Pith. sign in

REVIEW 2 major objections 6 minor 28 references

Swapping Scanner Context to Audit Tumor Segmentation

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 · glm-5.2

2026-07-09 21:11 UTC pith:4DYJS4TM

load-bearing objection CCT audit signal is never validated against ground-truth context sensitivity, making the core novelty indistinguishable from a training regularizer that helps OOD Dice. the 2 major comments →

arxiv 2607.07038 v1 pith:4DYJS4TM submitted 2026-07-08 cs.CV

TRACE-Seg3D: Counterfactual Context Auditing For Robust 3D Glioma Segmentation Under Institutional Shift

Nguyen Linh Dan Le , Nguyen Pham Hoang Le , Tran Dang Khoi This is my paper
classification cs.CV
keywords medical image segmentationdistribution shiftcounterfactual reasoningdomain generalizationgliomabrain tumorcausal representation learningauditability
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.

The paper argues that a 3D brain tumor segmentation model can be audited for reliability by holding the disease evidence fixed and swapping the imaging context, then measuring how much the prediction changes. The central object is Counterfactual Context Transport (CCT), a mechanism that factorizes a scan into a disease latent and a context latent, replaces the context latent with alternatives from a support bank, and reports a voxel-wise instability map alongside the segmentation. The paper claims this audit signal exposes context-sensitive failure modes that standard overlap metrics miss, and that enforcing the anatomical hierarchy of tumor subregions (enhancing tumor contained within tumor core, contained within whole tumor) during both training and inference improves cross-institution robustness, particularly for the small and fragile enhancing tumor region.

Core claim

By separating disease evidence from imaging context at the latent level and then systematically transporting the context while keeping disease fixed, the paper demonstrates that segmentation predictions can be paired with a case-level audit signal that quantifies context sensitivity. The paper reports that this factorization, combined with an anatomy-aware structural prior, yields the strongest cross-domain segmentation performance in both transfer directions between two glioma datasets (BraTS and UTSW), improving OOD mean DSC while controlling isolated false positives in the enhancing tumor subregion.

What carries the argument

Counterfactual Context Transport (CCT): a mechanism that keeps the disease latent z_d fixed and replaces the context latent z_c with alternatives sampled from a support bank, producing a consensus prediction, a voxel-wise instability map, and a stability-gated mask. This is paired with proxy-anchored factorization (supervising disease and context latents with weak proxy targets and adversarial reversal to reduce leakage) and an anatomy-aware structural prior (enforcing the nested hierarchy ET ⊆ TC ⊆ WT to remove implausible enhancing tumor islands).

Load-bearing premise

The learned latent variables that supposedly separate disease evidence from imaging context are not guaranteed to be cleanly disentangled. They are supervised by weak proxy signals (like intensity statistics and label volumes) that could correlate with both disease and context, meaning the context swap might inadvertently change disease information too, which would invalidate the audit signal.

What would settle it

If the proxy anchors for disease and context are correlated with each other, the factorization fails: swapping the context latent would also swap disease information, and the instability map would reflect proxy confounding rather than true context sensitivity. The paper does not test whether the proxies are independent enough to guarantee clean separation.

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

If this is right

  • If the instability map reliably flags context-sensitive regions, clinicians could use it to decide which parts of an automated segmentation to trust and which to manually review, particularly for small structures like enhancing tumor.
  • The counterfactual transport principle could extend to other medical imaging tasks where scanner or protocol shift is a concern, such as organ segmentation in CT or lesion detection in ultrasound, by auditing whether predictions are driven by anatomy or acquisition artifacts.
  • If the proxy-anchored factorization is effective enough, institutions could share context latents rather than raw data, enabling collaborative improvement of context banks without exchanging patient images.
  • The structural prior approach (enforcing nested anatomical hierarchy as a post-hoc and training constraint) could be adapted to other tasks with known anatomical containment relationships, such as retinal layer segmentation or cardiac structure delineation.

Where Pith is reading between the lines

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

  • If the proxy anchors for disease and context are themselves correlated (e.g., tumor volume correlates with scanner type because certain institutions see different case mixes), the factorization could fail silently, and the instability map would reflect proxy confounding rather than true context sensitivity. The paper acknowledges this risk but does not test it directly.
  • The context bank is built from source-domain training cases only; if the target domain contains a context type not represented in the bank, the audit signal may be uninformative because the transported contexts are too dissimilar to probe the relevant failure mode.
  • The improvement in cross-domain performance may partly come from the anatomy-aware structural prior acting as a regularizer that happens to help OOD generalization, rather than from the causal factorization itself; the ablation shows the prior has the clearest boundary effect, which is consistent with this possibility.
  • The audit signal's clinical utility depends on calibration: if the instability threshold is set too high, the audit provides no filtering; if too low, it flags too much. The paper does not explore how a clinician would calibrate this threshold in practice.

