Pith. sign in

REVIEW 3 major objections 5 minor 40 references

A training-free memory bank of reliable predictions lets vision-language models adapt to new medical imaging domains and beat fine-tuning-based test-time adaptation.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 17:14 UTC pith:BTETYZNA

load-bearing objection A well-executed training-free TTA pipeline with large reported gains; the main risk is that the memory-selection proxy is unvalidated and the evidence lacks code/error bars. the 3 major comments →

arxiv 2607.17693 v1 pith:BTETYZNA submitted 2026-07-20 cs.CV

Memory-Supported Synergistic Adaptation for Training-Free Test-Time Medical Image Segmentation

Lingrui Li , Nan Pu , Dong Zhao , Wenjing Li , Andrew P French , Zhun Zhong , Xin Chen This is my paper
classification cs.CV
keywords test-time adaptationmedical image segmentationvision-language modelstraining-free adaptationmemory bankprototype matchingdomain shiftBiomedCLIP
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

This paper proposes Memory-Supported Synergistic Adaptation (MSSA), a training-free test-time adaptation method for medical image segmentation that uses vision-language models (VLMs). The central claim is that instead of updating model parameters to adapt to a new domain, one can build an online memory bank of high-confidence image-text predictions and refine each new test image by aligning its features to prototypes from the most similar stored example. The authors show that this non-parametric approach consistently improves VLM-based segmentation and, on optic disc and lung segmentation benchmarks, surpasses fine-tuning-based TTA methods by large margins (up to +12.2% DSC and +11.7% mIoU). The significance is that it challenges the assumption that test-time adaptation requires parameter updates, and that it can be done without source data or labels.

Core claim

The paper's central discovery is that for medical image segmentation with VLMs, the degradation caused by fine-tuning on noisy pseudo-labels outweighs the adaptation benefit, and a training-free alternative can do better. MSSA operates in two stages. First, it generates coarse candidate segmentations by grounding image-text alignment (using a VLM to produce saliency maps that prompt a segmentation model), stabilized by majority voting across augmentations. Second, it curates a memory bank by admitting only candidates that pass a dual criterion: semantic alignment with the text prompt (BiomedCLIP similarity of the masked image) and spatial smoothness (shape regularity and boundary smoothness)

What carries the argument

The machinery is a memory bank of reliable image-text predictions, maintained with a dual-criteria score (semantic alignment Qsem plus spatial smoothness Qsmo) and an adaptive, monotonically non-decreasing threshold, coupled with a relevance-driven prototype alignment that uses DINOv2 feature similarity to select an anchor and ALPNet-style local prototypes to refine the query mask.

Load-bearing premise

The load-bearing premise is that the vision-language model's similarity score between a masked candidate and the text prompt reliably indicates whether that candidate is a correct segmentation; if that proxy is miscalibrated on the target domain, the memory bank will store wrong masks and the subsequent prototype matching will propagate those errors.

What would settle it

On a target domain where the text prompt is deliberately mismatched (e.g., asking for 'optic disc' on lung images) or where BiomedCLIP's ranking of candidates is swapped relative to Dice score, MSSA's memory bank would fill with wrong masks; comparing Qsem ranking against ground-truth DSC across a held-out set of target images would expose this failure.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Fine-tuning-based TTA is not necessary for VLM segmentation; curation of reliable pseudo-labels plus non-parametric matching achieves higher accuracy.
  • Adapting to a new medical domain can be done online with zero backpropagation, preserving the VLM's pretrained alignment.
  • The method is robust to substantial pathological shifts (e.g., COVID-19 chest X-rays), where fine-tuning methods degrade.
  • The adaptive thresholding mechanism makes the method relatively insensitive to the initial threshold setting.
  • The observed gains are consistent across diverse fundus and chest X-ray domains.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The success hinges on Qsem being a trustworthy proxy for segmentation quality; if the VLM's text-image alignment is miscalibrated on the target domain, the memory bank could fill with plausible but anatomically wrong masks.
  • This suggests a general design principle for VLM-based dense prediction: preserving pretrained cross-modal alignment is more valuable than optimizing on noisy predictions.
  • The same memory-prototype scheme could extend to other dense tasks (e.g., detection or panoptic segmentation) and other shift types, but the current evaluation is limited to two anatomically regular structures.
  • The latency trade-off (12.1 s/image) is acceptable for offline diagnostic use but may need pruning before real-time deployment.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes MSSA, a training-free test-time adaptation (TTA) method for VLM-based medical image segmentation. It builds on MedCLIP-SAMv2, adding Gaussian point selection and dual majority voting to stabilize image-text candidate masks, then constructs an online memory bank by filtering candidates with a semantic-alignment score (BiomedCLIP image-text similarity) plus a spatial smoothness score. New queries are segmented by retrieving the most DINOv2-similar anchor from the bank and performing non-parametric prototype matching (adapted from ALPNet). Experiments on optic-disc (five domains) and lung (three domains) benchmarks report large gains over training-based TTA baselines, e.g., +12.2% DSC and +11.7% mIoU average over TTCS on optic disc, with ablations and a short limitation/latency discussion.

