Pith. sign in

REVIEW 3 major objections 5 minor 67 references

SRPL-SFDA: SAM-Guided Reliable Pseudo-Labels for Source-Free Domain Adaptation in Medical Image Segmentation

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A SAM-guided pseudo-label pipeline for source-free medical image segmentation reaches Dice within about one point of supervised target-domain training on two multi-site MRI benchmarks.

desk verdict A solid empirical SFDA recipe with a disclosed but real target-validation leak in model selection; worth refereeing, but the 'close to supervised' claim needs a fully unsupervised protocol. read the letter →

arxiv 2506.09403 v1 pith:GHNY5FQ4 submitted 2025-06-11 cs.CV

Xinya Liu , Jianghao Wu , Tao Lu , Shaoting Zhang , Guotai Wang This is my paper
classification cs.CV
keywords source-freedomainadaptationmedicalimagesegmentationSegmentAnythingModelpseudo-labelrefinementtest-timeintensityenhancementconsistency-basedreliabilityMRIshiftreliability-awaretraining
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 claims that a frozen Segment Anything Model (SAM) can be turned into a reliable pseudo-label refiner for source-free domain adaptation (SFDA) in medical image segmentation, so that adapting a pretrained model to a new clinical site needs neither the source images nor target annotations. The method applies three intensity transforms to each target image, uses their averaged source-model predictions to draw a bounding box, and sends the three enhanced copies concatenated as an RGB-like image to SAM for refined boundaries. A consistency check over three SAM outputs selects a reliable pixel region, and a reliability-aware loss trains the target model there while entropy-minimizing the remaining ambiguous regions. On prostate and fetal-brain MRI benchmarks the adapted model reaches 82.22% and 94.33% Dice, compared with 83.02% and 95.53% for training directly on labeled target data. The authors note that their checkpoint selection uses a labeled target-domain validation set, a practical constraint acknowledged in the paper.

What carries the argument

The load-bearing mechanism is the T3IE triple used twice. For each target image, histogram equalization, domain-adaptive gamma correction (aligning image mean to dataset mean), and SAM-compatible gamma correction (pushing pixel statistics toward natural-image mean 0.5 and standard deviation 0.29) produce three enhanced copies. Averaged source-model predictions from these copies yield the initial pseudo-label and the box prompt; the copies concatenated as $X_{\text{RGB}}$ yield SAM's refined mask. The same three copies, run through SAM separately with an identical box, produce the CMSO agreement map: a pixel is reliable ($\Omega_C$) exactly when $R_{\text{He}}=R_{\gamma_D}=R_{\gamma_S}$, and unreliable otherwise. The RPSR loss completes the training: partial cross-entropy plus partial Dice on $\Omega_C$, and partial entropy minimization on $\Omega_U$ with weight $\lambda=10$.

What would settle it

On a held-out target test set with ground truth, compute the Dice of the consensus region $\Omega_C$ against the true mask while also recomputing SRPL-SFDA with the box prompt $B$ shifted by 10 pixels: if $\Omega_C$ is not more accurate than the full refined pseudo-label $R$, or if the 'reliable' labels move with the box, the CMSO reliability claim is falsified.

Watch

Extended reading notes

Core claim

The central claim is that SAM's zero-shot, prompt-based segmentation, although trained on natural images, can correct medical pseudo-labels tightly enough that source-free adaptation nearly matches supervised target training. The source model's averaged predictions over the three T3IE-enhanced copies of a target image give an initial pseudo-label; its bounding box becomes the prompt for SAM, and the same three enhanced copies concatenated into a pseudo-RGB image become SAM's input. The refined mask $R$ is then filtered by consistency: each enhanced copy is sent to SAM separately with the same box, and only pixels $\Omega_C$ where all three outputs agree are used as supervision, through a combination of partial cross-entropy and partial Dice, while the remaining pixels $\Omega_U$ are regularized by entropy minimization. With this recipe the method reaches 82.22% Dice on prostate sites D/E/F versus 83.02% for target-only supervised training and 94.33% on fetal brain versus 95.53%, and it reports lower average surface distance than the prior SFDA methods on the prostate benchmark.

Load-bearing premise

The method assumes that when the three SAM runs agree on a pixel, that pixel is correct, but all three runs share the same bounding-box prompt derived from the source model's pseudo-label, so agreement can certify shared bias rather than anatomical truth.

Editorial extensions

If this is right

  • A clinic deploying a segmentation model could adapt it to a new scanner using only unlabeled target scans, avoiding the transfer of patient data from the original site.
  • Future promptable segmenters can be dropped into the same recipe without retraining, since SAM itself is used frozen and only the target segmentation model is updated.
  • The threshold-free consensus rule means reliability estimation does not need a tuned confidence cutoff, making the method less sensitive to hyperparameter choice.
  • Because boundary-sensitive losses run only on the consensus region, the adapted model can learn sharper edges than the source model provides, which matters for contour-based clinical tasks.