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

2 major / 6 minor

Summary. The paper introduces TRACE-Seg3D, a framework for 3D glioma segmentation under institutional distribution shift. The core idea is to factorize the encoder's bottleneck into a disease-guided latent (z_d) and a context-guided latent (z_c) using proxy anchors and adversarial regularization, then perform Counterfactual Context Transport (CCT) at inference time by holding z_d fixed and swapping z_c from a support bank. This yields a consensus prediction, a voxel-wise instability map, and a stability-gated mask. An anatomy-aware structural prior enforces the ET⊆TC⊆WT hierarchy. Experiments on BraTS and UTSW show competitive in-distribution and cross-domain performance, with ablation studies isolating each component.

Significance. The paper addresses a clinically relevant problem: silent failures in 3D medical image segmentation under scanner and institutional shift. The framing of prediction-level auditability—pairing a segmentation with evidence about context sensitivity and anatomical plausibility—is well-motivated and distinct from prior causal segmentation work that uses causality primarily as a training objective. The inclusion of an anatomy-aware structural prior targeting ET false-positive control is a practical contribution. The code is stated to be publicly available, which supports reproducibility. The cross-domain improvements on BraTS↔UTSW are consistent across both transfer directions, strengthening the empirical claim.

major comments (2)
  1. §3.3, Eqs. (11)–(14), and §5.2: The paper's central novelty is the CCT audit signal (instability map u_cct, consensus p_cct, stability-gated p_gate), but no experiment validates that these audit outputs actually correlate with true context sensitivity or prediction reliability. Table 3 shows that removing CCT drops OOD DSC by 1.6 points, but CCT serves dual roles—as a training regularizer (Eqs. 11, 19) and as an inference-time audit—and the ablation does not separate these. A model trained with the CCT loss but evaluated without the audit outputs would clarify whether the OOD improvement comes from the training regularization or from the inference-time consensus. Additionally, the paper never reports whether p_gate is better calibrated than the ungated prediction, whether u_cct correlates with voxel-level errors under known context shifts, or whether CCT instability is more informative a
  2. §3.2, Eq. (7): The factorization into z_d and z_c relies on proxy anchors. For UTSW, metadata proxies (site, scanner, acquisition descriptors) are available. For BraTS, the paper states it uses 'source-split pseudo-proxies derived from image intensity, geometry, and label-volume summaries.' These pseudo-proxies may be correlated with both disease state and context, which would undermine the separation that CCT depends on. If z_c contains disease information, transporting it would alter the lesion evidence rather than purely the imaging context, invalidating the audit signal. The paper should report the residual cross-head accuracy after adversarial training (Eq. 8) or provide a quantitative measure of leakage between z_d and z_c to demonstrate that the factorization is empirically effective, not just architecturally encouraged.
minor comments (6)
  1. Table 1: The improvement margins over the strongest baselines (e.g., ICMSeg and CauSSL) are modest in some columns. The paper should clarify whether differences are statistically significant or whether the gains could be attributable to the MedNeXt backbone choice, since the backbone is stronger than several baselines.
  2. §3.2: The proxy anchors for BraTS are described vaguely. A supplementary table listing the exact pseudo-proxy definitions (which intensity statistics, which geometry features, which label-volume summaries) would help readers assess whether they meaningfully represent imaging context.
  3. Table 3: The 'Audit' column header is unclear. It is not defined in the text and the values (1.50, 1.70, etc.) have no stated metric. Clarify what is reported.
  4. §3.4, Eq. (15): The thresholds (WT=0.65, TC=0.30, ET=0.55) and the 32-voxel ET minimum are reported in §4.3 but not justified. Figure 4 shows sensitivity analysis, but the paper should state how these values were selected (e.g., validation set optimization) to rule out test-set tuning.
  5. Figure 6: The qualitative examples are useful but limited. Adding a case where the instability map correctly flags a region that is indeed wrong under OOD shift would more directly demonstrate the audit's value.
  6. Reference [20] (UTSW-Glioma) is cited with a 2026 date and a DOI; confirm that the dataset is properly released and publicly available at the time of submission.