Significance. If the empirical results are reproducible, this is a meaningful contribution: it is an early training-free TTA framework for VLM segmentation, avoiding parameter-update drift that limits fine-tuning-based TTA, and it is conceptually simple. The reported gains are large and consistent across domains, and the ablation study is informative, especially the demonstration in Table 3 (row 6) that a purely self-referential image-image variant can collapse on Domain D. The paper also states limitations on latency and organ geometry. However, the reliability of the memory-selection proxy is not validated, key hyperparameters are selected on the evaluation benchmarks without sensitivity analysis, and no code/data are provided; these gaps currently weaken the support for the central claim.

major comments (3)
  1. [Sec. 3.2, Eq. (3); Table 3] The semantic alignment score Qsem is computed with the same BiomedCLIP model that produced the saliency map and the candidate mask, making it a self-referential quality signal rather than an independent check. The manuscript provides no evidence that Qsem is a calibrated proxy for segmentation quality under domain shift (e.g., correlation with DSC on a labeled subset, an oracle-vs-proxy comparison, or prompt/threshold sensitivity). This is load-bearing because the memory bank admits candidates only through Eq. (4), and errors propagate to subsequent queries through Eqs. (7)-(10). The risk is concrete: Table 3 row 6 shows that the same selection mechanism, in the self-referential image-image variant, leads to catastrophic failure on Domain D (18.6 mIoU). Please add a direct validation of the selection proxy, or an analysis of bank composition and failure cases under domain shift.
  2. [Sec. 4.2; Table 3; contribution bullet (Sec. 1)] The method's hyperparameters (sigma=0.25, P=80, N_warm=10, memory bank size 15, choice of augmentation set) are fixed values with no sensitivity analysis and no separate validation split. This contradicts the contribution claim that the framework works "without introducing sensitive hyperparameter tuning." Additionally, all results are single runs with no error bars, and since TTA is inherently stream-order-dependent, the absence of variance over test-stream permutations is a concern. Please provide sensitivity results for the key hyperparameters (or a principled selection protocol) and report mean +/- std over several stream orders or random seeds.
  3. [General reproducibility] No code, data splits, or detailed algorithm pseudocode are provided; the project page link is not a code release. The pipeline is complex (augmentations, adaptive thresholding, memory FIFO, prototype matching), and the exact prompt templates and augmentation parameters are not fully specified. Without this material, the benchmark numbers cannot be independently reproduced. Please release code and exact dataset splits/prompt templates, or make the supplementary material sufficiently detailed for exact reproduction.
minor comments (5)
  1. [Table 2] In the TTCS row, the MC and SZ columns are identical (65.2 DSC / 43.1 mIoU). This is likely a copy-paste error; please correct and recompute the averages.
  2. [Algorithm 1] The formulas for shape regularity Rs and boundary smoothness Bs are not given; only verbal descriptions are provided. Please specify the exact equations so the score can be reproduced.
  3. [Eq. (8)] The function f_theta is described as an indicator function, but the notation f_theta[Y(u,v)=c] is not defined. Please clarify, e.g., use 1[Y(u,v)=c] or define f_theta explicitly.
  4. [Sec. 4.2] The phrase "sigma empirically determined as 0.25" lacks a procedure. Please state the validation criterion and the range explored.
  5. [Fig. 1(d)] The performance comparison figure lacks axis labels and error bars; please clarify what is being plotted and improve the caption.