Reading between the lines

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

  • Beyond the paper: the same consensus-based selection could be applied to other promptable segmenters such as medical-SAM variants, with only the input-transform statistics needing recalibration.
  • Beyond the paper: the SAM-compatible gamma target of mean 0.5 and standard deviation 0.29 is tuned to natural images; on CT, PET, or ultrasound the same T3IE idea would likely need modality-specific reference statistics.
  • Beyond the paper: since the paper selects checkpoints with a labeled target validation set, a deployment-ready variant would need a label-free selection proxy, such as consensus-region entropy or agreement on the unlabeled training set.
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

3 major / 5 minor

Summary. The paper proposes SRPL-SFDA, a source-free domain adaptation method for medical image segmentation. The method has three main components: Test-Time Tri-branch Intensity Enhancement (T3IE), which applies histogram equalization and two gamma corrections to target images before inference; SAM-guided pseudo-label refinement, where the source model's averaged T3IE prediction is used to form a bounding-box prompt and the three T3IE channels are concatenated as an SAM-compatible RGB input; and Consistency of Multiple SAM Outputs (CMSO), which identifies reliable pseudo-label pixels as those where the three SAM outputs agree, followed by a Reliability-Aware Pseudo-Label Supervision and Regularization (RPSR) loss. Experiments are reported on two MRI datasets, prostate (sites A/B to D/E/F, plus site C) and fetal brain (TrueFISP to HASTE). The reported results show that SRPL-SFDA outperforms four SFDA baselines and approaches the Dice of supervised target-only training and fine-tuning, with ablations on pseudo-label quality, loss components, hyper-parameter lambda, and SAM prompt types. The code is publicly available.

Significance. If the reported numbers are valid under a genuinely source-free protocol, the paper makes a useful empirical contribution: it demonstrates a concrete recipe for improving pseudo-labels with a frozen SAM through intensity enhancement and consistency-based reliable-region mining, and the ablation studies are internally coherent and informative. The release of code and the model-agnostic framing are additional strengths. However, the headline claim of approaching supervised performance is currently conditional on using target-domain validation labels for hyper-parameter and checkpoint selection, which is disclosed in Section 4.2. This weakens the source-free claim as stated and must be addressed before the result can be taken at face value.

major comments (3)
  1. [Sec. 4.2] The implementation details state: 'According to the best performance on the validation set of the target domain, the hyper-parameter setting was lambda=10.0, and the corresponding checkpoint was employed for inference.' This uses target-domain labels for both hyper-parameter tuning and model selection, which is not a source-free protocol. The abstract's claim that SRPL-SFDA 'is close to that of supervised training in the target domain' is therefore an upper bound under oracle selection, not an established result for a method that never sees target labels. I request that the authors either (a) re-run the main comparisons without any target-label-based selection, for example by fixing lambda via the source validation set or by reporting the last-epoch model and the sensitivity of test Dice to lambda, or (b) clearly reframe all headline results as 'oracle-selected' and remove or qualify the comparison to supervised training. This is load-bearing because it concerns the central claim of the paper.
  2. [Sec. 3.2.1, Eq. (7)] The CMSO reliability criterion assumes that agreement among R_He, R_gammaD, and R_gammaS certifies correctness. However, all three SAM outputs are produced from the same bounding-box prompt B, which is derived from the same source-model pseudo-label Y. A biased or incomplete box prompt is therefore shared by all three outputs, so agreement can reflect consistency rather than correctness. The Discussion acknowledges this possibility but provides no quantitative evidence that the consensus region is actually more accurate than the non-consensus region. Please add an analysis on the target validation set that compares segmentation accuracy inside Omega_C versus Omega_U, or an experiment that perturbs the box prompt and shows that consensus remains correlated with correctness. Without this, the RPSR supervision in Eq. (13) may be training on trusted noise in precisely the way the method is designed to avoid.
  3. [Sec. 4.2, Table 1] The target validation sets are very small: 4 volumes for prostate sites D/E/F and 5 volumes for fetal brain. Selecting lambda and a checkpoint from 300 epochs on such small sets risks overfitting the validation set, and the reported test Dice may therefore be inflated relative to what a truly unsupervised protocol would achieve. The paper reports a single random split with no repeated-seed or repeated-split variability. Please report results across multiple data splits or random seeds, or at least show that the test performance is flat around lambda=10.0 and that checkpoint selection is not a sharp peak on the validation curve. This is important for assessing whether the reported gains over baselines are robust rather than an artifact of small-sample oracle selection.
minor comments (5)
  1. [Sec. 4.4.4] The text at the start of this subsection is corrupted: 'We conducThese values represent the overall distributioe performance' is an incomplete and duplicated fragment that makes the prompt-type ablation description unreadable. Please rewrite this paragraph.
  2. [Tables 2, 3, and 5] The final method in Table 5 reports Dice of 84.46% on prostate and 92.58% on fetal brain, whereas the same method in Tables 2 and 3 reports 82.22% and 94.33%. It appears Table 5 is evaluated on the target validation set while Tables 2 and 3 are on the test set, but this is not stated in the captions or text. Please clarify the evaluation protocol for each table, as the discrepancy is currently confusing.
  3. [Sec. 4.3] The 'SAM(X)-BB prompt' baseline uses ground-truth bounding boxes expanded by 5-10 pixels to simulate user interaction. This is an oracle-like upper reference for SAM's prompt-based capability rather than an SFDA baseline, and the comparison should be described as such: it shows how much of the gain comes from the pseudo-label-derived box rather than from user-provided box prompts.
  4. [Sec. 4.3] The claim that SRPL-SFDA is 'close to supervised training' is based on Dice gaps of 0.80-3.08 percentage points, but no significance test is reported against 'Target only' or 'Fine-tune.' Please either add such tests or soften the wording to describe the observed Dice gap without implying statistical equivalence.
  5. [Sec. 2.2] The related work discusses DPL, CLR, and CCMT but the experimental comparison includes only PTBN, TENT, AdaMI, and UPL-SFDA. Please state explicitly why the additional pseudo-label-based SFDA methods are not included in the comparison, or add them if feasible.