Circularity Check

0 steps flagged

No significant circularity found: the derivation chain is self-contained against external benchmarks

full rationale

The paper's central claims are empirically validated against external benchmarks (BraTS, UTSW) and compared against independent baselines (nnU-Net, UNETR, MedNeXt, CSDG, CauSSL, etc.). The three mechanisms—proxy-anchored factorization (M1), Counterfactual Context Transport (M2), and anatomy-aware structural prior (M3)—are defined by their own equations (Eqs. 7-9, 10-14, 15-16) and are not circularly defined in terms of their own outputs. The CCT consensus prediction (Eq. 12) and instability map (Eq. 13) are computed from transported-context predictions (Eq. 10), which are genuine forward passes through the decoder with replaced context latents—not definitions that reduce to their inputs. The structural prior (Eq. 15) enforces ET⊆TC⊆WT as an independent anatomical constraint, not a tautological restatement of the segmentation output. The proxy heads (Eq. 7) supervise latents using disease/context proxies (region volumes, site metadata, intensity statistics) that are external to the segmentation prediction itself. The adversarial loss (Eq. 8) and orthogonality penalty (Eq. 9) are standard regularization terms. No self-citation chain is load-bearing: the paper cites external work (IRM [1], causal representation learning [24], CausalVAE [27], counterfactual generative networks [23]) as motivation, not as mathematical premises that force the conclusion. The OOD performance improvements (Table 1) are measured against held-out target domains, not fitted quantities. The ablation (Table 3) independently removes each component. While the audit signal's clinical validity is not separately validated (a correctness concern, not circularity), the CCT audit outputs are computed from genuine counterfactual interventions on learned latents, not from quantities that are definitionally equivalent to what they claim to measure. The derivation is self-contained.

Axiom & Free-Parameter Ledger

5 free parameters · 3 axioms · 2 invented entities

The framework relies on several hand-tuned parameters (thresholds, margins) and core domain assumptions about the separability of disease and context. The invented entities (context bank, instability map) are validated only internally.

free parameters (5)
  • Region thresholds (tau_WT, tau_TC, tau_ET) = 0.65, 0.30, 0.55
    Decision thresholds for the anatomical prior, chosen empirically.
  • ET component minimum support = 32 voxels
    Hand-tuned threshold for removing small ET islands.
  • CCT stability margin (m_c) = 0.03
    Parameter in the stability penalty loss (Eq. 19).
  • Audit threshold (tau_u) = 0.05
    Threshold for the stability-gated prediction (Eq. 14).
  • Loss weights (lambda_reg, lambda_cct, lambda_stab, lambda_px, lambda_adv, lambda_orth) = Not specified
    Weights for the composite loss function (Eq. 20), values not provided in the paper.
axioms (3)
  • domain assumption Disease evidence and imaging context can be separated into distinct latent variables z_d and z_c.
    The entire framework depends on this factorization being achievable via proxy anchors and adversarial training, as stated in Sec 3.2.
  • domain assumption The anatomical hierarchy ET ⊆ TC ⊆ WT holds for glioma segmentation.
    Used as a structural prior in Sec 3.4 to enforce plausibility.
  • ad hoc to paper Context latents from a source-domain support bank can meaningfully audit context sensitivity on target-domain cases.
    The CCT mechanism (Sec 3.3) assumes that transporting source contexts is sufficient to expose spurious dependencies, even when the target shift is unknown.
invented entities (2)
  • Context bank B_ctx no independent evidence
    purpose: Stores context latents from support cases for use in Counterfactual Context Transport.
    A constructed data structure specific to this method; its effectiveness is validated only through the paper's own experiments.
  • Instability map u_cct no independent evidence
    purpose: Voxel-wise map of context-induced prediction variance.
    A novel audit signal introduced by the paper; no external validation of its correlation with true clinical unreliability is provided.

pith-pipeline@v1.1.0-glm · 15708 in / 2371 out tokens · 225280 ms · 2026-07-09T21:11:22.527650+00:00 · methodology

0 comments
read the original abstract

Medical image segmentation models can achieve strong benchmark performance while remaining sensitive to scanner, protocol, and institutional variation. These context shifts alter image appearance without changing the underlying lesion, allowing models to exploit nuisance cues that Dice and HD95 fail to expose. We present TRACE-Seg3D, a counterfactual context auditing framework for robust 3D medical image segmentation. TRACE-Seg3D preserves lesion-relevant evidence and systematically varies imaging context to quantify prediction stability under controlled context shifts. The framework pairs each segmentation with audit evidence for context sensitivity and anatomical plausibility, enabling case-level reliability assessment beyond overlap-based evaluation. Experiments on BraTS and UTSW glioma segmentation benchmarks demonstrate competitive in-distribution and cross-domain performance. TRACE-Seg3D also exposes context-sensitive failure modes missed by conventional metrics. These results establish counterfactual context auditing as a practical route toward transparent and reliable 3D medical image segmentation under distribution shift. Our code is available at https://github.com/danleneurocom/Counterfactual-Representation-Network.