Circularity Check

0 steps flagged

No significant circularity: MSSA's central results are benchmarked against external ground truth; the Qsem self-score is a selection heuristic, not a derived prediction.

full rationale

MSSA is an empirically evaluated pipeline, not a derivation whose output is forced by its inputs. The headline gains (Tables 1–2) are measured against human ground-truth masks on public datasets, so the central claims are externally falsifiable. The closest thing to a self-referential loop is the memory-bank selector: Eq. 3 computes Qsem = BiomedCLIP(I_masked, T) with the same BiomedCLIP model that produced the initial saliency map and candidate mask, and Eq. 4 admits candidates with high Qsem + Qsmo. This is an internal-consistency heuristic rather than a circular definition: the final prediction is not Qsem, but is produced by DINOv2 prototype matching (Eqs. 7–10) and can disagree with the score. The paper itself demonstrates non-circularity: Table 3 rows 5–6 show a purely self-referential image–image variant 'causing a catastrophic failure on Domain D, even with NMC,' which shows self-consistency alone is not equated with correctness and that the full method's success depends on the external benchmark. Section 4.5 also openly limits the approach to organs with regular geometries. Self-citations [14,15,39,40] appear only in related-work context (TTA, source-free adaptation, cross-image consistency) and no load-bearing claim reduces to them; no uniqueness theorem or imported ansatz is involved. Hence no circular step can be exhibited; the mild self-referential design of the selector warrants at most a score of 1, not a finding of circularity.

Axiom & Free-Parameter Ledger

6 free parameters · 4 axioms · 0 invented entities

Central claim rests on several untested domain assumptions about foundation-model features and filtering scores, plus hand-set hyperparameters. No new physical entities are introduced.

free parameters (6)
  • sigma (Gaussian point-selection spread) = 0.25
    Controls the Gaussian weighting in Eq. (1); chosen empirically on the target benchmarks (Sec. 4.2).
  • P (memory selection percentile) = 80
    Adaptive threshold percentile in Eq. (6); hand-set to 80.
  • N_warm (warm-up length) = 10
    Number of samples before percentile thresholding activates; hand-set.
  • Memory bank capacity = 15
    FIFO memory size; hand-set in Sec. 4.2 without a sensitivity study.
  • tau_init (initial selection threshold) = not specified
    Initial threshold used before warm-up; value not reported, claimed to be superseded after warm-up.
  • Augmentation set for SAM majority voting = 4 geometric variants
    Number and type of SAM augmentations chosen by hand; affects stability and latency.
axioms (4)
  • domain assumption BiomedCLIP image-text similarity of a masked candidate (Qsem) is a valid proxy for segmentation quality under target-domain shift.
    Used as the primary filter in Sec. 3.2 Eq. (3); if VLM similarity is miscalibrated under shift, the memory bank can be filled with semantically plausible but spatially wrong masks.
  • domain assumption DINOv2 cosine feature similarity selects an anchor whose mask and prototypes transfer to the query under cross-site/pathology shift.
    Anchor selection relies on DINOv2 feature similarity (Eq. 7); no evidence is given that nearest-neighbor in feature space implies mask-protototype transferability in the target domain.
  • domain assumption Shape regularity and boundary smoothness scores correlate with clinically useful anatomical masks.
    Qsmo in Algorithm 1 assumes well-formed anatomical structures are smooth and consolidated; pathological or irregular structures could be down-weighted.
  • domain assumption The GPT-4-generated text prompt is a faithful description of the target anatomy for BiomedCLIP.
    All groundings depend on the prompt; prompt errors or ambiguity would propagate to candidates and memory.

pith-pipeline@v1.3.0-alltime-deepseek · 13376 in / 13687 out tokens · 145165 ms · 2026-08-01T17:14:45.837361+00:00 · methodology

