation 3; empirically set; Fig. 6(c) shows 0.005 optimal.
  • r (propagation step size) = 4
    Equations 4 and 7; Fig. 6(d) shows best at r=3 or 4.
  • tau_Delta (stable interval threshold) = 15, 30, 45 for spans 1, 2, 3
    Equation 5; Fig. 6(e) shows best tau_Delta between 15 and 30.
  • kappa (temperature in feature aggregation loss) = 1 abdominal, 10 prostate
    Equation 1; set in Section V-A3.
  • m (bounding box margin)
    Defined in Section IV-C1; no numeric value is reported, leaving an unstated degree of freedom for replication.
assumptions (5)
  • domain assumption Feature affinity property: same-class target features produced by the source model stay close under domain shift.
    Section III and Fig. 2(a). This motivates propagation over target features in Eq. 4. Only illustrated by t-SNE, not quantitatively validated.
  • domain assumption MedSAM output stability near ground-truth box prompts.
    Section IV-C and Fig. 1(e). Used as termination criterion via Eq. 5, and contradicted by some failure cases in Fig. 7.
  • domain assumption MedSAM features separate foreground from background in target images.
    Section III and Section IV-C2. Used to propagate over MedSAM feature space via Eq. 7; the paper notes MedSAM lacks category semantics.
  • domain assumption Organs are spatially connected in each slice.
    Section IV-D justifies keeping only the largest connected component per class; may fail for fragmented structures.
  • domain assumption Frozen source classifier weights are plausible class prototypes.
    Section IV-B follows [16], [35] and uses c_k from H^s as prototypes in Eq. 2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Leveraging Segment Anything Model for Source-Free Domain Adaptation via Dual Feature Guided Auto-Prompting." pith.science (2026). https://pith.science/paper/JF7AJCJ3

@misc{pith2026250508527,
  author       = {Pith},
  title        = {Pith review of: Leveraging Segment Anything Model for Source-Free Domain Adaptation via Dual Feature Guided Auto-Prompting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JF7AJCJ3}},
  note         = {Machine review of arXiv:2505.08527}
}
read the original abstract

Source-free domain adaptation (SFDA) for segmentation aims at adapting a model trained in the source domain to perform well in the target domain with only the source model and unlabeled target data. Inspired by the recent success of Segment Anything Model (SAM) which exhibits the generality of segmenting images of various modalities and in different domains given human-annotated prompts like bounding boxes or points, we for the first time explore the potentials of Segment Anything Model for SFDA via automatedly finding an accurate bounding box prompt. We find that the bounding boxes directly generated with existing SFDA approaches are defective due to the domain gap. To tackle this issue, we propose a novel Dual Feature Guided (DFG) auto-prompting approach to search for the box prompt. Specifically, the source model is first trained in a feature aggregation phase, which not only preliminarily adapts the source model to the target domain but also builds a feature distribution well-prepared for box prompt search. In the second phase, based on two feature distribution observations, we gradually expand the box prompt with the guidance of the target model feature and the SAM feature to handle the class-wise clustered target features and the class-wise dispersed target features, respectively. To remove the potentially enlarged false positive regions caused by the over-confident prediction of the target model, the refined pseudo-labels produced by SAM are further postprocessed based on connectivity analysis. Experiments on 3D and 2D datasets indicate that our approach yields superior performance compared to conventional methods. Code is available at https://github.com/xmed-lab/DFG.

Figures

Figures reproduced from arXiv: 2505.08527 by the authors.

Figure 1
Figure 1. Take spleen in a target domain image in MRI→CT adaptation as an example. (a) MedSAM [21] requires an accurate bounding box prompt. Neither a too-small nor a too-large bounding box leads to a decent segmentation result. (b) Segmentation results of [16] and the corresponding bounding boxes, produced by different output probability thresholds. Due to the domain gap and limited knowledge from source model and target unl… view at source ↗
Figure 2
Figure 2. T-SNE visualization of pixel features. (a) The features of a target domain image produced by the source model (left) and the target model after the feature aggregation phase (right). (b) The features of a target domain image produced by MedSAM. and feature distribution property from those of the foundation models for segmentation. Therefore, these methods can not be easily extended to SFDA for medical image segmenta… view at source ↗
Figure 3
Figure 3. Overview of our proposed Dual Feature Guided auto-prompting (DFG) framework. Feature aggregation (FA) forces target features to form clusters and prepares for box prompt search. Target model feature guided box prompt search (TBS) and MedSAM feature guided box prompt search (MBS) find an accurate box prompt via propagation in the feature spaces to obtain refined pseudo-labels. Finally, the target model is trained wit… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison of different SFDA methods. The top two rows are segmentation results of BTCV images in CHAOS→BTCV. The 3rd and 4th rows are segmentation results of CHAOS images in BTCV→CHAOS. The 5th row is segmentation results of Curvas image in CHAOS→Curvas. T…
Figure 5
Figure 5. Figure 5: Visualization of box prompt search and pseudo-labels. From left to right: Target images; Ground truth; The predictions of target model after feature aggregation; The features produced by target model after feature aggregation; The procedure of box prompt search, with t…
Figure 6
Figure 6. Figure 6: Performance of our method with different hyper-parameter values. Dice scores are given. Spleen R . kidney L. kidney Liver Stable interval Image Ground truth Box prompt search MedSAM pred. change Our box prompt GT box prompt [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Examples of failure cases of box prompt search. From left to right: Target images; Ground truth; Ground truth box prompt (red box), and box prompt search procedure (blue boxes), with boxes changing from small to large (For clarity, only the first, the last, and part of…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 35 canonical work pages