Circularity Check

1 steps flagged · score 2.0 of 10

No classic derivation circularity; CMSO's shared box prompt creates a self-referential reliability criterion, and target-validation selection is a disclosed protocol risk rather than a circular step.

  1. other [Sec. 3.1.3 and Sec. 3.2.1 (Eq. 7), with the limitation acknowledged in Sec. 5]
    "Specifically, we send X_He, X_gammaD and X_gammaS to SAM respectively with the same bounding box prompt B derived from Y. ... The region with a consensus among the three outputs are treated as reliable part, which is denoted as Omega_C."

    The three SAM outputs in Eq. (7) are all generated with the same box prompt B, and B is itself computed from the raw pseudo-label Y (Sec. 3.1.3). Agreement among R_He, R_gammaD and R_gammaS therefore tests consistency of SAM under intensity perturbations while holding the prompt fixed; it cannot detect errors that originate in Y and propagate through B. The 'reliable' set Omega_C is thus defined as the set where a shared, Y-dependent prompt yields agreeing masks, so the reliability claim reduces to self-consistency of a single prompt chain rather than an independent check of label correctness. The paper's own Discussion attempts to justify this but does not break the dependence: all three outputs share the same potentially biased prompt.

full rationale

This is an empirical method paper, not a derivation paper, so the classic circularity failure modes largely do not apply. The method's core components—T3IE, SAM prompting from pseudo-label boxes, CMSO, and RPSR—are disclosed heuristics evaluated against external baselines and ablations. The self-extension of RPL-SFDA (Liu et al., 2024) is explicitly disclosed and is not load-bearing: the central novelty is the SAM-guided refinement, which is compared with external SFDA methods as well as with the prior RPL-SFDA. The only step with a genuine self-referential flavor is CMSO: the reliability region is defined as agreement among three SAM outputs that all share a box prompt derived from the same source-model pseudo-label Y. That is a shared-input confound, and the paper's defense in Sec. 5 does not remove it; however, it is an assumption about what consistency means, not a fitted parameter renamed as a prediction or a uniqueness theorem imported from the authors' prior work. Separately, the paper acknowledges using a labeled target validation set for hyperparameter and checkpoint selection (Sec. 4.2), which is a protocol limitation and a threat to the 'close to supervised training' claim, but it is not circularity in the derivation-chain sense. Overall, the reported gains are empirically contingent and the central claim has independent content, so the appropriate circularity score is low.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The method depends on one target-tuned hyperparameter, lambda, plus a heuristic box-prompt margin and externally adopted natural-image statistics. The main domain assumptions are that source-model box prompts are good enough for SAM, that consistency across SAM outputs implies correctness, and that T3IE preserves anatomy. No new physical or ontological entities are introduced; T3IE, CMSO, and RPSR are algorithmic components rather than posited entities.