0 comments
read the original abstract

Test-time adaptation (TTA) aims to mitigate distribution shifts by adapting models with unlabeled target data at inference time. While TTA with vision-language models (VLMs) has shown promising results in classification, extending it to medical image segmentation remains challenging. In this setting, the adaptation gains from optimizing on VLM-generated predictions are often outweighed by the degradation to the VLM's strong pretrained features caused by noisy, update-driven learning, resulting in limited and unstable improvements. We therefore propose Memory-Supported Synergistic Adaptation (MSSA), a novel training-free TTA framework for medical image segmentation. Without updating model parameters, MSSA dynamically selects reliable image-text predictions to construct an online memory, uses them as text-guided semantic priors, and couples them with cross-image structural alignment for robust adaptation. Specifically, MSSA consists of (i) a noise-aware memory construction module that filters and stabilizes cross-modal predictions, and (ii) a relevance-driven prototype alignment module that aligns the target sample with structurally consistent memory samples and their reliable predictions to improve adaptation. Extensive experiments on multiple medical segmentation benchmarks demonstrate that MSSA consistently improves VLM-based segmentation models and outperforms existing fine-tuning-based TTA methods by a clear margin, with gains of up to 12.2% DSC and 11.7% mIoU. Project page: https://lingrayy.github.io/MSSA/ .

Figures

Figures reproduced from arXiv: 2607.17693 by Andrew P French, Dong Zhao, Lingrui Li, Nan Pu, Wenjing Li, Xin Chen, Zhun Zhong.

Figure 1
Figure 1. Figure 1: Comparison of Test-Time Adaptation (TTA) paradigms. (a) Zero-shot VLM segmentation [13] relies on image–text alignment and often suffers from unstable lo￾calization under domain shifts. (b) Fine-tuning-based TTA [4] updates model parame￾ters using noisy image–text pseudo-labels, leading to performance drift. (c) Our MSSA achieves training-free adaptation through synergistic image–text grounding and image– … view at source ↗
Figure 2
Figure 2. Figure 2: Motivation of MSSA. (a) Image–text grounding provides semantic priors for localization but becomes coarse under domain shifts. (b) Image–image consistency en￾forces structural alignment, yet lacks semantic anchoring to reliably localize foreground regions in the target domain.(c) MSSA synergistically combines both, achieving robust and accurate segmentation. Adaptation (MSSA) (see [PITH_FULL_IMAGE:figures… view at source ↗
Figure 3
Figure 3. Figure 3: The overall framework of MSSA operating in two stages. (a) Image-Text Coarse Grounding produces initial coarse masks from input images and text prompts, stabi￾lized via our dual majority voting. (b) Noise-Aware Memory Construction filters these candidates by evaluating their semantic alignment and spatial smoothness, populating a dynamic memory bank with only the highest-quality samples. Image-Image Proto￾… view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison of MSSA against state-of-the-art methods on optic disc datasets. confirms the effectiveness of our selective memory bank in providing stable and accurate guidance. 32 Image SAMAug SaLIP MedCLIPSAMv1 MedCLIPSAMv2 TTCS MSSA(Ours) GT MC SZ COVID -QU-Ex [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative comparison of MSSA against other TTA methods on lung datasets. 4.4 Ablation Study We performed extensive ablation studies to verify the contribution of each com￾ponent in our framework. Component-wise Analysis. The progression of results in [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative comparison of different paradigms on Optic Disc [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

40 extracted references · 11 linked inside Pith

  1. [1]

    arXiv preprint arXiv:2303.08774 (2023)

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F.L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al.: Gpt-4 technical report. arXiv preprint arXiv:2303.08774 (2023)

    Pith/arXiv arXiv 2023

  2. [2]

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

    Aleem, S., Wang, F., Maniparambil, M., Arazo, E., Dietlmeier, J., Curran, K., Connor, N.E., Little, S.: Test-time adaptation with salip: A cascade of sam and clip for zero-shot medical image segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 5184–5193 (2024)

    2024

  3. [3]

    arXiv preprint arXiv:2407.07042 (2024)

    Ayzenberg, L., Giryes, R., Greenspan, H.: Protosam: One-shot medical image seg- mentation with foundational models. arXiv preprint arXiv:2407.07042 (2024)

    Pith/arXiv arXiv 2024

  4. [4]