Figures

Figures reproduced from arXiv: 2607.07038 by Nguyen Linh Dan Le, Nguyen Pham Hoang Le, Tran Dang Khoi.

Figure 1
Figure 1. Figure 1: The overall framework of TRACE-Seg3D. Given a multi-modal 3D MRI volume x, the model first extracts multi-scale features with a MedNeXt-style residual convolutional encoder–decoder. The encoder produces hierarchical features h 1 , . . . , h5 , where the deepest bottleneck feature is globally pooled to form disease and context latents zd and zc. These latents are supervised by disease and context proxy head… view at source ↗
Figure 2
Figure 2. Figure 2: Causal audit overview. (a,b) Structural abstraction: context C and lesion state D jointly generate the image X, while the anatomical tar￾get Y ⋆ depends on lesion evidence. (c) Proxy-anchored segmentation learns disease-guided and context-guided la￾tents (z i d, zi c) to condition the segmenta￾tion network; dashed arrows denote reg￾ularization losses. (d) CCT fixes disease evidence and transports context l… view at source ↗
Figure 3
Figure 3. Figure 3: Structural-prior visualization. Exam￾ple prediction before and after applying the anatomy-aware structural prior. The single-component rows further clarify the function of each module. The factoriza￾tion ablation mainly weakens subregion robustness, consis￾tent with the need to separate lesion evidence from context￾dependent appearance. Remov￾ing CCT degrades OOD perfor￾mance and eliminates the audit outpu… view at source ↗
Figure 4
Figure 4. Figure 4: Hyperparameter sensitivity. Sensitivity curves for region thresholds, and ET component filtering [PITH_FULL_IMAGE:figures/full_fig_p012_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Qualitative ET exam￾ples. Ranging on all sizes. ET is the most fragile region in the WT/TC/ET hierarchy because it is small, clinically meaningful, and easily confused with context-induced enhancement [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Segmentation and context-instability visualization. [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

28 extracted references · 28 canonical work pages · 2 internal anchors

  1. [1]

    Invariant Risk Minimization

    Arjovsky, M., Bottou, L., Gulrajani, I., Lopez-Paz, D.: Invariant risk minimization. arXiv preprint arXiv:1907.02893 (2019) 2, 4

    work page internal anchor Pith review Pith/arXiv arXiv 1907

  2. [2]

    Scientific Data 4, 170117 (2017).https://doi.org/10.1038/sdata.2017.1179

    Bakas, S., Akbari, H., Sotiras, A., Bilello, M., Rozycki, M., Kirby, J.S., Freymann, J.B., Farahani, K., Davatzikos, C.: Advancing the cancer genome atlas glioma mri collections with expert segmentation labels and radiomic features. Scientific Data 4, 170117 (2017).https://doi.org/10.1038/sdata.2017.1179

    work page doi:10.1038/sdata.2017.1179 2017

  3. [3]

    Bakas, S., Reyes, M., Jakab, A., Bauer, S., et al.: Identifying the best machine learning algorithms for brain tumor segmentation, progression assessment, and overall survival prediction in the brats challenge (2018) 9

    work page 2018

  4. [4]

    In: 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)

    Chen, B., Zhu, Y., Ao, Y., Caprara, S., Sutter, R., Rätsch, G., Konukoglu, E., Susmelj, A.: Generalizable single-source cross-modality medical image segmenta- tion via invariant causal mechanisms. In: 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV). pp. 3592–3602. IEEE (2025) 5, 10, 11

    work page 2025

  5. [5]

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

    Chen, C., Liu, X., Ding, M., Zheng, J., Li, J.: 3d dilated multi-fiber network for real-time brain tumor segmentation in mri. In: International conference on medical image computing and computer-assisted intervention. pp. 184–192. Springer (2019) 4, 10, 11

    work page 2019

  6. [6]