free parameters (3)
  • lambda (weight of entropy minimization, Eq. 13) = 10.0
    Chosen as best on target-domain validation sets in Section 4.4.3, i.e., selected using target labels; affects the balance between reliable pseudo-label supervision and entropy regularization.
  • SAM box-prompt expansion margin = 5-10 pixels
    Heuristic margin applied to the bounding box from the source pseudo-label, Section 4.3 and Table 6; not tuned systematically.
  • Natural-image target mean and std for gamma correction = mu=0.5, sigma=0.29
    Adopted in Section 3.1.1 from Sener et al. 2022; used to set SAM-compatible gamma correction and the three-channel SAM input. These are external constants, not fitted to the target data, but they are not universal constants either.
assumptions (5)
  • domain assumption Source model pseudo-labels, after T3IE averaging, provide bounding-box prompts accurate enough for SAM to refine rather than mislead.
    Section 3.1.3: SAM refinement uses a box prompt from the source pseudo-label Y; if Y is too poor, the prompt is wrong and refinement fails.
  • domain assumption Agreement among three SAM outputs under T3IE perturbations identifies semantically correct reliable regions.
    Section 3.2.1, Eq. 7: CMSO labels any pixel where R_He, R_gammaD, and R_gammaS agree as reliable; the same box prompt B is used for all three, so agreement can encode shared prompt bias.
  • domain assumption T3IE transformations preserve segmentation-relevant structures while reducing the medical-to-natural domain gap.
    Section 3.1.1: histogram equalization and two gamma corrections are assumed to increase contrast without distorting anatomy.
  • ad hoc to paper A labeled target-domain validation set may be used for checkpoint selection and hyperparameter tuning.
    Section 4.2 and Discussion: the adapted model checkpoint is chosen by best target validation performance; this is outside the unlabeled SFDA protocol but is disclosed.
  • domain assumption Natural-image pixel statistics (mean 0.5, std 0.29) are representative for aligning medical images to SAM's training distribution.
    Section 3.1.1, Eq. 3; cited to Sener et al. 2022, whose relevance to global natural-image statistics is unclear.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SRPL-SFDA: SAM-Guided Reliable Pseudo-Labels for Source-Free Domain Adaptation in Medical Image Segmentation." pith.science (2026). https://pith.science/paper/GHNY5FQ4

@misc{pith2026250609403,
  author       = {Pith},
  title        = {Pith review of: SRPL-SFDA: SAM-Guided Reliable Pseudo-Labels for Source-Free Domain Adaptation in Medical Image Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GHNY5FQ4}},
  note         = {Machine review of arXiv:2506.09403}
}
read the original abstract

Domain Adaptation (DA) is crucial for robust deployment of medical image segmentation models when applied to new clinical centers with significant domain shifts. Source-Free Domain Adaptation (SFDA) is appealing as it can deal with privacy concerns and access constraints on source-domain data during adaptation to target-domain data. However, SFDA faces challenges such as insufficient supervision in the target domain with unlabeled images. In this work, we propose a Segment Anything Model (SAM)-guided Reliable Pseudo-Labels method for SFDA (SRPL-SFDA) with three key components: 1) Test-Time Tri-branch Intensity Enhancement (T3IE) that not only improves quality of raw pseudo-labels in the target domain, but also leads to SAM-compatible inputs with three channels to better leverage SAM's zero-shot inference ability for refining the pseudo-labels; 2) A reliable pseudo-label selection module that rejects low-quality pseudo-labels based on Consistency of Multiple SAM Outputs (CMSO) under input perturbations with T3IE; and 3) A reliability-aware training procedure in the unlabeled target domain where reliable pseudo-labels are used for supervision and unreliable parts are regularized by entropy minimization. Experiments conducted on two multi-domain medical image segmentation datasets for fetal brain and the prostate respectively demonstrate that: 1) SRPL-SFDA effectively enhances pseudo-label quality in the unlabeled target domain, and improves SFDA performance by leveraging the reliability-aware training; 2) SRPL-SFDA outperformed state-of-the-art SFDA methods, and its performance is close to that of supervised training in the target domain. The code of this work is available online: https://github.com/HiLab-git/SRPL-SFDA.

Figures

Figures reproduced from arXiv: 2506.09403 by the authors.

Figure 2
Figure 2. Visual comparison between different SFDA methods on the prostate and fetal brain segmentation tasks. rectly using SAM for user interaction-based inference. In addition, we found that the performance of our method decreased when training from scratch in the target do￾main, demonstrating the usefulness of source model for initialization. In terms of average ASSD, our method achieved the lowest value of 2.02 mm among t… view at source ↗
Figure 3
Figure 3. Visual comparison of pseudo-labels in the target domain obtained by di [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figure 4
Figure 4. Effect of different λ values on the validation sets of different target domains. ASSD metrics. For the Prostate dataset, SAM(X)-box achieves a Dice score of 78.32% and an ASSD of 2.94 mm, significantly better than those of SAM(X)-point and SAM(X)-mask. A similar trend is observed on the Fe￾tal brain dataset, where SAM(X)-box achieves a Dice score of 90.74% and an ASSD of 1.69 mm, compared to 81.62% and 10.37 mm for … view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

67 extracted references · 57 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  4. [4]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  5. [5]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize ":" * " " *...

  6. [6]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize ":" * " " *...

  7. [7]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize "" * " " * ...

  8. [8]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize ":" * " " *...