  1. [1]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. Springer, 2015, pp. 234–241

  2. [2]

    Medical image segmentation using deep learning: A survey,

    R. Wang, T. Lei, R. Cui, B. Zhang, H. Meng, and A. K. Nandi, “Medical image segmentation using deep learning: A survey,” IET Image Proc. , vol. 16, no. 5, pp. 1243–1267, 2022

  3. [3]

    A-eval: A benchmark for cross-dataset and cross- modality evaluation of abdominal multi-organ segmentation,

    Z. Huang, Z. Deng, J. Ye, H. Wang, Y . Su, T. Li, H. Sun, J. Cheng, J. Chen, J. He et al., “A-eval: A benchmark for cross-dataset and cross- modality evaluation of abdominal multi-organ segmentation,” Medical Image Analysis , p. 103499, 2025

  4. [4]

    Patch-based output space adversarial learning for joint optic disc and cup segmentation,

    S. Wang, L. Yu, X. Yang, C.-W. Fu, and P.-A. Heng, “Patch-based output space adversarial learning for joint optic disc and cup segmentation,” IEEE Trans. Med. Imaging , vol. 38, no. 11, pp. 2485–2495, 2019

  5. [5]

    Do we really need to access the source data? source hypothesis transfer for unsupervised domain adaptation,

    J. Liang, D. Hu, and J. Feng, “Do we really need to access the source data? source hypothesis transfer for unsupervised domain adaptation,” in Proc. Int. Conf. Mach. Learn. PMLR, 2020, pp. 6028–6039

  6. [6]

    Source-free domain adaptation via distribution estimation,

    N. Ding, Y . Xu, Y . Tang, C. Xu, Y . Wang, and D. Tao, “Source-free domain adaptation via distribution estimation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. , 2022, pp. 7212–7222

  7. [7]

    Uncertainty reduction for model adaptation in semantic segmentation,

    P. T. Sivaprasad and F. Fleuret, “Uncertainty reduction for model adaptation in semantic segmentation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., no. CONF. IEEE, 2021, pp. 9608–9618

  8. [8]

    Uncertainty-guided source-free domain adaptation,

    S. Roy, M. Trapp, A. Pilzer, J. Kannala, N. Sebe, E. Ricci, and A. Solin, “Uncertainty-guided source-free domain adaptation,” in Proc. Eur . Conf. Comput. Vis. Springer, 2022, pp. 537–555

Show all 43 references
  1. [9]

    Class relationship embedded learning for source-free unsupervised domain adaptation,

    Y . Zhang, Z. Wang, and W. He, “Class relationship embedded learning for source-free unsupervised domain adaptation,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit. , 2023, pp. 7619–7629

  2. [10]

    Source-free domain adaptive fundus image segmentation with denoised pseudo-labeling,

    C. Chen, Q. Liu, Y . Jin, Q. Dou, and P.-A. Heng, “Source-free domain adaptive fundus image segmentation with denoised pseudo-labeling,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. Springer, 2021, pp. 225–235

  3. [11]

    Denoising for relaxing: Unsupervised domain adaptive fundus image segmentation without source data,

    Z. Xu, D. Lu, Y . Wang, J. Luo, D. Wei, Y . Zheng, and R. K.-y. Tong, “Denoising for relaxing: Unsupervised domain adaptive fundus image segmentation without source data,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. Springer, 2022, pp. 214–224

  4. [12]

    Upl-sfda: Uncertainty-aware pseudo label guided source-free domain adaptation for medical image segmentation,

    J. Wu, G. Wang, R. Gu, T. Lu, Y . Chen, W. Zhu, T. Vercauteren, S. Ourselin, and S. Zhang, “Upl-sfda: Uncertainty-aware pseudo label guided source-free domain adaptation for medical image segmentation,” IEEE Trans. Med. Imaging , 2023

  5. [13]

    Source free domain adaptation for medical image segmentation with fourier style mining,