    In: Medical Image Computing and Computer Assisted Inter- vention – MICCAI 2024

    Chen,Y.,Wei,M.,Zheng,Z.,Hu,J.,Shi,Y.,Xiong,S.,Zhu,X.X.,Mou,L.:Causal- clipseg: Unlocking clip’s potential in referring medical image segmentation with causal intervention. In: Medical Image Computing and Computer Assisted Inter- vention – MICCAI 2024. vol. LNCS 15003, pp. 77–87. Springer Nature Switzerland (October 2024) 5, 10, 11

    work page 2024

  7. [7]

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

    Çiçek, Ö., Abdulkadir, A., Lienkamp, S.S., Brox, T., Ronneberger, O.: 3d u-net: learning dense volumetric segmentation from sparse annotation. In: International conference on medical image computing and computer-assisted intervention. pp. 424–432. Springer (2016) 2, 4

    work page 2016

  8. [8]

    In: International MICCAI brainlesion workshop

    Hatamizadeh, A., Nath, V., Tang, Y., Yang, D., Roth, H.R., Xu, D.: Swin unetr: Swin transformers for semantic segmentation of brain tumors in mri images. In: International MICCAI brainlesion workshop. pp. 272–284. Springer (2021) 2, 4

    work page 2021

  9. [9]

    In: Proceedings of the IEEE/CVF winter conference on applications of computer vi- sion

    Hatamizadeh, A., Tang, Y., Nath, V., Yang, D., Myronenko, A., Landman, B., Roth, H.R., Xu, D.: Unetr: Transformers for 3d medical image segmentation. In: Proceedings of the IEEE/CVF winter conference on applications of computer vi- sion. pp. 574–584 (2022) 2, 4, 10, 11

    work page 2022

  10. [10]

    Nature methods18(2), 203–211 (2021) 2, 4, 10, 11

    Isensee, F., Jaeger, P.F., Kohl, S.A., Petersen, J., Maier-Hein, K.H.: nnu-net: a self-configuring method for deep learning-based biomedical image segmentation. Nature methods18(2), 203–211 (2021) 2, 4, 10, 11

    work page 2021

  11. [11]

    Advances in neural information processing sys- tems31(2018) 4

    Kohl, S., Romera-Paredes, B., Meyer, C., De Fauw, J., Ledsam, J.R., Maier-Hein, K., Eslami, S., Jimenez Rezende, D., Ronneberger, O.: A probabilistic u-net for segmentation of ambiguous images. Advances in neural information processing sys- tems31(2018) 4

    work page 2018

  12. [12]

    In: International Conference on Medical Image Computing and Computer-Assisted Intervention

    Liu, H., Li, Q., Nie, W., Xu, Z., Liu, A.: Causal intervention for brain tumor segmentation. In: International Conference on Medical Image Computing and Computer-Assisted Intervention. pp. 160–170. Springer (2024) 2, 5

    work page 2024

  13. [13]

    IEEE Transactions on Medical Imaging (2025) 10, 11

    Lv, P., Wang, Y., Liu, M., Zhang, Z., Ma, Y., Liu, L., Meijering, E.: Ciseg: Unsu- pervised cross-modality adaptation for 3d medical image segmentation via causal intervention. IEEE Transactions on Medical Imaging (2025) 10, 11

    work page 2025

  14. [14]

    In: Inter- 16 N.L.D

    Mehta, R., De Sousa Ribeiro, F., Xia, T., Roschewitz, M., Santhirasekaram, A., Marshall, D.C., Glocker, B.: Cf-seg: Counterfactuals meet segmentation. In: Inter- 16 N.L.D. Le et al. national Conference on Medical Image Computing and Computer-Assisted Inter- vention. pp. 117–127. Springer (2025) 2, 5, 10, 11

    work page 2025

  15. [15]

    IEEE Transactions on Medical Imaging34(10), 1993–2024 (2015).https://doi.org/ 10.1109/TMI.2014.23776949

    Menze, B.H., Jakab, A., Bauer, S., Kalpathy-Cramer, J., Farahani, K., Kirby, J., et al.: The multimodal brain tumor image segmentation benchmark (brats). IEEE Transactions on Medical Imaging34(10), 1993–2024 (2015).https://doi.org/ 10.1109/TMI.2014.23776949

    work page doi:10.1109/tmi.2014.23776949 1993

  16. [16]

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

    Miao, J., Chen, C., Liu, F., Wei, H., Heng, P.A.: Caussl: Causality-inspired semi-supervised learning for medical image segmentation. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 21426–21437 (2023) 2, 5, 10, 11

    work page 2023

  17. [17]

    In: 2016 fourth international confer- ence on 3D vision (3DV)