    In: 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

    Chen, H., Xu, Y., Xu, Y., Zhang, Y., Cui, L.: Test-time medical image segmenta- tion using clip-guided sam adaptation. In: 2024 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). pp. 1866–1873. IEEE (2024) 16 L.Li et al

    2024

  5. [5]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recog- nition

    Chen, Z., Pan, Y., Ye, Y., Lu, M., Xia, Y.: Each test image deserves a specific prompt: Continual test-time adaptation for 2d medical image segmentation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recog- nition. pp. 11184–11193 (2024)

    2024

  6. [6]

    Chowdhury, M.E., Rahman, T., Khandakar, A., Mazhar, R., Kadir, M.A., Mahbub, Z.B., Islam, K.R., Khan, M.S., Iqbal, A., Al Emadi, N., et al.: Can ai help in screening viral and covid-19 pneumonia? Ieee Access8, 132665–132676 (2020)

    2020

  7. [7]

    arXiv preprint arXiv:2307.01187 (2023)

    Dai, H., Ma, C., Yan, Z., Liu, Z., Shi, E., Li, Y., Shu, P., Wei, X., Zhao, L., Wu, Z., et al.: Samaug: Point prompt augmentation for segment anything model. arXiv preprint arXiv:2307.01187 (2023)

    Pith/arXiv arXiv 2023

  8. [8]

    In: 2011 24th International Symposium on Computer-based Medical Systems (CBMS)

    Fumero, F., Alayón, S., Sanchez, J.L., Sigut, J., Gonzalez-Hernandez, M.: Rim- one: An open retinal image database for optic nerve evaluation. In: 2011 24th International Symposium on Computer-based Medical Systems (CBMS). pp. 1–6. IEEE (2011)

    2011

  9. [9]

    ICLR1(2), 3 (2022)

    Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W., et al.: Lora: Low-rank adaptation of large language models. ICLR1(2), 3 (2022)

    2022

  10. [10]

    IEEE transactions on medical imaging33(2), 233–245 (2013)

    Jaeger, S., Karargyris, A., Candemir, S., Folio, L., Siegelman, J., Callaghan, F., Xue, Z., Palaniappan, K., Singh, R.K., Antani, S., et al.: Automatic tuberculosis screening using chest radiographs. IEEE transactions on medical imaging33(2), 233–245 (2013)

    2013

  11. [11]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Kirillov, A., Mintun, E., Ravi, N., Mao, H., Rolland, C., Gustafson, L., Xiao, T., Whitehead, S., Berg, A.C., Lo, W.Y., et al.: Segment anything. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 4015–4026 (2023)

    2023

  12. [12]

    In: International conference on medical image computing and computer-assisted intervention

    Koleilat, T., Asgariandehkordi, H., Rivaz, H., Xiao, Y.: Medclip-sam: Bridging text and image towards universal medical image segmentation. In: International conference on medical image computing and computer-assisted intervention. pp. 643–653. Springer (2024)

    2024

  13. [13]

    Medical Image Analysis p

    Koleilat, T., Asgariandehkordi, H., Rivaz, H., Xiao, Y.: Medclip-samv2: Towards universal text-driven medical image segmentation. Medical Image Analysis p. 103749 (2025)

    2025

  14. [14]

    In: 2025 IEEE International Conference on Bioinformatics and Biomedicine (BIBM)

    Li, L., Zhou, Y., Pu, N., Chen, X., Zhong, Z.: Multi-scale global-instance prompt tuning for continual test-time adaptation in medical image segmentation. In: 2025 IEEE International Conference on Bioinformatics and Biomedicine (BIBM). pp. 2395–2402. IEEE (2025)

    2025

  15. [15]

    In: Proceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision

    Li, L., Zhou, Y., Yang, G.: Robust source-free domain adaptation for fundus image segmentation. In: Proceedings of the IEEE/CVF Winter Conference on Applica- tions of Computer Vision. pp. 7840–7849 (2024)

    2024

  16. [16]

    arXiv preprint arXiv:2305.13310 (2023)

    Liu, Y., Zhu, M., Li, H., Chen, H., Wang, X., Shen, C.: Matcher: Segment anything with one shot using all-purpose feature matching. arXiv preprint arXiv:2305.13310 (2023)