Show all 67 references
  1. [9]

    , author Kervadec, H

    author Bateson, M. , author Kervadec, H. , author Dolz, J. , author Lombaert, H. , & author Ayed, I. B. ( year 2022 ). title Source-free domain adaptation for image segmentation . journal Medical Image Analysis \/ , volume 82 \/ , pages 102617

    2022

  2. [10]

    , author Kervadec, H

    author Bateson, M. , author Kervadec, H. , author Dolz, J. , author Lombaert, H. , & author Ben Ayed, I. ( year 2020 ). title Source-relaxed domain adaptation for image segmentation . In booktitle MICCAI \/ (pp. pages 490--499 ). organization Springer

    2020

  3. [11]

    , author Yan, H

    author Cai, T. , author Yan, H. , author Ding, K. , author Zhang, Y. , & author Zhou, Y. ( year 2024 ). title WSPolyp-SAM : Weakly supervised and self-guided fine-tuning of SAM for colonoscopy polyp segmentation . journal Applied Sciences \/ , volume 14 \/ , pages 5007

    2024

  4. [12]

    , author Dou, Q

    author Chen, C. , author Dou, Q. , author Chen, H. , author Qin, J. , & author Heng, P. A. ( year 2020 ). title Unsupervised bidirectional cross-modality adaptation via deeply synergistic image and feature alignment for medical image segmentation . journal IEEE Transactions on...

    2020

  5. [13]

    , author Liu, Q

    author Chen, C. , author Liu, Q. , author Jin, Y. , author Dou, Q. , & author Heng, P.-A. ( year 2021 ). title Source-free domain adaptive fundus image segmentation with denoised pseudo-labeling . In booktitle MICCAI \/ (pp. pages 225--235 ). organization Springer

    2021

  6. [14]

    , author Miao, J

    author Chen, C. , author Miao, J. , author Wu, D. , author Zhong, A. , author Yan, Z. , author Kim, S. , author Hu, J. , author Liu, Z. , author Sun, L. , author Li, X. et al. ( year 2024 a ). title Ma-sam: Modality-agnostic sam adaptation for 3d medical image segmentation . j...

    2024

  7. [15]

    , author Mei, J

    author Chen, J. , author Mei, J. , author Li, X. , author Lu, Y. , author Yu, Q. , author Wei, Q. , author Luo, X. , author Xie, Y. , author Adeli, E. , author Wang, Y. et al. ( year 2024 b ). title TransUNet: Rethinking the U-Net architecture design for medical image segmenta...

    2024

  8. [16]

    , author Wei, Q

    author Cheng, Z. , author Wei, Q. , author Zhu, H. , author Wang, Y. , author Qu, L. , author Shao, W. , & author Zhou, Y. ( year 2024 ). title Unleashing the potential of SAM for medical adaptation via hierarchical decoding . In booktitle CVPR \/ (pp. pages 3511--3522 )

    2024

  9. [17]

    , author Fang, H

    author Diao, Y. , author Fang, H. , author Yu, H. , author Li, F. , & author Xu, Y. ( year 2025 ). title Multimodal invariant feature prompt network for brain tumor segmentation with missing modalities . journal Neurocomputing \/ , volume 616 \/ , pages 128847

    2025

  10. [18]

    , author Chen, Z.-M

    author Ge, Y. , author Chen, Z.-M. , author Zhang, G. , author Heidari, A. A. , author Chen, H. , & author Teng, S. ( year 2023 ). title Unsupervised domain adaptation via style adaptation and boundary enhancement for medical semantic segmentation . journal Neurocomputing \/ ,...

    2023

  11. [19]

    , author Zhong, Y

    author Gong, S. , author Zhong, Y. , author Ma, W. , author Li, J. , author Wang, Z. , author Zhang, J. , author Heng, P.-A. , & author Dou, Q. ( year 2024 ). title 3DSAM-adapter : Holistic adaptation of SAM from 2D to 3D for promptable tumor segmentation . journal Medical Ima...

    2024

  12. [20]

    , & author Liu, M

    author Guan, H. , & author Liu, M. ( year 2021 ). title Domain adaptation for medical image analysis: A survey . journal IEEE Transactions on Biomedical Engineering \/ , volume 69 \/ , pages 1173--1185

    2021

  13. [21]

    , author Qi, L

    author Han, X. , author Qi, L. , author Yu, Q. , author Zhou, Z. , author Zheng, Y. , author Shi, Y. , & author Gao, Y. ( year 2021 ). title Deep symmetric adaptation network for cross-modality medical image segmentation . journal IEEE Transactions on Medical Imaging \/ , volu...

    2021

  14. [22]

    , author Carass, A

    author He, Y. , author Carass, A. , author Zuo, L. , author Dewey, B. E. , & author Prince, J. L. ( year 2021 ). title Autoencoder based self-supervised test-time adaptation for medical image analysis . journal Medical Image Analysis \/ , volume 72 \/ , pages 102136

    2021

  15. [23]

    , & author Zhu, Y

    author Hu, X. , & author Zhu, Y. ( year 2023 ). title Dual frame-level and region-level alignment for unsupervised video domain adaptation . journal Neurocomputing \/ , volume 550 \/ , pages 126454

    2023

  16. [24]

    , author Yang, X

    author Huang, Y. , author Yang, X. , author Liu, L. , author Zhou, H. , author Chang, A. , author Zhou, X. , author Chen, R. , author Yu, J. , author Chen, J. , author Chen, C. et al. ( year 2024 ). title Segment anything model for medical images? journal Medical Image Analysi...