    Milletari, F., Navab, N., Ahmadi, S.A.: V-net: Fully convolutional neural networks for volumetric medical image segmentation. In: 2016 fourth international confer- ence on 3D vision (3DV). pp. 565–571. Ieee (2016) 2, 4

    work page 2016

  18. [18]

    IEEE Transactions on Medical Imaging42(4), 1095–1106 (2022) 2, 5, 10, 11

    Ouyang, C., Chen, C., Li, S., Li, Z., Qin, C., Bai, W., Rueckert, D.: Causality- inspired single-source domain generalization for medical image segmentation. IEEE Transactions on Medical Imaging42(4), 1095–1106 (2022) 2, 5, 10, 11

    work page 2022

  19. [19]

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

    Perera, S., Navard, P., Yilmaz, A.: Segformer3d: an efficient transformer for 3d medical image segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 4981–4988 (2024) 2, 4, 10, 11

    work page 2024

  20. [20]

    The Cancer Imaging Archive (2026).https://doi.org/10.7937/ DFAE-1B86, version 1, Data set 9

    Reddy, D., Saadat, N., Holcomb, J., Wagner, B., Truong, N., Bowerman, J., Hatan- paa, K., Patel, T., Pinho, M., Yu, F., Zhang, K., Lodhi, S., Madhuranthakam, A., Bangalore Yogananda, C.G., Maldjian, J.: The university of texas southwest- ern glioma mri dataset with molecular marker characterization and segmentations (utsw-glioma). The Cancer Imaging Archi...

    work page 2026

  21. [21]

    In: International Conference on Medical image computing and computer-assisted intervention

    Ronneberger, O., Fischer, P., Brox, T.: U-net: Convolutional networks for biomedi- cal image segmentation. In: International Conference on Medical image computing and computer-assisted intervention. pp. 234–241. Springer (2015) 2, 4

    work page 2015

  22. [22]

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

    Roy, S., Koehler, G., Ulrich, C., Baumgartner, M., Petersen, J., Isensee, F., Jaeger, P.F., Maier-Hein, K.H.: Mednext: transformer-driven scaling of convnets for med- ical image segmentation. In: International conference on medical image computing and computer-assisted intervention. pp. 405–415. Springer (2023) 2, 4, 10, 11

    work page 2023

  23. [23]

    In: 9th International Conference on Learning Representations (ICLR 2021) (2021) 2, 4

    Sauer, A., Geiger, A.: Counterfactual generative networks. In: 9th International Conference on Learning Representations (ICLR 2021) (2021) 2, 4

    work page 2021

  24. [24]

    Proceedings of the IEEE109(5), 612–634 (2021) 2, 4

    Schölkopf, B., Locatello, F., Bauer, S., Ke, N.R., Kalchbrenner, N., Goyal, A., Ben- gio, Y.: Toward causal representation learning. Proceedings of the IEEE109(5), 612–634 (2021) 2, 4

    work page 2021

  25. [25]

    IEEE Transactions on Medical Imaging (2025) 10, 11

    Shen, Z., Cao, P., Zhou, Q., Yang, J., Zaiane, O.R.: Collaborative learning of augmentationanddisentanglementforsemi-superviseddomaingeneralizedmedical image segmentation. IEEE Transactions on Medical Imaging (2025) 10, 11

    work page 2025

  26. [26]

    Advances in neural information processing systems34, 12077–12090 (2021) 4

    Xie, E., Wang, W., Yu, Z., Anandkumar, A., Alvarez, J.M., Luo, P.: Segformer: Simple and efficient design for semantic segmentation with transformers. Advances in neural information processing systems34, 12077–12090 (2021) 4

    work page 2021

  27. [27]

    CausalVAE: Structured Causal Disentanglement in Variational Autoencoder

    Yang, M., Liu, F., Chen, Z., Shen, X., Hao, J., Wang, J.: Causalvae: Structured causal disentanglement in variational autoencoder. arXiv preprint arXiv:2004.08697 (2020) 2, 4

    work page internal anchor Pith review Pith/arXiv arXiv 2004

  28. [28]

    IEEE Journal of Biomedical and Health Informatics30(2), 1429– 1442 (2025) 10, 11

    Zhu, Z., Dong, Q., Luo, G., Wang, W., Dong, S., Wang, K., Tian, Y., Wang, G., Li, S.: Causality-adjusted data augmentation for domain continual medical image segmentation. IEEE Journal of Biomedical and Health Informatics30(2), 1429– 1442 (2025) 10, 11

    work page 2025