    Pith/arXiv arXiv 2023

  17. [17]

    Advances in Neural Informa- tion Processing Systems36, 6690–6702 (2023)

    Ma, X., Wang, Y., Liu, H., Guo, T., Wang, Y.: When visual prompt tuning meets source-free domain adaptive semantic segmentation. Advances in Neural Informa- tion Processing Systems36, 6690–6702 (2023)

    2023

  18. [18]

    arXiv preprint arXiv:2507.16337 (2025)

    Mao, X., Xing, X., Meng, F., Liu, J., Bai, F., Nie, Q., Meng, M.: One polyp iden- tifies all: One-shot polyp segmentation with sam via cascaded priors and iterative prompt evolution. arXiv preprint arXiv:2507.16337 (2025)

    Pith/arXiv arXiv 2025

  19. [19]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Mirza, M.J., Micorek, J., Possegger, H., Bischof, H.: The norm must go on: Dy- namic unsupervised domain adaptation by normalization. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 14765– 14775 (2022) MSSA for Training-Free Test-Time Medical Image Segmentation 17

    2022

  20. [20]

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

    Nie, J., Xing, Y., Zhang, G., Yan, P., Xiao, A., Tan, Y.P., Kot, A.C., Lu, S.: Cross- domain few-shot segmentation via iterative support-query correspondence mining. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 3380–3390 (2024)

    2024

  21. [21]

    In: International conference on machine learning

    Niu, S., Wu, J., Zhang, Y., Chen, Y., Zheng, S., Zhao, P., Tan, M.: Efficient test- time model adaptation without forgetting. In: International conference on machine learning. pp. 16888–16905. PMLR (2022)

    2022

  22. [22]

    arXiv preprint arXiv:2304.07193 (2023)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 (2023)

    Pith/arXiv arXiv 2023

  23. [23]

    Medical Image Analysis59, 101570 (2020)

    Orlando, J.I., Fu, H., Breda, J.B., van Keer, K., Bathula, D.R., Diaz-Pinto, A., Fang, R., Heng, P.A., Kim, J., Lee, J., et al.: Refuge challenge: A unified frame- work for evaluating automated methods for glaucoma assessment from fundus pho- tographs. Medical Image Analysis59, 101570 (2020)

    2020

  24. [24]

    In: European conference on computer vision

    Ouyang, C., Biffi, C., Chen, C., Kart, T., Qiu, H., Rueckert, D.: Self-supervision with superpixels: Training few-shot medical image segmentation without annota- tion. In: European conference on computer vision. pp. 762–780. Springer (2020)

    2020

  25. [25]

    In: International conference on machine learning

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al.: Learning transferable visual models from natural language supervision. In: International conference on machine learning. pp. 8748–8763. PmLR (2021)

    2021

  26. [26]

    EURASIP Journal on Image and Video Processing2016(1), 35 (2016)

    Rahman, S., Rahman, M.M., Abdullah-Al-Wadud, M., Al-Quaderi, G.D., Shoyaib, M.: An adaptive gamma correction for image enhancement. EURASIP Journal on Image and Video Processing2016(1), 35 (2016)

    2016

  27. [27]

    Computers in biology and medicine132, 104319 (2021)

    Rahman, T., Khandakar, A., Qiblawey, Y., Tahir, A., Kiranyaz, S., Kashem, S.B.A., Islam, M.T., Al Maadeed, S., Zughaier, S.M., Khan, M.S., et al.: Exploring the effect of image enhancement techniques on covid-19 detection using chest x-ray images. Computers in biology and medicine132, 104319 (2021)

    2021

  28. [28]

    JSM Biomedical Imaging Data Papers2(1), 1004 (2015)

    Sivaswamy, J., Krishnadas, S., Chakravarty, A., Joshi, G., Tabish, A.S., et al.: A comprehensive retinal image dataset for the assessment of glaucoma from the optic nerve head analysis. JSM Biomedical Imaging Data Papers2(1), 1004 (2015)

    2015

  29. [29]

    arXiv preprint arXiv:2006.10726 (2020)