    2024

  17. [25]

    , author Erdil, E

    author Karani, N. , author Erdil, E. , author Chaitanya, K. , & author Konukoglu, E. ( year 2021 ). title Test-time adaptable neural networks for robust medical image segmentation . journal Medical Image Analysis \/ , volume 68 \/ , pages 101907

    2021

  18. [26]

    , author Mintun, E

    author Kirillov, A. , author Mintun, E. , author Ravi, N. , author Mao, H. , author Rolland, C. , author Gustafson, L. , author Xiao, T. , author Whitehead, S. , author Berg, A. C. , author Lo, W.-Y. et al. ( year 2023 ). title Segment anything . In booktitle CVPR \/ (pp. page...

    2023

  19. [27]

    , author Yu, Z

    author Li, J. , author Yu, Z. , author Du, Z. , author Zhu, L. , & author Shen, H. T. ( year 2024 ). title A comprehensive survey on source-free domain adaptation . journal IEEE Transactions on Pattern Analysis and Machine Intelligence \/ , volume 46 \/ , pages 5743 -- 5762

    2024

  20. [28]

    , author Hu, D

    author Liang, J. , author Hu, D. , author Feng, J. , & author He, R. ( year 2022 ). title Dine: Domain adaptation from single and multiple black-box predictors . In booktitle CVPR \/ (pp. pages 8003--8013 )

    2022

  21. [29]

    , author Hu, D

    author Liang, J. , author Hu, D. , author Wang, Y. , author He, R. , & author Feng, J. ( year 2021 ). title Source data-absent unsupervised domain adaptation through hypothesis transfer and labeling transfer . journal IEEE Transactions on Pattern Analysis and Machine Intellige...

    2021

  22. [30]

    , author Del Bue, A

    author Litrico, M. , author Del Bue, A. , & author Morerio, P. ( year 2023 ). title Guiding pseudo-labels with uncertainty estimation for source-free unsupervised domain adaptation . In booktitle CVPR \/ (pp. pages 7640--7650 )

    2023

  23. [31]

    , author Tran, C

    author Liu, P. , author Tran, C. T. , author Kong, B. , & author Fang, R. ( year 2022 a ). title CADA: M ulti-scale collaborative adversarial domain adaptation for unsupervised optic disc and cup segmentation . journal Neurocomputing \/ , volume 469 \/ , pages 209--220

    2022

  24. [32]

    , author Dou, Q

    author Liu, Q. , author Dou, Q. , author Yu, L. , & author Heng, P. A. ( year 2020 ). title MS-Net : Multi-site network for improving prostate segmentation with heterogeneous MRI data . journal IEEE Transactions on Medical Imaging \/ , volume 39 \/ , pages 2713 -- 2724

    2020

  25. [33]

    , author Wu, J

    author Liu, X. , author Wu, J. , author Luo, X. , author Liao, W. , author Zhang, S. , author Zhang, S. , & author Wang, G. ( year 2024 ). title RPL-SFDA : Reliable pseudo label-guided source-free cross-modality adaptation for NPC GTV segmentation . In booktitle ISBI \/ (pp. p...

    2024

  26. [34]

    , author Xing, F

    author Liu, X. , author Xing, F. , author El Fakhri, G. , & author Woo, J. ( year 2023 ). title Memory consistent unsupervised off-the-shelf model adaptation for source-relaxed medical image segmentation . journal Medical Image Analysis \/ , volume 83 \/ , pages 102641

    2023

  27. [35]

    , author Yoo, C

    author Liu, X. , author Yoo, C. , author Xing, F. , author Kuo, C.-C. J. , author El Fakhri, G. , author Kang, J.-W. , & author Woo, J. ( year 2022 b ). title Unsupervised black-box model domain adaptation for brain tumor segmentation . journal Frontiers in Neuroscience \/ , v...

    2022

  28. [36]

    , author Yoo, C

    author Liu, X. , author Yoo, C. , author Xing, F. , author Kuo, C. J. , author El Fakhri, G. , author Kang, J.-W. , & author Woo, J. ( year 2022 c ). title Unsupervised domain adaptation for segmentation with black-box source model . In booktitle Proceedings of SPIE--the Inter...