    C. Yang, X. Guo, Z. Chen, and Y . Yuan, “Source free domain adaptation for medical image segmentation with fourier style mining,” Med. Image Anal., vol. 79, p. 102457, 2022

  6. [14]

    Fvp: Fourier visual prompting for source-free unsupervised domain adaptation of medical image segmentation,

    Y . Wang, J. Cheng, Y . Chen, S. Shao, L. Zhu, Z. Wu, T. Liu, and H. Zhu, “Fvp: Fourier visual prompting for source-free unsupervised domain adaptation of medical image segmentation,” IEEE Trans. Med. Imaging, 2023

  7. [15]

    Source- free domain adaptation for image segmentation,

    M. Bateson, H. Kervadec, J. Dolz, H. Lombaert, and I. B. Ayed, “Source- free domain adaptation for image segmentation,” Med. Image Anal. , vol. 82, p. 102617, 2022

  8. [16]

    Source- free domain adaptation for medical image segmentation via prototype- anchored feature alignment and contrastive learning,

    Q. Yu, N. Xi, J. Yuan, Z. Zhou, K. Dang, and X. Ding, “Source- free domain adaptation for medical image segmentation via prototype- anchored feature alignment and contrastive learning,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. , 2023, pp. 3–12

  9. [17]

    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 Proc. IEEE Int. Conf. Comput. Vis. , 2023, pp. 4015–4026

  10. [18]

    Customized segment anything model for medical image segmentation,

    K. Zhang and D. Liu, “Customized segment anything model for medical image segmentation,” arXiv preprint arXiv:2304.13785 , 2023

  11. [19]

    Ma-sam: Modality-agnostic sam adaptation for 3d medical image segmentation,

    C. Chen, J. Miao, D. Wu, Z. Yan, S. Kim, J. Hu, A. Zhong, Z. Liu, L. Sun, X. Li et al., “Ma-sam: Modality-agnostic sam adaptation for 3d medical image segmentation,” arXiv preprint arXiv:2309.08842 , 2023

  12. [20]

    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

  13. [21]

    Segment anything in medical images,

    J. Ma, Y . He, F. Li, L. Han, C. You, and B. Wang, “Segment anything in medical images,” Nat. Commun. , vol. 15, no. 1, p. 654, 2024

  14. [22]

    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

  15. [23]

    Beyond adapting sam: Towards end-to-end ultrasound image segmentation via auto prompting,

    X. Lin, Y . Xiang, L. Yu, and Z. Yan, “Beyond adapting sam: Towards end-to-end ultrasound image segmentation via auto prompting,” in International Conference on Medical Image Computing and Computer- Assisted Intervention . Springer, 2024, pp. 24–34

  16. [24]

    Medsam-u: Uncertainty-guided auto multi-prompt adaptation for reliable medsam,

    N. Zhou, K. Zou, K. Ren, M. Luo, L. He, M. Wang, Y . Chen, Y . Zhang, H. Chen, and H. Fu, “Medsam-u: Uncertainty-guided auto multi-prompt adaptation for reliable medsam,” arXiv preprint arXiv:2409.00924, 2024

  17. [25]

    Source-free domain adaptive fundus image segmentation with class-balanced mean teacher,

    L. Tang, K. Li, C. He, Y . Zhang, and X. Li, “Source-free domain adaptive fundus image segmentation with class-balanced mean teacher,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. Springer, 2023, pp. 684–694

  18. [26]

    Adapting off-the- shelf source segmenter for target medical image segmentation,

    X. Liu, F. Xing, C. Yang, G. El Fakhri, and J. Woo, “Adapting off-the- shelf source segmenter for target medical image segmentation,” in Proc. Int. Conf. Med. Image Comput. Comput.-Assist. Intervent. Springer, 2021, pp. 549–559

  19. [27]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark et al. , “Learning transferable visual models from natural language supervision,” in Proc. Int. Conf. Mach. Learn. PMLR, 2021, pp. 8748–8763

  20. [28]

    Scaling up visual and vision-language representation learning with noisy text supervision,

    C. Jia, Y . Yang, Y . Xia, Y .-T. Chen, Z. Parekh, H. Pham, Q. Le, Y .- H. Sung, Z. Li, and T. Duerig, “Scaling up visual and vision-language representation learning with noisy text supervision,” in Proc. Int. Conf. Mach. Learn. PMLR, 2021, pp. 4904–4916

  21. [29]

    Seggpt: Segmenting everything in context,

    X. Wang, X. Zhang, Y . Cao, W. Wang, C. Shen, and T. Huang, “Seggpt: Segmenting everything in context,” arXiv preprint arXiv:2304.03284 , 2023

  22. [30]

    Rethinking the role of pre-trained networks in source-free domain adaptation,

    W. Zhang, L. Shen, and C.-S. Foo, “Rethinking the role of pre-trained networks in source-free domain adaptation,” in Proc. IEEE Int. Conf. Comput. Vis., 2023, pp. 18 841–18 851