    Wang, D., Shelhamer, E., Liu, S., Olshausen, B., Darrell, T.: Tent: Fully test-time adaptation by entropy minimization. arXiv preprint arXiv:2006.10726 (2020)

    Pith/arXiv arXiv 2006

  30. [30]

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

    Wang, Q., Fink, O., Van Gool, L., Dai, D.: Continual test-time domain adaptation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 7201–7211 (2022)

    2022

  31. [31]

    Advances in Neural Information Processing Systems36, 16009–16027 (2023)

    Wang, Y., Rudner, T.G., Wilson, A.G.: Visual explanations of image-text repre- sentations via multi-modal information bottleneck attribution. Advances in Neural Information Processing Systems36, 16009–16027 (2023)

    2023

  32. [32]

    IEEE Transactions on Pattern Analysis and Machine Intelligence45(7), 8827–8844 (2023)

    Wu, L., Fang, L., He, X., He, M., Ma, J., Zhong, Z.: Querying labeled for unlabeled: Cross-image semantic consistency guided semi-supervised semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence45(7), 8827–8844 (2023)

    2023

  33. [33]

    In: Proceed- ings of the AAAI Conference on Artificial Intelligence

    Yang, S., Wu, J., Liu, J., Li, X., Zhang, Q., Pan, M., Gan, Y., Chen, Z., Zhang, S.: Exploring sparse visual prompt for domain adaptive dense prediction. In: Proceed- ings of the AAAI Conference on Artificial Intelligence. pp. 16334–16342 (2024)

    2024

  34. [34]

    arXiv preprint arXiv:2305.03048 (2023) 18 L.Li et al

    Zhang, R., Jiang, Z., Guo, Z., Yan, S., Pan, J., Ma, X., Dong, H., Gao, P., Li, H.: Personalize segment anything model with one shot. arXiv preprint arXiv:2305.03048 (2023) 18 L.Li et al

    Pith/arXiv arXiv 2023

  35. [35]

    arXiv preprint arXiv:2303.009152(3), 6 (2023)

    Zhang, S., Xu, Y., Usuyama, N., Bagga, J., Tinn, R., Preston, S., Rao, R., Wei, M., Valluri, N., Wong, C., et al.: Large-scale domain-specific pretraining for biomedical vision-language processing. arXiv preprint arXiv:2303.009152(3), 6 (2023)

    Pith/arXiv arXiv 2023

  36. [36]

    arXiv preprint arXiv:2311.07877 (2023)

    Zhang, Y., Sun, Y., Zheng, S., Shui, Z., Zhu, C., Yang, L.: Test-time training for se- mantic segmentation with output contrastive loss. arXiv preprint arXiv:2311.07877 (2023)

    Pith/arXiv arXiv 2023

  37. [37]

    In: 2010 Annual international conference of the IEEE engineering in medicine and biology

    Zhang, Z., Yin, F.S., Liu, J., Wong, W.K., Tan, N.M., Lee, B.H., Cheng, J., Wong, T.Y.: Origa-light: An online retinal fundus image database for glaucoma analysis and research. In: 2010 Annual international conference of the IEEE engineering in medicine and biology. pp. 3065–3068. IEEE (2010)

    2010

  38. [38]

    arXiv preprint arXiv:2301.13018 (2023)

    Zhao, B., Chen, C., Xia, S.T.: Delta: degradation-free fully test-time adaptation. arXiv preprint arXiv:2301.13018 (2023)

    Pith/arXiv arXiv 2023

  39. [39]

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

    Zhao, D., Wang, S., Zang, Q., Jiao, L., Sebe, N., Zhong, Z.: Stable neighbor denoising for source-free domain adaptive segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 23416– 23427 (2024)

    2024

  40. [40]

    IEEE TPAMI47(11), 10378–10395 (2025).https://doi.org/10.1109/ TPAMI.2025.3596943

    Zhao, D., Zang, Q., Pu, N., Wang, S., Sebe, N., Zhong, Z.: Secov2: Semantic connectivity-driven pseudo-labeling for robust cross-domain semantic segmenta- tion. IEEE TPAMI47(11), 10378–10395 (2025).https://doi.org/10.1109/ TPAMI.2025.3596943

    arXiv 2025