    2022

  29. [37]

    , author Zhang, W

    author Liu, Y. , author Zhang, W. , & author Wang, J. ( year 2021 ). title Source-free domain adaptation for semantic segmentation . In booktitle CVPR \/ (pp. pages 1215--1224 )

    2021

  30. [38]

    , author He, Y

    author Ma, J. , author He, Y. , author Li, F. , author Han, L. , author You, C. , & author Wang, B. ( year 2024 ). title Segment anything in medical images . journal Nature Communications \/ , volume 15 \/ , pages 654

    2024

  31. [39]

    , author Padhy, S

    author Nado, Z. , author Padhy, S. , author Sculley, D. , author D'Amour, A. , author Lakshminarayanan, B. , & author Snoek, J. ( year 2020 ). title Evaluating prediction-time batch normalization for robustness under covariate shift . journal arXiv preprint arXiv:2006.10963 \/ ,

    2020 arXiv

  32. [40]

    , author Efros, A

    author Park, T. , author Efros, A. A. , author Zhang, R. , & author Zhu, J.-Y. ( year 2020 ). title Contrastive learning for unpaired image-to-image translation . In booktitle ECCV \/ (pp. pages 319--345 ). organization Springer

    2020

  33. [41]

    author Raychaudhuri, D. S. , author Ta, C.-K. , author Dutta, A. , author Lal, R. , & author Roy-Chowdhury, A. K. ( year 2023 ). title Prior-guided source-free domain adaptation for human pose estimation . In booktitle ICCV \/ (pp. pages 14996--15006 )

    2023

  34. [42]

    , author Fischer, P

    author Ronneberger, O. , author Fischer, P. , & author Brox, T. ( year 2015 ). title U-Net : Convolutional networks for biomedical image segmentation . In booktitle MICCAI \/ (pp. pages 234--241 ). organization Springer

    2015

  35. [43]

    , author Chatterjee, D

    author Sener, F. , author Chatterjee, D. , author Shelepov, D. , author He, K. , author Singhania, D. , author Wang, R. , & author Yao, A. ( year 2022 ). title Assembly101: A large-scale multi-view video dataset for understanding procedural activities . In booktitle CVPR \/ (p...

    2022

  36. [44]

    , author Wang, X

    author Sun, Y. , author Wang, X. , author Liu, Z. , author Miller, J. , author Efros, A. , & author Hardt, M. ( year 2020 ). title Test-time training with self-supervision for generalization under distribution shifts . In booktitle ICML \/ (pp. pages 9229--9248 )

    2020

  37. [45]

    , & author Luo, T

    author Sunkara, R. , & author Luo, T. ( year 2022 ). title No more strided convolutions or pooling: A new CNN building block for low-resolution images and small objects . In booktitle ECML-PKDD \/ (pp. pages 443--459 ). organization Springer

    2022

  38. [46]

    , author Vray, G

    author Tomar, D. , author Vray, G. M. G. , author Bozorgtabar, B. , & author Thiran, J.-P. ( year 2022 ). title OptTTA : Learnable test-time augmentation for source-free medical image segmentation under domain shift . In booktitle MIDL \/ (pp. pages 1192--1217 ). organization PMLR

    2022

  39. [47]

    , author Jain, H

    author Vu, T.-H. , author Jain, H. , author Bucher, M. , author Cord, M. , & author P \'e rez, P. ( year 2019 ). title Advent: Adversarial entropy minimization for domain adaptation in semantic segmentation . In booktitle CVPR \/ (pp. pages 2517--2526 )

    2019

  40. [48]

    , author Li, Z

    author Wagner, F. , author Li, Z. , author Saha, P. , & author Kamnitsas, K. ( year 2023 ). title Post-deployment adaptation with access to source data via federated learning and source-target remote gradient alignment . In booktitle MLMI \/ (pp. pages 253--263 ). organization...

    2023

  41. [49]

    , author Shelhamer, E

    author Wang, D. , author Shelhamer, E. , author Liu, S. , author Olshausen, B. , & author Darrell, T. ( year 2021 ). title TENT: Fully test-time adaptation by entropy minimization . In booktitle ICLR \/ (pp. pages 1--15 )

    2021

  42. [50]

    , author Li, W

    author Wang, G. , author Li, W. , author Aertsen, M. , author Deprest, J. , author Ourselin, S. , & author Vercauteren, T. ( year 2019 ). title Aleatoric uncertainty estimation with test-time augmentation for medical image segmentation with convolutional neural networks . jour...

    2019

  43. [51]

    , author Luo, X

    author Wang, G. , author Luo, X. , author Gu, R. , author Yang, S. , author Qu, Y. , author Zhai, S. , author Zhao, Q. , author Li, K. , & author Zhang, S. ( year 2023 ). title PyMIC : A deep learning toolkit for annotation-efficient medical image segmentation . journal Comput...

    2023

  44. [52]

    , author Zuluaga, M

    author Wang, G. , author Zuluaga, M. A. , author Li, W. , author Pratt, R. , author Patel, P. A. , author Aertsen, M. , author Doel, T. , author David, A. L. , author Deprest, J. , author Ourselin, S. et al. ( year 2018 ). title DeepIGeoS: A deep interactive geodesic framework...