  23. [31]

    The unreasonable effectiveness of large language-vision models for source- free video domain adaptation,

    G. Zara, A. Conti, S. Roy, S. Lathuili `ere, P. Rota, and E. Ricci, “The unreasonable effectiveness of large language-vision models for source- free video domain adaptation,” in Proc. IEEE Int. Conf. Comput. Vis. , 2023, pp. 10 307–10 317

  24. [32]

    Source-free domain adaptation with frozen multimodal foundation model,

    S. Tang, W. Su, M. Ye, and X. Zhu, “Source-free domain adaptation with frozen multimodal foundation model,” arXiv preprint, 2023

  25. [33]

    Exploiting the intrinsic neighborhood structure for source-free domain adaptation,

    S. Yang, J. van de Weijer, L. Herranz, S. Jui et al. , “Exploiting the intrinsic neighborhood structure for source-free domain adaptation,” in Proc. Adv. Neural Inf. Process. Syst. , vol. 34, 2021, pp. 29 393–29 405

  26. [34]

    Bmd: A general class-balanced multicentric dynamic prototype strategy for source-free domain adaptation,

    S. Qu, G. Chen, J. Zhang, Z. Li, W. He, and D. Tao, “Bmd: A general class-balanced multicentric dynamic prototype strategy for source-free domain adaptation,” arXiv preprint arXiv:2204.02811 , 2022

  27. [35]

    Exploiting chain rule and bayes’ theorem to compare probability distributions,

    H. Zheng and M. Zhou, “Exploiting chain rule and bayes’ theorem to compare probability distributions,” in Proc. Adv. Neural Inf. Process. Syst., vol. 34, 2021, pp. 14 993–15 006

  28. [36]

    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

  29. [37]

    Chaos challenge- combined (ct-mr) healthy abdominal organ segmentation,

    A. E. Kavur, N. S. Gezer, M. Barıs ¸, S. Aslan, P.-H. Conze, V . Groza, D. D. Pham, S. Chatterjee, P. Ernst, S. ¨Ozkan et al. , “Chaos challenge- combined (ct-mr) healthy abdominal organ segmentation,” Med. Image Anal., vol. 69, p. 101950, 2021

  30. [38]

    Miccai calibration and uncertainty for multirater volume assessment in multiorgan segmentation challenge,

    M. Riera i Mar ´ın, J. Garc ´ıa L ´opez, J. Kleiss, S. O K, A. Galdr ´an, M. May, M. A. Gonz ´alez-Ballester, J. Rodr ´ıguez Comas, M. chmidt, C. Hessman, A. Aubanell, and A. Antol ´ın, “Miccai calibration and uncertainty for multirater volume assessment in multiorgan segmenta...

  31. [39]

    Domain adaptation meets zero-shot learning: an annotation-efficient approach to multi-modality medical image segmentation,

    C. Bian, C. Yuan, K. Ma, S. Yu, D. Wei, and Y . Zheng, “Domain adaptation meets zero-shot learning: an annotation-efficient approach to multi-modality medical image segmentation,” IEEE Trans. Med. Imaging, vol. 41, no. 5, pp. 1043–1056, 2021

  32. [40]

    Unsupervised bidirectional cross-modality adaptation via deeply synergistic image and feature alignment for medical image segmentation,

    C. Chen, Q. Dou, H. Chen, J. Qin, and P. A. Heng, “Unsupervised bidirectional cross-modality adaptation via deeply synergistic image and feature alignment for medical image segmentation,” IEEE Trans. Med. Imaging, vol. 39, no. 7, pp. 2494–2505, 2020

  33. [41]

    Nci- isbi 2013 challenge: Automated segmentation of prostate structures,

    N. Bloch, A. Madabhushi, H. Huisman, J. Freymann, J. Kirby, M. Grauer, A. Enquobahrie, C. Jaffe, L. Clarke, and K. Farahani, “Nci- isbi 2013 challenge: Automated segmentation of prostate structures,” The Cancer Imaging Archive , 2015

  34. [42]

    Ms-net: Multi-site network for improving prostate segmentation with heterogeneous mri data,

    Q. Liu, Q. Dou, L. Yu, and P. A. Heng, “Ms-net: Multi-site network for improving prostate segmentation with heterogeneous mri data,” IEEE Trans. Med. Imaging , 2020

  35. [43]

    Variability of manual segmen- tation of the prostate in axial t2-weighted mri: a multi-reader study,

    A. S. Becker, K. Chaitanya, K. Schawkat, U. J. Muehlematter, A. M. H¨otker, E. Konukoglu, and O. F. Donati, “Variability of manual segmen- tation of the prostate in axial t2-weighted mri: a multi-reader study,” Eur . J. Radiol., vol. 121, p. 108716, 2019

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.