    2018

  45. [53]

    , author Du, Y

    author Wang, Q. , author Du, Y. , author Fan, H. , & author Ma, C. ( year 2022 ). title Towards collaborative appearance and semantic adaptation for medical image segmentation . journal Neurocomputing \/ , volume 491 \/ , pages 633--643

    2022

  46. [54]

    , author Zhang, X

    author Wen, Z. , author Zhang, X. , & author Ye, C. ( year 2023 ). title Source-free domain adaptation for medical image segmentation via selectively updated mean teacher . In booktitle IPMI \/ (pp. pages 225--236 ). organization Springer

    2023

  47. [55]

    author Wolterink, J. M. , author Dinkla, A. M. , author Savenije, M. H. , author Seevinck, P. R. , author van den Berg, C. A. , & author I s gum, I. ( year 2017 ). title Deep MR to CT synthesis using unpaired data . In booktitle SASHIMI \/ (pp. pages 14--23 ). organization Springer

    2017

  48. [56]

    , author Gu, R

    author Wu, J. , author Gu, R. , author Lu, T. , author Zhang, S. , & author Wang, G. ( year 2023 a ). title UPL-TTA : Uncertainty-aware pseudo label guided fully test time adaptation for fetal brain segmentation . In booktitle IPMI \/ (pp. pages 237--249 ). organization Springer

    2023

  49. [57]

    , author Ji, W

    author Wu, J. , author Ji, W. , author Liu, Y. , author Fu, H. , author Xu, M. , author Xu, Y. , & author Jin, Y. ( year 2023 b ). title Medical SAM adapter : Adapting segment anything model for medical image segmentation . journal arXiv preprint arXiv:2304.12620 \/ ,

    2023 arXiv

  50. [58]

    , author Wang, G

    author Wu, J. , author Wang, G. , author Gu, R. , author Lu, T. , author Chen, Y. , author Zhu, W. , author Vercauteren, T. , author Ourselin, S. , & author Zhang, S. ( year 2023 c ). title UPL-SFDA : Uncertainty-aware pseudo label guided source-free domain adaptation for medi...

    2023

  51. [59]

    , author Zheng, H

    author Xia, Q. , author Zheng, H. , author Zou, H. , author Luo, D. , author Tang, H. , author Li, L. , & author Jiang, B. ( year 2025 ). title A comprehensive review of deep learning for medical image segmentation . journal Neurocomputing \/ , volume 613 \/ , pages 128740

    2025

  52. [60]

    , author Bennett, T

    author Xing, F. , author Bennett, T. , & author Ghosh, D. ( year 2019 ). title Adversarial domain adaptation and pseudo-labeling for cross-modality microscopy image quantification . In booktitle MICCAI \/ (pp. pages 740--749 ). organization Springer

    2019

  53. [61]

    , author Cornish, T

    author Xing, F. , author Cornish, T. C. , author Bennett, T. D. , & author Ghosh, D. ( year 2020 ). title Bidirectional mapping-based domain adaptation for nucleus detection in cross-modality microscopy images . journal IEEE Transactions on Medical Imaging \/ , volume 40 \/ , ...

    2020

  54. [62]

    , author Lu, D

    author Xu, Z. , author Lu, D. , author Wang, Y. , author Luo, J. , author Wei, D. , author Zheng, Y. , & author Tong, R. K.-Y. ( year 2022 ). title Denoising for relaxing : Unsupervised domain adaptive fundus image segmentation without source data . In booktitle MICCAI \/ (pp....

    2022

  55. [63]

    , author Guo, X

    author Yang, C. , author Guo, X. , author Chen, Z. , & author Yuan, Y. ( year 2022 ). title Source free domain adaptation for medical image segmentation with fourier style mining . journal Medical Image Analysis \/ , volume 79 \/ , pages 102457

    2022

  56. [64]

    , author Ren, Y

    author Yang, Z. , author Ren, Y. , author Wu, Z. , author Zeng, M. , author Xu, J. , author Yang, Y. , author Pu, X. , author Philip, S. Y. , & author He, L. ( year 2023 ). title DC-FUDA : Improving deep clustering via fully unsupervised domain adaptation . journal Neurocomput...

    2023

  57. [65]

    , author Lu, X

    author Zhang, H. , author Lu, X. , author Ma, P. , author Liu, J. , author Lian, J. , & author Ma, Y. ( year 2025 ). title Cluster fusion based cross teaching for semi-supervised medical image segmentation . journal Neurocomputing \/ , volume 618 \/ , pages 129147

    2025

  58. [66]

    , author Ma, D

    author Zhang, Y. , author Ma, D. , & author Wu, X. ( year 2024 ). title Source-free domain adaptation framework based on confidence constrained mean teacher for fundus image segmentation . journal Neurocomputing \/ , (p. pages 129262 )

    2024

  59. [67]

    , author Park, T

    author Zhu, J.-Y. , author Park, T. , author Isola, P. , & author Efros, A. A. ( year 2017 ). title Unpaired image-to-image translation using cycle-consistent adversarial networks . In booktitle ICCV \/ (pp. pages 2223--2232 )

    2017

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.